{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9998823021695635,
  "eval_steps": 187,
  "global_step": 9558,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003138608811644239,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0002,
      "loss": 1.6202,
      "step": 1
    },
    {
      "epoch": 0.0003138608811644239,
      "eval_loss": 3.076237440109253,
      "eval_runtime": 148.221,
      "eval_samples_per_second": 6.747,
      "eval_steps_per_second": 6.747,
      "step": 1
    },
    {
      "epoch": 0.0003138608811644239,
      "mmlu_eval_accuracy": 0.44607718879653785,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.3333333333333333,
      "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": 0.1,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.27906976744186046,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.75,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.2727272727272727,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.6046511627906976,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.48484848484848486,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.36470588235294116,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.36231884057971014,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "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": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 3.0195472597842277,
      "step": 1
    },
    {
      "epoch": 0.0006277217623288478,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0002,
      "loss": 1.6781,
      "step": 2
    },
    {
      "epoch": 0.0009415826434932716,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.6066,
      "step": 3
    },
    {
      "epoch": 0.0012554435246576955,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.6144,
      "step": 4
    },
    {
      "epoch": 0.0015693044058221193,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.5999,
      "step": 5
    },
    {
      "epoch": 0.0018831652869865433,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.4956,
      "step": 6
    },
    {
      "epoch": 0.002197026168150967,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.5835,
      "step": 7
    },
    {
      "epoch": 0.002510887049315391,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.4302,
      "step": 8
    },
    {
      "epoch": 0.002824747930479815,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.5676,
      "step": 9
    },
    {
      "epoch": 0.0031386088116442386,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.573,
      "step": 10
    },
    {
      "epoch": 0.0034524696928086625,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.4498,
      "step": 11
    },
    {
      "epoch": 0.0037663305739730865,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.7236,
      "step": 12
    },
    {
      "epoch": 0.0040801914551375105,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.5408,
      "step": 13
    },
    {
      "epoch": 0.004394052336301934,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4307,
      "step": 14
    },
    {
      "epoch": 0.004707913217466358,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.584,
      "step": 15
    },
    {
      "epoch": 0.005021774098630782,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.4904,
      "step": 16
    },
    {
      "epoch": 0.005335634979795206,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.848,
      "step": 17
    },
    {
      "epoch": 0.00564949586095963,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.5244,
      "step": 18
    },
    {
      "epoch": 0.005963356742124054,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.4058,
      "step": 19
    },
    {
      "epoch": 0.006277217623288477,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.5462,
      "step": 20
    },
    {
      "epoch": 0.0065910785044529015,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.6042,
      "step": 21
    },
    {
      "epoch": 0.006904939385617325,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.6849,
      "step": 22
    },
    {
      "epoch": 0.007218800266781749,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 2.0465,
      "step": 23
    },
    {
      "epoch": 0.007532661147946173,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.6686,
      "step": 24
    },
    {
      "epoch": 0.007846522029110597,
      "grad_norm": 0.875,
      "learning_rate": 0.0002,
      "loss": 1.7766,
      "step": 25
    },
    {
      "epoch": 0.008160382910275021,
      "grad_norm": 0.75,
      "learning_rate": 0.0002,
      "loss": 1.9928,
      "step": 26
    },
    {
      "epoch": 0.008474243791439445,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002,
      "loss": 2.442,
      "step": 27
    },
    {
      "epoch": 0.008788104672603868,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.9709,
      "step": 28
    },
    {
      "epoch": 0.009101965553768292,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.1932,
      "step": 29
    },
    {
      "epoch": 0.009415826434932715,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.1419,
      "step": 30
    },
    {
      "epoch": 0.00972968731609714,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.1,
      "step": 31
    },
    {
      "epoch": 0.010043548197261564,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002,
      "loss": 2.4471,
      "step": 32
    },
    {
      "epoch": 0.010357409078425988,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002,
      "loss": 2.6097,
      "step": 33
    },
    {
      "epoch": 0.010671269959590411,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 2.2288,
      "step": 34
    },
    {
      "epoch": 0.010985130840754835,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002,
      "loss": 2.6281,
      "step": 35
    },
    {
      "epoch": 0.01129899172191926,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.9545,
      "step": 36
    },
    {
      "epoch": 0.011612852603083684,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0002,
      "loss": 2.2779,
      "step": 37
    },
    {
      "epoch": 0.011926713484248107,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0002,
      "loss": 2.5123,
      "step": 38
    },
    {
      "epoch": 0.01224057436541253,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002,
      "loss": 2.5084,
      "step": 39
    },
    {
      "epoch": 0.012554435246576954,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 2.5873,
      "step": 40
    },
    {
      "epoch": 0.012868296127741378,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002,
      "loss": 2.4971,
      "step": 41
    },
    {
      "epoch": 0.013182157008905803,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002,
      "loss": 2.3611,
      "step": 42
    },
    {
      "epoch": 0.013496017890070227,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002,
      "loss": 2.9962,
      "step": 43
    },
    {
      "epoch": 0.01380987877123465,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 2.3759,
      "step": 44
    },
    {
      "epoch": 0.014123739652399074,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 2.4454,
      "step": 45
    },
    {
      "epoch": 0.014437600533563497,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 2.1861,
      "step": 46
    },
    {
      "epoch": 0.014751461414727923,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 2.5668,
      "step": 47
    },
    {
      "epoch": 0.015065322295892346,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 2.0805,
      "step": 48
    },
    {
      "epoch": 0.01537918317705677,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002,
      "loss": 2.8306,
      "step": 49
    },
    {
      "epoch": 0.015693044058221195,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002,
      "loss": 2.2581,
      "step": 50
    },
    {
      "epoch": 0.016006904939385617,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.5381,
      "step": 51
    },
    {
      "epoch": 0.016320765820550042,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.3098,
      "step": 52
    },
    {
      "epoch": 0.016634626701714464,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.5292,
      "step": 53
    },
    {
      "epoch": 0.01694848758287889,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.6597,
      "step": 54
    },
    {
      "epoch": 0.017262348464043314,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.4327,
      "step": 55
    },
    {
      "epoch": 0.017576209345207736,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4566,
      "step": 56
    },
    {
      "epoch": 0.01789007022637216,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3754,
      "step": 57
    },
    {
      "epoch": 0.018203931107536583,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.7264,
      "step": 58
    },
    {
      "epoch": 0.01851779198870101,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.4453,
      "step": 59
    },
    {
      "epoch": 0.01883165286986543,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.463,
      "step": 60
    },
    {
      "epoch": 0.019145513751029856,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4478,
      "step": 61
    },
    {
      "epoch": 0.01945937463219428,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.5525,
      "step": 62
    },
    {
      "epoch": 0.019773235513358703,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.466,
      "step": 63
    },
    {
      "epoch": 0.020087096394523128,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.3745,
      "step": 64
    },
    {
      "epoch": 0.02040095727568755,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.4817,
      "step": 65
    },
    {
      "epoch": 0.020714818156851975,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.4156,
      "step": 66
    },
    {
      "epoch": 0.0210286790380164,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.7504,
      "step": 67
    },
    {
      "epoch": 0.021342539919180822,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.4968,
      "step": 68
    },
    {
      "epoch": 0.021656400800345248,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.585,
      "step": 69
    },
    {
      "epoch": 0.02197026168150967,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.5652,
      "step": 70
    },
    {
      "epoch": 0.022284122562674095,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2941,
      "step": 71
    },
    {
      "epoch": 0.02259798344383852,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.7452,
      "step": 72
    },
    {
      "epoch": 0.022911844325002942,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.5938,
      "step": 73
    },
    {
      "epoch": 0.023225705206167367,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.781,
      "step": 74
    },
    {
      "epoch": 0.02353956608733179,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.792,
      "step": 75
    },
    {
      "epoch": 0.023853426968496214,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.9774,
      "step": 76
    },
    {
      "epoch": 0.02416728784966064,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 2.4913,
      "step": 77
    },
    {
      "epoch": 0.02448114873082506,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.0286,
      "step": 78
    },
    {
      "epoch": 0.024795009611989487,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 2.3414,
      "step": 79
    },
    {
      "epoch": 0.02510887049315391,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.8796,
      "step": 80
    },
    {
      "epoch": 0.025422731374318334,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.9177,
      "step": 81
    },
    {
      "epoch": 0.025736592255482756,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.7417,
      "step": 82
    },
    {
      "epoch": 0.02605045313664718,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.3617,
      "step": 83
    },
    {
      "epoch": 0.026364314017811606,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 2.418,
      "step": 84
    },
    {
      "epoch": 0.026678174898976028,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0002,
      "loss": 2.7761,
      "step": 85
    },
    {
      "epoch": 0.026992035780140453,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 2.3215,
      "step": 86
    },
    {
      "epoch": 0.027305896661304875,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.1868,
      "step": 87
    },
    {
      "epoch": 0.0276197575424693,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 2.0753,
      "step": 88
    },
    {
      "epoch": 0.027933618423633726,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002,
      "loss": 2.2398,
      "step": 89
    },
    {
      "epoch": 0.028247479304798147,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 2.4929,
      "step": 90
    },
    {
      "epoch": 0.028561340185962573,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 2.4407,
      "step": 91
    },
    {
      "epoch": 0.028875201067126995,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002,
      "loss": 2.7458,
      "step": 92
    },
    {
      "epoch": 0.02918906194829142,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 2.0544,
      "step": 93
    },
    {
      "epoch": 0.029502922829455845,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 2.6945,
      "step": 94
    },
    {
      "epoch": 0.029816783710620267,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 2.3824,
      "step": 95
    },
    {
      "epoch": 0.030130644591784692,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.2367,
      "step": 96
    },
    {
      "epoch": 0.030444505472949114,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002,
      "loss": 2.6735,
      "step": 97
    },
    {
      "epoch": 0.03075836635411354,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.3055,
      "step": 98
    },
    {
      "epoch": 0.031072227235277965,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 2.432,
      "step": 99
    },
    {
      "epoch": 0.03138608811644239,
      "grad_norm": 0.875,
      "learning_rate": 0.0002,
      "loss": 3.0268,
      "step": 100
    },
    {
      "epoch": 0.03169994899760681,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.5252,
      "step": 101
    },
    {
      "epoch": 0.032013809878771234,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.5605,
      "step": 102
    },
    {
      "epoch": 0.03232767075993566,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.5084,
      "step": 103
    },
    {
      "epoch": 0.032641531641100084,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.462,
      "step": 104
    },
    {
      "epoch": 0.03295539252226451,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.6249,
      "step": 105
    },
    {
      "epoch": 0.03326925340342893,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.641,
      "step": 106
    },
    {
      "epoch": 0.03358311428459335,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.4551,
      "step": 107
    },
    {
      "epoch": 0.03389697516575778,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.4532,
      "step": 108
    },
    {
      "epoch": 0.034210836046922204,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.5638,
      "step": 109
    },
    {
      "epoch": 0.03452469692808663,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.6085,
      "step": 110
    },
    {
      "epoch": 0.03483855780925105,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4586,
      "step": 111
    },
    {
      "epoch": 0.03515241869041547,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.4463,
      "step": 112
    },
    {
      "epoch": 0.0354662795715799,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.6576,
      "step": 113
    },
    {
      "epoch": 0.03578014045274432,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.4852,
      "step": 114
    },
    {
      "epoch": 0.03609400133390874,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.3461,
      "step": 115
    },
    {
      "epoch": 0.03640786221507317,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.3754,
      "step": 116
    },
    {
      "epoch": 0.03672172309623759,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.5055,
      "step": 117
    },
    {
      "epoch": 0.03703558397740202,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.5216,
      "step": 118
    },
    {
      "epoch": 0.03734944485856644,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.5402,
      "step": 119
    },
    {
      "epoch": 0.03766330573973086,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.5661,
      "step": 120
    },
    {
      "epoch": 0.037977166620895286,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.533,
      "step": 121
    },
    {
      "epoch": 0.03829102750205971,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.6953,
      "step": 122
    },
    {
      "epoch": 0.03860488838322414,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.3169,
      "step": 123
    },
    {
      "epoch": 0.03891874926438856,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.7129,
      "step": 124
    },
    {
      "epoch": 0.03923261014555298,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.3726,
      "step": 125
    },
    {
      "epoch": 0.039546471026717406,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.6792,
      "step": 126
    },
    {
      "epoch": 0.03986033190788183,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 2.3559,
      "step": 127
    },
    {
      "epoch": 0.040174192789046256,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.5121,
      "step": 128
    },
    {
      "epoch": 0.04048805367021068,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.715,
      "step": 129
    },
    {
      "epoch": 0.0408019145513751,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 2.0793,
      "step": 130
    },
    {
      "epoch": 0.041115775432539525,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.7691,
      "step": 131
    },
    {
      "epoch": 0.04142963631370395,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.973,
      "step": 132
    },
    {
      "epoch": 0.041743497194868376,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.9109,
      "step": 133
    },
    {
      "epoch": 0.0420573580760328,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002,
      "loss": 2.0848,
      "step": 134
    },
    {
      "epoch": 0.04237121895719722,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002,
      "loss": 2.6422,
      "step": 135
    },
    {
      "epoch": 0.042685079838361645,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 2.3111,
      "step": 136
    },
    {
      "epoch": 0.04299894071952607,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.421,
      "step": 137
    },
    {
      "epoch": 0.043312801600690495,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.2324,
      "step": 138
    },
    {
      "epoch": 0.04362666248185492,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.1292,
      "step": 139
    },
    {
      "epoch": 0.04394052336301934,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 2.945,
      "step": 140
    },
    {
      "epoch": 0.044254384244183764,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 2.5678,
      "step": 141
    },
    {
      "epoch": 0.04456824512534819,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0002,
      "loss": 2.4902,
      "step": 142
    },
    {
      "epoch": 0.044882106006512615,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002,
      "loss": 3.1968,
      "step": 143
    },
    {
      "epoch": 0.04519596688767704,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002,
      "loss": 2.4699,
      "step": 144
    },
    {
      "epoch": 0.04550982776884146,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.2869,
      "step": 145
    },
    {
      "epoch": 0.045823688650005884,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.2249,
      "step": 146
    },
    {
      "epoch": 0.04613754953117031,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 2.4014,
      "step": 147
    },
    {
      "epoch": 0.046451410412334734,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 2.3175,
      "step": 148
    },
    {
      "epoch": 0.04676527129349916,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 2.7023,
      "step": 149
    },
    {
      "epoch": 0.04707913217466358,
      "grad_norm": 1.375,
      "learning_rate": 0.0002,
      "loss": 3.2566,
      "step": 150
    },
    {
      "epoch": 0.047392993055828,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.4692,
      "step": 151
    },
    {
      "epoch": 0.04770685393699243,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.4202,
      "step": 152
    },
    {
      "epoch": 0.048020714818156854,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.4694,
      "step": 153
    },
    {
      "epoch": 0.04833457569932128,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.5723,
      "step": 154
    },
    {
      "epoch": 0.0486484365804857,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.5976,
      "step": 155
    },
    {
      "epoch": 0.04896229746165012,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3935,
      "step": 156
    },
    {
      "epoch": 0.04927615834281455,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.45,
      "step": 157
    },
    {
      "epoch": 0.04959001922397897,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.4176,
      "step": 158
    },
    {
      "epoch": 0.0499038801051434,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.498,
      "step": 159
    },
    {
      "epoch": 0.05021774098630782,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.5622,
      "step": 160
    },
    {
      "epoch": 0.05053160186747224,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.5955,
      "step": 161
    },
    {
      "epoch": 0.05084546274863667,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.6085,
      "step": 162
    },
    {
      "epoch": 0.05115932362980109,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.4561,
      "step": 163
    },
    {
      "epoch": 0.05147318451096551,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3197,
      "step": 164
    },
    {
      "epoch": 0.051787045392129936,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4097,
      "step": 165
    },
    {
      "epoch": 0.05210090627329436,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4671,
      "step": 166
    },
    {
      "epoch": 0.05241476715445879,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.713,
      "step": 167
    },
    {
      "epoch": 0.05272862803562321,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.4736,
      "step": 168
    },
    {
      "epoch": 0.05304248891678763,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.5112,
      "step": 169
    },
    {
      "epoch": 0.053356349797952056,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.9055,
      "step": 170
    },
    {
      "epoch": 0.05367021067911648,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.502,
      "step": 171
    },
    {
      "epoch": 0.05398407156028091,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.6173,
      "step": 172
    },
    {
      "epoch": 0.05429793244144533,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.6522,
      "step": 173
    },
    {
      "epoch": 0.05461179332260975,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.6627,
      "step": 174
    },
    {
      "epoch": 0.054925654203774175,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.718,
      "step": 175
    },
    {
      "epoch": 0.0552395150849386,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.5639,
      "step": 176
    },
    {
      "epoch": 0.055553375966103026,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.8627,
      "step": 177
    },
    {
      "epoch": 0.05586723684726745,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.9256,
      "step": 178
    },
    {
      "epoch": 0.05618109772843187,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.9848,
      "step": 179
    },
    {
      "epoch": 0.056494958609596295,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 2.0576,
      "step": 180
    },
    {
      "epoch": 0.05680881949076072,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 2.289,
      "step": 181
    },
    {
      "epoch": 0.057122680371925146,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 2.5124,
      "step": 182
    },
    {
      "epoch": 0.05743654125308957,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 2.1902,
      "step": 183
    },
    {
      "epoch": 0.05775040213425399,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.3745,
      "step": 184
    },
    {
      "epoch": 0.058064263015418414,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 3.0241,
      "step": 185
    },
    {
      "epoch": 0.05837812389658284,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 2.6551,
      "step": 186
    },
    {
      "epoch": 0.058691984777747265,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 2.6666,
      "step": 187
    },
    {
      "epoch": 0.058691984777747265,
      "eval_loss": 1.8769614696502686,
      "eval_runtime": 123.4185,
      "eval_samples_per_second": 8.103,
      "eval_steps_per_second": 8.103,
      "step": 187
    },
    {
      "epoch": 0.058691984777747265,
      "mmlu_eval_accuracy": 0.44123983572722647,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.3333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.6833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.5,
      "mmlu_eval_accuracy_human_aging": 0.5652173913043478,
      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.2727272727272727,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.627906976744186,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.37142857142857144,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_psychology": 0.36231884057971014,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "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": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 0.9604393265299513,
      "step": 187
    },
    {
      "epoch": 0.05900584565891169,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 2.422,
      "step": 188
    },
    {
      "epoch": 0.05931970654007611,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 2.571,
      "step": 189
    },
    {
      "epoch": 0.059633567421240534,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 2.2629,
      "step": 190
    },
    {
      "epoch": 0.05994742830240496,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.8845,
      "step": 191
    },
    {
      "epoch": 0.060261289183569385,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.1545,
      "step": 192
    },
    {
      "epoch": 0.06057515006473381,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 2.5156,
      "step": 193
    },
    {
      "epoch": 0.06088901094589823,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 2.517,
      "step": 194
    },
    {
      "epoch": 0.06120287182706265,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.2571,
      "step": 195
    },
    {
      "epoch": 0.06151673270822708,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 2.2549,
      "step": 196
    },
    {
      "epoch": 0.061830593589391504,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 2.275,
      "step": 197
    },
    {
      "epoch": 0.06214445447055593,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.0545,
      "step": 198
    },
    {
      "epoch": 0.06245831535172035,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 3.1062,
      "step": 199
    },
    {
      "epoch": 0.06277217623288478,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 2.6846,
      "step": 200
    },
    {
      "epoch": 0.0630860371140492,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.5702,
      "step": 201
    },
    {
      "epoch": 0.06339989799521362,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.6625,
      "step": 202
    },
    {
      "epoch": 0.06371375887637805,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.423,
      "step": 203
    },
    {
      "epoch": 0.06402761975754247,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.5124,
      "step": 204
    },
    {
      "epoch": 0.0643414806387069,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.499,
      "step": 205
    },
    {
      "epoch": 0.06465534151987132,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.5965,
      "step": 206
    },
    {
      "epoch": 0.06496920240103574,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3176,
      "step": 207
    },
    {
      "epoch": 0.06528306328220017,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.4845,
      "step": 208
    },
    {
      "epoch": 0.06559692416336459,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.3401,
      "step": 209
    },
    {
      "epoch": 0.06591078504452902,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.4552,
      "step": 210
    },
    {
      "epoch": 0.06622464592569344,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4667,
      "step": 211
    },
    {
      "epoch": 0.06653850680685786,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.3008,
      "step": 212
    },
    {
      "epoch": 0.06685236768802229,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.7032,
      "step": 213
    },
    {
      "epoch": 0.0671662285691867,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.4191,
      "step": 214
    },
    {
      "epoch": 0.06748008945035114,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3981,
      "step": 215
    },
    {
      "epoch": 0.06779395033151556,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3225,
      "step": 216
    },
    {
      "epoch": 0.06810781121267998,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.4282,
      "step": 217
    },
    {
      "epoch": 0.06842167209384441,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.7299,
      "step": 218
    },
    {
      "epoch": 0.06873553297500883,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.4266,
      "step": 219
    },
    {
      "epoch": 0.06904939385617326,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.4593,
      "step": 220
    },
    {
      "epoch": 0.06936325473733768,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.7093,
      "step": 221
    },
    {
      "epoch": 0.0696771156185021,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.7773,
      "step": 222
    },
    {
      "epoch": 0.06999097649966653,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.7139,
      "step": 223
    },
    {
      "epoch": 0.07030483738083095,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.8133,
      "step": 224
    },
    {
      "epoch": 0.07061869826199538,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.6248,
      "step": 225
    },
    {
      "epoch": 0.0709325591431598,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.7408,
      "step": 226
    },
    {
      "epoch": 0.07124642002432421,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.7419,
      "step": 227
    },
    {
      "epoch": 0.07156028090548865,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.7289,
      "step": 228
    },
    {
      "epoch": 0.07187414178665306,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 2.2417,
      "step": 229
    },
    {
      "epoch": 0.07218800266781748,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.8025,
      "step": 230
    },
    {
      "epoch": 0.07250186354898192,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 2.0277,
      "step": 231
    },
    {
      "epoch": 0.07281572443014633,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002,
      "loss": 2.6918,
      "step": 232
    },
    {
      "epoch": 0.07312958531131077,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.0573,
      "step": 233
    },
    {
      "epoch": 0.07344344619247518,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 2.506,
      "step": 234
    },
    {
      "epoch": 0.0737573070736396,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 2.009,
      "step": 235
    },
    {
      "epoch": 0.07407116795480403,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 2.2538,
      "step": 236
    },
    {
      "epoch": 0.07438502883596845,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 2.4865,
      "step": 237
    },
    {
      "epoch": 0.07469888971713289,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 2.3167,
      "step": 238
    },
    {
      "epoch": 0.0750127505982973,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 2.3738,
      "step": 239
    },
    {
      "epoch": 0.07532661147946172,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.0565,
      "step": 240
    },
    {
      "epoch": 0.07564047236062615,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 2.5195,
      "step": 241
    },
    {
      "epoch": 0.07595433324179057,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.3961,
      "step": 242
    },
    {
      "epoch": 0.076268194122955,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 2.8956,
      "step": 243
    },
    {
      "epoch": 0.07658205500411942,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 2.5885,
      "step": 244
    },
    {
      "epoch": 0.07689591588528384,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.8268,
      "step": 245
    },
    {
      "epoch": 0.07720977676644827,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.9894,
      "step": 246
    },
    {
      "epoch": 0.07752363764761269,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 2.0273,
      "step": 247
    },
    {
      "epoch": 0.07783749852877712,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.2477,
      "step": 248
    },
    {
      "epoch": 0.07815135940994154,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 2.5224,
      "step": 249
    },
    {
      "epoch": 0.07846522029110596,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002,
      "loss": 3.2266,
      "step": 250
    },
    {
      "epoch": 0.0787790811722704,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.4904,
      "step": 251
    },
    {
      "epoch": 0.07909294205343481,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.4279,
      "step": 252
    },
    {
      "epoch": 0.07940680293459924,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.5421,
      "step": 253
    },
    {
      "epoch": 0.07972066381576366,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.5092,
      "step": 254
    },
    {
      "epoch": 0.08003452469692808,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.5434,
      "step": 255
    },
    {
      "epoch": 0.08034838557809251,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.4732,
      "step": 256
    },
    {
      "epoch": 0.08066224645925693,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.4807,
      "step": 257
    },
    {
      "epoch": 0.08097610734042136,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.5032,
      "step": 258
    },
    {
      "epoch": 0.08128996822158578,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.5402,
      "step": 259
    },
    {
      "epoch": 0.0816038291027502,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4766,
      "step": 260
    },
    {
      "epoch": 0.08191768998391463,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4982,
      "step": 261
    },
    {
      "epoch": 0.08223155086507905,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3532,
      "step": 262
    },
    {
      "epoch": 0.08254541174624348,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.3225,
      "step": 263
    },
    {
      "epoch": 0.0828592726274079,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4933,
      "step": 264
    },
    {
      "epoch": 0.08317313350857232,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.4604,
      "step": 265
    },
    {
      "epoch": 0.08348699438973675,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.337,
      "step": 266
    },
    {
      "epoch": 0.08380085527090117,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.465,
      "step": 267
    },
    {
      "epoch": 0.0841147161520656,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.5809,
      "step": 268
    },
    {
      "epoch": 0.08442857703323002,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.921,
      "step": 269
    },
    {
      "epoch": 0.08474243791439444,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.5733,
      "step": 270
    },
    {
      "epoch": 0.08505629879555887,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.5355,
      "step": 271
    },
    {
      "epoch": 0.08537015967672329,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 1.6797,
      "step": 272
    },
    {
      "epoch": 0.08568402055788772,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.6325,
      "step": 273
    },
    {
      "epoch": 0.08599788143905214,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.5476,
      "step": 274
    },
    {
      "epoch": 0.08631174232021656,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.6788,
      "step": 275
    },
    {
      "epoch": 0.08662560320138099,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 2.0317,
      "step": 276
    },
    {
      "epoch": 0.08693946408254541,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.8197,
      "step": 277
    },
    {
      "epoch": 0.08725332496370984,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.7157,
      "step": 278
    },
    {
      "epoch": 0.08756718584487426,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 2.2781,
      "step": 279
    },
    {
      "epoch": 0.08788104672603868,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 2.1994,
      "step": 280
    },
    {
      "epoch": 0.08819490760720311,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 2.1835,
      "step": 281
    },
    {
      "epoch": 0.08850876848836753,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.9166,
      "step": 282
    },
    {
      "epoch": 0.08882262936953196,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.1354,
      "step": 283
    },
    {
      "epoch": 0.08913649025069638,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 2.3373,
      "step": 284
    },
    {
      "epoch": 0.0894503511318608,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 2.5996,
      "step": 285
    },
    {
      "epoch": 0.08976421201302523,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.9957,
      "step": 286
    },
    {
      "epoch": 0.09007807289418965,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002,
      "loss": 2.4719,
      "step": 287
    },
    {
      "epoch": 0.09039193377535408,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 2.1207,
      "step": 288
    },
    {
      "epoch": 0.0907057946565185,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 2.365,
      "step": 289
    },
    {
      "epoch": 0.09101965553768292,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002,
      "loss": 2.3941,
      "step": 290
    },
    {
      "epoch": 0.09133351641884735,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.1109,
      "step": 291
    },
    {
      "epoch": 0.09164737730001177,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002,
      "loss": 2.3109,
      "step": 292
    },
    {
      "epoch": 0.0919612381811762,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 2.3769,
      "step": 293
    },
    {
      "epoch": 0.09227509906234062,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 2.5643,
      "step": 294
    },
    {
      "epoch": 0.09258895994350504,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 2.6575,
      "step": 295
    },
    {
      "epoch": 0.09290282082466947,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.0201,
      "step": 296
    },
    {
      "epoch": 0.09321668170583389,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.6622,
      "step": 297
    },
    {
      "epoch": 0.09353054258699832,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 2.7948,
      "step": 298
    },
    {
      "epoch": 0.09384440346816274,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002,
      "loss": 2.6774,
      "step": 299
    },
    {
      "epoch": 0.09415826434932716,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002,
      "loss": 3.0224,
      "step": 300
    },
    {
      "epoch": 0.09447212523049159,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.4284,
      "step": 301
    },
    {
      "epoch": 0.094785986111656,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.4059,
      "step": 302
    },
    {
      "epoch": 0.09509984699282044,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.4115,
      "step": 303
    },
    {
      "epoch": 0.09541370787398486,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4218,
      "step": 304
    },
    {
      "epoch": 0.09572756875514928,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.4053,
      "step": 305
    },
    {
      "epoch": 0.09604142963631371,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.6277,
      "step": 306
    },
    {
      "epoch": 0.09635529051747813,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3927,
      "step": 307
    },
    {
      "epoch": 0.09666915139864256,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.3535,
      "step": 308
    },
    {
      "epoch": 0.09698301227980698,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3208,
      "step": 309
    },
    {
      "epoch": 0.0972968731609714,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.4483,
      "step": 310
    },
    {
      "epoch": 0.09761073404213583,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.8075,
      "step": 311
    },
    {
      "epoch": 0.09792459492330025,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.2494,
      "step": 312
    },
    {
      "epoch": 0.09823845580446468,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4753,
      "step": 313
    },
    {
      "epoch": 0.0985523166856291,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.5693,
      "step": 314
    },
    {
      "epoch": 0.09886617756679351,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.441,
      "step": 315
    },
    {
      "epoch": 0.09918003844795795,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.6538,
      "step": 316
    },
    {
      "epoch": 0.09949389932912237,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.5327,
      "step": 317
    },
    {
      "epoch": 0.0998077602102868,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2743,
      "step": 318
    },
    {
      "epoch": 0.10012162109145122,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.282,
      "step": 319
    },
    {
      "epoch": 0.10043548197261563,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2735,
      "step": 320
    },
    {
      "epoch": 0.10074934285378007,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.5729,
      "step": 321
    },
    {
      "epoch": 0.10106320373494448,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.4801,
      "step": 322
    },
    {
      "epoch": 0.1013770646161089,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.6091,
      "step": 323
    },
    {
      "epoch": 0.10169092549727334,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.8584,
      "step": 324
    },
    {
      "epoch": 0.10200478637843775,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.507,
      "step": 325
    },
    {
      "epoch": 0.10231864725960219,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.6124,
      "step": 326
    },
    {
      "epoch": 0.1026325081407666,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.7112,
      "step": 327
    },
    {
      "epoch": 0.10294636902193102,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 2.1519,
      "step": 328
    },
    {
      "epoch": 0.10326022990309545,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.941,
      "step": 329
    },
    {
      "epoch": 0.10357409078425987,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.8767,
      "step": 330
    },
    {
      "epoch": 0.1038879516654243,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.9287,
      "step": 331
    },
    {
      "epoch": 0.10420181254658872,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.7912,
      "step": 332
    },
    {
      "epoch": 0.10451567342775314,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.8166,
      "step": 333
    },
    {
      "epoch": 0.10482953430891757,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.3756,
      "step": 334
    },
    {
      "epoch": 0.10514339519008199,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.9608,
      "step": 335
    },
    {
      "epoch": 0.10545725607124642,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 2.2499,
      "step": 336
    },
    {
      "epoch": 0.10577111695241084,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 2.3659,
      "step": 337
    },
    {
      "epoch": 0.10608497783357526,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 2.304,
      "step": 338
    },
    {
      "epoch": 0.1063988387147397,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 2.0312,
      "step": 339
    },
    {
      "epoch": 0.10671269959590411,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.2391,
      "step": 340
    },
    {
      "epoch": 0.10702656047706854,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.0306,
      "step": 341
    },
    {
      "epoch": 0.10734042135823296,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0002,
      "loss": 2.5958,
      "step": 342
    },
    {
      "epoch": 0.10765428223939738,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002,
      "loss": 2.9164,
      "step": 343
    },
    {
      "epoch": 0.10796814312056181,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 2.3001,
      "step": 344
    },
    {
      "epoch": 0.10828200400172623,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 2.5496,
      "step": 345
    },
    {
      "epoch": 0.10859586488289066,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 2.1181,
      "step": 346
    },
    {
      "epoch": 0.10890972576405508,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 2.3313,
      "step": 347
    },
    {
      "epoch": 0.1092235866452195,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.5624,
      "step": 348
    },
    {
      "epoch": 0.10953744752638393,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.3083,
      "step": 349
    },
    {
      "epoch": 0.10985130840754835,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 3.233,
      "step": 350
    },
    {
      "epoch": 0.11016516928871278,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3294,
      "step": 351
    },
    {
      "epoch": 0.1104790301698772,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.5399,
      "step": 352
    },
    {
      "epoch": 0.11079289105104162,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4208,
      "step": 353
    },
    {
      "epoch": 0.11110675193220605,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4345,
      "step": 354
    },
    {
      "epoch": 0.11142061281337047,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.424,
      "step": 355
    },
    {
      "epoch": 0.1117344736945349,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.6212,
      "step": 356
    },
    {
      "epoch": 0.11204833457569932,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.4564,
      "step": 357
    },
    {
      "epoch": 0.11236219545686374,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.5485,
      "step": 358
    },
    {
      "epoch": 0.11267605633802817,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.2384,
      "step": 359
    },
    {
      "epoch": 0.11298991721919259,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.4083,
      "step": 360
    },
    {
      "epoch": 0.11330377810035702,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3978,
      "step": 361
    },
    {
      "epoch": 0.11361763898152144,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.4216,
      "step": 362
    },
    {
      "epoch": 0.11393149986268586,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.5479,
      "step": 363
    },
    {
      "epoch": 0.11424536074385029,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.4693,
      "step": 364
    },
    {
      "epoch": 0.11455922162501471,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4133,
      "step": 365
    },
    {
      "epoch": 0.11487308250617914,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1875,
      "step": 366
    },
    {
      "epoch": 0.11518694338734356,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.5695,
      "step": 367
    },
    {
      "epoch": 0.11550080426850798,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.1122,
      "step": 368
    },
    {
      "epoch": 0.11581466514967241,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.5259,
      "step": 369
    },
    {
      "epoch": 0.11612852603083683,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.4507,
      "step": 370
    },
    {
      "epoch": 0.11644238691200126,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.6836,
      "step": 371
    },
    {
      "epoch": 0.11675624779316568,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.6754,
      "step": 372
    },
    {
      "epoch": 0.1170701086743301,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.4691,
      "step": 373
    },
    {
      "epoch": 0.11738396955549453,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.7798,
      "step": 374
    },
    {
      "epoch": 0.11738396955549453,
      "eval_loss": 1.8649098873138428,
      "eval_runtime": 144.7095,
      "eval_samples_per_second": 6.91,
      "eval_steps_per_second": 6.91,
      "step": 374
    },
    {
      "epoch": 0.11738396955549453,
      "mmlu_eval_accuracy": 0.43153515218737043,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.3125,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.22727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.0,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
      "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": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.5930232558139535,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.4,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.4904409643094731,
      "step": 374
    },
    {
      "epoch": 0.11769783043665895,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.6591,
      "step": 375
    },
    {
      "epoch": 0.11801169131782338,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.9671,
      "step": 376
    },
    {
      "epoch": 0.1183255521989878,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.9548,
      "step": 377
    },
    {
      "epoch": 0.11863941308015222,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.869,
      "step": 378
    },
    {
      "epoch": 0.11895327396131665,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 2.1079,
      "step": 379
    },
    {
      "epoch": 0.11926713484248107,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 2.2094,
      "step": 380
    },
    {
      "epoch": 0.1195809957236455,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 2.0853,
      "step": 381
    },
    {
      "epoch": 0.11989485660480992,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 2.1766,
      "step": 382
    },
    {
      "epoch": 0.12020871748597434,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.873,
      "step": 383
    },
    {
      "epoch": 0.12052257836713877,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.6411,
      "step": 384
    },
    {
      "epoch": 0.12083643924830319,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 2.0356,
      "step": 385
    },
    {
      "epoch": 0.12115030012946762,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 2.3583,
      "step": 386
    },
    {
      "epoch": 0.12146416101063204,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.0772,
      "step": 387
    },
    {
      "epoch": 0.12177802189179646,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 2.1947,
      "step": 388
    },
    {
      "epoch": 0.12209188277296089,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.741,
      "step": 389
    },
    {
      "epoch": 0.1224057436541253,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.6193,
      "step": 390
    },
    {
      "epoch": 0.12271960453528974,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.921,
      "step": 391
    },
    {
      "epoch": 0.12303346541645416,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.7521,
      "step": 392
    },
    {
      "epoch": 0.12334732629761858,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 2.1829,
      "step": 393
    },
    {
      "epoch": 0.12366118717878301,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.3879,
      "step": 394
    },
    {
      "epoch": 0.12397504805994743,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 2.2603,
      "step": 395
    },
    {
      "epoch": 0.12428890894111186,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 2.0838,
      "step": 396
    },
    {
      "epoch": 0.12460276982227628,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.8756,
      "step": 397
    },
    {
      "epoch": 0.1249166307034407,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 2.1248,
      "step": 398
    },
    {
      "epoch": 0.1252304915846051,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.3589,
      "step": 399
    },
    {
      "epoch": 0.12554435246576956,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.4052,
      "step": 400
    },
    {
      "epoch": 0.12585821334693398,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3253,
      "step": 401
    },
    {
      "epoch": 0.1261720742280984,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.5794,
      "step": 402
    },
    {
      "epoch": 0.12648593510926281,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.5086,
      "step": 403
    },
    {
      "epoch": 0.12679979599042723,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4157,
      "step": 404
    },
    {
      "epoch": 0.12711365687159168,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.402,
      "step": 405
    },
    {
      "epoch": 0.1274275177527561,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.5706,
      "step": 406
    },
    {
      "epoch": 0.12774137863392052,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.441,
      "step": 407
    },
    {
      "epoch": 0.12805523951508493,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.4242,
      "step": 408
    },
    {
      "epoch": 0.12836910039624935,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.3529,
      "step": 409
    },
    {
      "epoch": 0.1286829612774138,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.5687,
      "step": 410
    },
    {
      "epoch": 0.12899682215857822,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.6228,
      "step": 411
    },
    {
      "epoch": 0.12931068303974264,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.5197,
      "step": 412
    },
    {
      "epoch": 0.12962454392090705,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.5823,
      "step": 413
    },
    {
      "epoch": 0.12993840480207147,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.5927,
      "step": 414
    },
    {
      "epoch": 0.13025226568323592,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.481,
      "step": 415
    },
    {
      "epoch": 0.13056612656440034,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.696,
      "step": 416
    },
    {
      "epoch": 0.13087998744556475,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.4169,
      "step": 417
    },
    {
      "epoch": 0.13119384832672917,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.6478,
      "step": 418
    },
    {
      "epoch": 0.1315077092078936,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.5207,
      "step": 419
    },
    {
      "epoch": 0.13182157008905804,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3612,
      "step": 420
    },
    {
      "epoch": 0.13213543097022246,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.5294,
      "step": 421
    },
    {
      "epoch": 0.13244929185138687,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.5707,
      "step": 422
    },
    {
      "epoch": 0.1327631527325513,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.6781,
      "step": 423
    },
    {
      "epoch": 0.1330770136137157,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.5306,
      "step": 424
    },
    {
      "epoch": 0.13339087449488016,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.6768,
      "step": 425
    },
    {
      "epoch": 0.13370473537604458,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.7467,
      "step": 426
    },
    {
      "epoch": 0.134018596257209,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.5438,
      "step": 427
    },
    {
      "epoch": 0.1343324571383734,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.4618,
      "step": 428
    },
    {
      "epoch": 0.13464631801953783,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.926,
      "step": 429
    },
    {
      "epoch": 0.13496017890070228,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 2.0031,
      "step": 430
    },
    {
      "epoch": 0.1352740397818667,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 2.17,
      "step": 431
    },
    {
      "epoch": 0.1355879006630311,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.0235,
      "step": 432
    },
    {
      "epoch": 0.13590176154419553,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 2.2579,
      "step": 433
    },
    {
      "epoch": 0.13621562242535995,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 2.141,
      "step": 434
    },
    {
      "epoch": 0.1365294833065244,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 2.2934,
      "step": 435
    },
    {
      "epoch": 0.13684334418768881,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 2.5946,
      "step": 436
    },
    {
      "epoch": 0.13715720506885323,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 2.1508,
      "step": 437
    },
    {
      "epoch": 0.13747106595001765,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.1596,
      "step": 438
    },
    {
      "epoch": 0.13778492683118207,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 2.1806,
      "step": 439
    },
    {
      "epoch": 0.13809878771234652,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.7799,
      "step": 440
    },
    {
      "epoch": 0.13841264859351093,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.95,
      "step": 441
    },
    {
      "epoch": 0.13872650947467535,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.7952,
      "step": 442
    },
    {
      "epoch": 0.13904037035583977,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 2.338,
      "step": 443
    },
    {
      "epoch": 0.1393542312370042,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.9228,
      "step": 444
    },
    {
      "epoch": 0.13966809211816864,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 2.6961,
      "step": 445
    },
    {
      "epoch": 0.13998195299933305,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 2.2991,
      "step": 446
    },
    {
      "epoch": 0.14029581388049747,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 2.3858,
      "step": 447
    },
    {
      "epoch": 0.1406096747616619,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.1153,
      "step": 448
    },
    {
      "epoch": 0.1409235356428263,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 2.3709,
      "step": 449
    },
    {
      "epoch": 0.14123739652399075,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.7573,
      "step": 450
    },
    {
      "epoch": 0.14155125740515517,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.489,
      "step": 451
    },
    {
      "epoch": 0.1418651182863196,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.4093,
      "step": 452
    },
    {
      "epoch": 0.142178979167484,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.6179,
      "step": 453
    },
    {
      "epoch": 0.14249284004864843,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.56,
      "step": 454
    },
    {
      "epoch": 0.14280670092981287,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4018,
      "step": 455
    },
    {
      "epoch": 0.1431205618109773,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.4601,
      "step": 456
    },
    {
      "epoch": 0.1434344226921417,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.5944,
      "step": 457
    },
    {
      "epoch": 0.14374828357330613,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.4949,
      "step": 458
    },
    {
      "epoch": 0.14406214445447055,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.4648,
      "step": 459
    },
    {
      "epoch": 0.14437600533563497,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.544,
      "step": 460
    },
    {
      "epoch": 0.1446898662167994,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3839,
      "step": 461
    },
    {
      "epoch": 0.14500372709796383,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.4504,
      "step": 462
    },
    {
      "epoch": 0.14531758797912825,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.4584,
      "step": 463
    },
    {
      "epoch": 0.14563144886029267,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.427,
      "step": 464
    },
    {
      "epoch": 0.14594530974145709,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.451,
      "step": 465
    },
    {
      "epoch": 0.14625917062262153,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4653,
      "step": 466
    },
    {
      "epoch": 0.14657303150378595,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.5181,
      "step": 467
    },
    {
      "epoch": 0.14688689238495037,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3313,
      "step": 468
    },
    {
      "epoch": 0.1472007532661148,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.5084,
      "step": 469
    },
    {
      "epoch": 0.1475146141472792,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.5112,
      "step": 470
    },
    {
      "epoch": 0.14782847502844365,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.5859,
      "step": 471
    },
    {
      "epoch": 0.14814233590960807,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.4651,
      "step": 472
    },
    {
      "epoch": 0.1484561967907725,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.9876,
      "step": 473
    },
    {
      "epoch": 0.1487700576719369,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.4273,
      "step": 474
    },
    {
      "epoch": 0.14908391855310132,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.7127,
      "step": 475
    },
    {
      "epoch": 0.14939777943426577,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.6046,
      "step": 476
    },
    {
      "epoch": 0.1497116403154302,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.8459,
      "step": 477
    },
    {
      "epoch": 0.1500255011965946,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.656,
      "step": 478
    },
    {
      "epoch": 0.15033936207775903,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 2.6007,
      "step": 479
    },
    {
      "epoch": 0.15065322295892344,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.8374,
      "step": 480
    },
    {
      "epoch": 0.1509670838400879,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.9358,
      "step": 481
    },
    {
      "epoch": 0.1512809447212523,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 2.38,
      "step": 482
    },
    {
      "epoch": 0.15159480560241673,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.0626,
      "step": 483
    },
    {
      "epoch": 0.15190866648358115,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 2.383,
      "step": 484
    },
    {
      "epoch": 0.15222252736474556,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.1199,
      "step": 485
    },
    {
      "epoch": 0.15253638824591,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 2.2974,
      "step": 486
    },
    {
      "epoch": 0.15285024912707443,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 2.605,
      "step": 487
    },
    {
      "epoch": 0.15316411000823885,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 2.3385,
      "step": 488
    },
    {
      "epoch": 0.15347797088940326,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 2.0768,
      "step": 489
    },
    {
      "epoch": 0.15379183177056768,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.1478,
      "step": 490
    },
    {
      "epoch": 0.15410569265173213,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 2.6997,
      "step": 491
    },
    {
      "epoch": 0.15441955353289655,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.8556,
      "step": 492
    },
    {
      "epoch": 0.15473341441406097,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.7356,
      "step": 493
    },
    {
      "epoch": 0.15504727529522538,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 2.4964,
      "step": 494
    },
    {
      "epoch": 0.1553611361763898,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 2.2954,
      "step": 495
    },
    {
      "epoch": 0.15567499705755425,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 2.0918,
      "step": 496
    },
    {
      "epoch": 0.15598885793871867,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.8699,
      "step": 497
    },
    {
      "epoch": 0.15630271881988309,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 2.4207,
      "step": 498
    },
    {
      "epoch": 0.1566165797010475,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 2.2597,
      "step": 499
    },
    {
      "epoch": 0.15693044058221192,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0002,
      "loss": 3.121,
      "step": 500
    },
    {
      "epoch": 0.15724430146337637,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.4077,
      "step": 501
    },
    {
      "epoch": 0.1575581623445408,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.5515,
      "step": 502
    },
    {
      "epoch": 0.1578720232257052,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4571,
      "step": 503
    },
    {
      "epoch": 0.15818588410686962,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.4592,
      "step": 504
    },
    {
      "epoch": 0.15849974498803404,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4867,
      "step": 505
    },
    {
      "epoch": 0.1588136058691985,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.3015,
      "step": 506
    },
    {
      "epoch": 0.1591274667503629,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.4929,
      "step": 507
    },
    {
      "epoch": 0.15944132763152732,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.407,
      "step": 508
    },
    {
      "epoch": 0.15975518851269174,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.5229,
      "step": 509
    },
    {
      "epoch": 0.16006904939385616,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.5365,
      "step": 510
    },
    {
      "epoch": 0.1603829102750206,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.3955,
      "step": 511
    },
    {
      "epoch": 0.16069677115618503,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.1882,
      "step": 512
    },
    {
      "epoch": 0.16101063203734944,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.518,
      "step": 513
    },
    {
      "epoch": 0.16132449291851386,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3745,
      "step": 514
    },
    {
      "epoch": 0.16163835379967828,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.6599,
      "step": 515
    },
    {
      "epoch": 0.16195221468084273,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.4509,
      "step": 516
    },
    {
      "epoch": 0.16226607556200714,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.7153,
      "step": 517
    },
    {
      "epoch": 0.16257993644317156,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.5206,
      "step": 518
    },
    {
      "epoch": 0.16289379732433598,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.627,
      "step": 519
    },
    {
      "epoch": 0.1632076582055004,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3938,
      "step": 520
    },
    {
      "epoch": 0.16352151908666485,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.4612,
      "step": 521
    },
    {
      "epoch": 0.16383537996782926,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.4233,
      "step": 522
    },
    {
      "epoch": 0.16414924084899368,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.4416,
      "step": 523
    },
    {
      "epoch": 0.1644631017301581,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 2.0068,
      "step": 524
    },
    {
      "epoch": 0.16477696261132252,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.4914,
      "step": 525
    },
    {
      "epoch": 0.16509082349248697,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.7713,
      "step": 526
    },
    {
      "epoch": 0.16540468437365138,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.949,
      "step": 527
    },
    {
      "epoch": 0.1657185452548158,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.9683,
      "step": 528
    },
    {
      "epoch": 0.16603240613598022,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.2225,
      "step": 529
    },
    {
      "epoch": 0.16634626701714464,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 2.2496,
      "step": 530
    },
    {
      "epoch": 0.16666012789830909,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.4913,
      "step": 531
    },
    {
      "epoch": 0.1669739887794735,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 2.0728,
      "step": 532
    },
    {
      "epoch": 0.16728784966063792,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.7206,
      "step": 533
    },
    {
      "epoch": 0.16760171054180234,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 2.1716,
      "step": 534
    },
    {
      "epoch": 0.16791557142296676,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 2.1605,
      "step": 535
    },
    {
      "epoch": 0.1682294323041312,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 2.0504,
      "step": 536
    },
    {
      "epoch": 0.16854329318529562,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.9017,
      "step": 537
    },
    {
      "epoch": 0.16885715406646004,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.2211,
      "step": 538
    },
    {
      "epoch": 0.16917101494762446,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.132,
      "step": 539
    },
    {
      "epoch": 0.16948487582878888,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.0407,
      "step": 540
    },
    {
      "epoch": 0.16979873670995332,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 2.3361,
      "step": 541
    },
    {
      "epoch": 0.17011259759111774,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 2.0002,
      "step": 542
    },
    {
      "epoch": 0.17042645847228216,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 2.2588,
      "step": 543
    },
    {
      "epoch": 0.17074031935344658,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 2.4182,
      "step": 544
    },
    {
      "epoch": 0.171054180234611,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002,
      "loss": 2.4358,
      "step": 545
    },
    {
      "epoch": 0.17136804111577544,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 2.3013,
      "step": 546
    },
    {
      "epoch": 0.17168190199693986,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 2.2304,
      "step": 547
    },
    {
      "epoch": 0.17199576287810428,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.9786,
      "step": 548
    },
    {
      "epoch": 0.1723096237592687,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 2.5713,
      "step": 549
    },
    {
      "epoch": 0.17262348464043312,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.4127,
      "step": 550
    },
    {
      "epoch": 0.17293734552159756,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4719,
      "step": 551
    },
    {
      "epoch": 0.17325120640276198,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4504,
      "step": 552
    },
    {
      "epoch": 0.1735650672839264,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.4127,
      "step": 553
    },
    {
      "epoch": 0.17387892816509082,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.5099,
      "step": 554
    },
    {
      "epoch": 0.17419278904625524,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.6744,
      "step": 555
    },
    {
      "epoch": 0.17450664992741968,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.4517,
      "step": 556
    },
    {
      "epoch": 0.1748205108085841,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.4458,
      "step": 557
    },
    {
      "epoch": 0.17513437168974852,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.3115,
      "step": 558
    },
    {
      "epoch": 0.17544823257091294,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4375,
      "step": 559
    },
    {
      "epoch": 0.17576209345207736,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.3176,
      "step": 560
    },
    {
      "epoch": 0.1760759543332418,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2423,
      "step": 561
    },
    {
      "epoch": 0.1760759543332418,
      "eval_loss": 1.8500475883483887,
      "eval_runtime": 123.2087,
      "eval_samples_per_second": 8.116,
      "eval_steps_per_second": 8.116,
      "step": 561
    },
    {
      "epoch": 0.1760759543332418,
      "mmlu_eval_accuracy": 0.42976112199050265,
      "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": 0.3448275862068966,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.2692307692307692,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.0,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.5,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.27906976744186046,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3448275862068966,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.65,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.6046511627906976,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.2857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.3155592114687122,
      "step": 561
    },
    {
      "epoch": 0.17638981521440622,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.5559,
      "step": 562
    },
    {
      "epoch": 0.17670367609557064,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.5174,
      "step": 563
    },
    {
      "epoch": 0.17701753697673506,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4279,
      "step": 564
    },
    {
      "epoch": 0.17733139785789948,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.5815,
      "step": 565
    },
    {
      "epoch": 0.17764525873906392,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.6468,
      "step": 566
    },
    {
      "epoch": 0.17795911962022834,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.6738,
      "step": 567
    },
    {
      "epoch": 0.17827298050139276,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4164,
      "step": 568
    },
    {
      "epoch": 0.17858684138255718,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.4132,
      "step": 569
    },
    {
      "epoch": 0.1789007022637216,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.5096,
      "step": 570
    },
    {
      "epoch": 0.17921456314488604,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.762,
      "step": 571
    },
    {
      "epoch": 0.17952842402605046,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.6598,
      "step": 572
    },
    {
      "epoch": 0.17984228490721488,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.4627,
      "step": 573
    },
    {
      "epoch": 0.1801561457883793,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.8129,
      "step": 574
    },
    {
      "epoch": 0.18047000666954371,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.5752,
      "step": 575
    },
    {
      "epoch": 0.18078386755070816,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.8668,
      "step": 576
    },
    {
      "epoch": 0.18109772843187258,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.8953,
      "step": 577
    },
    {
      "epoch": 0.181411589313037,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.1257,
      "step": 578
    },
    {
      "epoch": 0.18172545019420142,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.8638,
      "step": 579
    },
    {
      "epoch": 0.18203931107536583,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.9372,
      "step": 580
    },
    {
      "epoch": 0.18235317195653028,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.8314,
      "step": 581
    },
    {
      "epoch": 0.1826670328376947,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.9337,
      "step": 582
    },
    {
      "epoch": 0.18298089371885912,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 2.2673,
      "step": 583
    },
    {
      "epoch": 0.18329475460002354,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 2.1699,
      "step": 584
    },
    {
      "epoch": 0.18360861548118795,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 2.3086,
      "step": 585
    },
    {
      "epoch": 0.1839224763623524,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.2748,
      "step": 586
    },
    {
      "epoch": 0.18423633724351682,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 2.226,
      "step": 587
    },
    {
      "epoch": 0.18455019812468124,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.8794,
      "step": 588
    },
    {
      "epoch": 0.18486405900584565,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 2.7864,
      "step": 589
    },
    {
      "epoch": 0.18517791988701007,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 2.5467,
      "step": 590
    },
    {
      "epoch": 0.18549178076817452,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 2.1094,
      "step": 591
    },
    {
      "epoch": 0.18580564164933894,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.2149,
      "step": 592
    },
    {
      "epoch": 0.18611950253050336,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.5329,
      "step": 593
    },
    {
      "epoch": 0.18643336341166777,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.8252,
      "step": 594
    },
    {
      "epoch": 0.1867472242928322,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 2.2788,
      "step": 595
    },
    {
      "epoch": 0.18706108517399664,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 2.0439,
      "step": 596
    },
    {
      "epoch": 0.18737494605516106,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.8777,
      "step": 597
    },
    {
      "epoch": 0.18768880693632548,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 2.5919,
      "step": 598
    },
    {
      "epoch": 0.1880026678174899,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 2.3098,
      "step": 599
    },
    {
      "epoch": 0.1883165286986543,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 2.7208,
      "step": 600
    },
    {
      "epoch": 0.18863038957981876,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.5596,
      "step": 601
    },
    {
      "epoch": 0.18894425046098318,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.4349,
      "step": 602
    },
    {
      "epoch": 0.1892581113421476,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.5032,
      "step": 603
    },
    {
      "epoch": 0.189571972223312,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4491,
      "step": 604
    },
    {
      "epoch": 0.18988583310447643,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.4891,
      "step": 605
    },
    {
      "epoch": 0.19019969398564088,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.35,
      "step": 606
    },
    {
      "epoch": 0.1905135548668053,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.5648,
      "step": 607
    },
    {
      "epoch": 0.19082741574796971,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.3777,
      "step": 608
    },
    {
      "epoch": 0.19114127662913413,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3868,
      "step": 609
    },
    {
      "epoch": 0.19145513751029855,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.4256,
      "step": 610
    },
    {
      "epoch": 0.191768998391463,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.4123,
      "step": 611
    },
    {
      "epoch": 0.19208285927262742,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.5571,
      "step": 612
    },
    {
      "epoch": 0.19239672015379183,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.6238,
      "step": 613
    },
    {
      "epoch": 0.19271058103495625,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.5166,
      "step": 614
    },
    {
      "epoch": 0.19302444191612067,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4666,
      "step": 615
    },
    {
      "epoch": 0.19333830279728512,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.5169,
      "step": 616
    },
    {
      "epoch": 0.19365216367844953,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3763,
      "step": 617
    },
    {
      "epoch": 0.19396602455961395,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3067,
      "step": 618
    },
    {
      "epoch": 0.19427988544077837,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.6093,
      "step": 619
    },
    {
      "epoch": 0.1945937463219428,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.3269,
      "step": 620
    },
    {
      "epoch": 0.19490760720310724,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.4157,
      "step": 621
    },
    {
      "epoch": 0.19522146808427165,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.4526,
      "step": 622
    },
    {
      "epoch": 0.19553532896543607,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.6507,
      "step": 623
    },
    {
      "epoch": 0.1958491898466005,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.7315,
      "step": 624
    },
    {
      "epoch": 0.1961630507277649,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.7668,
      "step": 625
    },
    {
      "epoch": 0.19647691160892936,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.7066,
      "step": 626
    },
    {
      "epoch": 0.19679077249009377,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.5872,
      "step": 627
    },
    {
      "epoch": 0.1971046333712582,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.4538,
      "step": 628
    },
    {
      "epoch": 0.1974184942524226,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.8707,
      "step": 629
    },
    {
      "epoch": 0.19773235513358703,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 2.0015,
      "step": 630
    },
    {
      "epoch": 0.19804621601475147,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 2.126,
      "step": 631
    },
    {
      "epoch": 0.1983600768959159,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 2.274,
      "step": 632
    },
    {
      "epoch": 0.1986739377770803,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 2.1807,
      "step": 633
    },
    {
      "epoch": 0.19898779865824473,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 2.1142,
      "step": 634
    },
    {
      "epoch": 0.19930165953940915,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.6402,
      "step": 635
    },
    {
      "epoch": 0.1996155204205736,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 2.4704,
      "step": 636
    },
    {
      "epoch": 0.199929381301738,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 2.5016,
      "step": 637
    },
    {
      "epoch": 0.20024324218290243,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0002,
      "loss": 2.6837,
      "step": 638
    },
    {
      "epoch": 0.20055710306406685,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.4928,
      "step": 639
    },
    {
      "epoch": 0.20087096394523127,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 2.2994,
      "step": 640
    },
    {
      "epoch": 0.20118482482639571,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.6428,
      "step": 641
    },
    {
      "epoch": 0.20149868570756013,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 2.0929,
      "step": 642
    },
    {
      "epoch": 0.20181254658872455,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.5125,
      "step": 643
    },
    {
      "epoch": 0.20212640746988897,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.3,
      "step": 644
    },
    {
      "epoch": 0.2024402683510534,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.3705,
      "step": 645
    },
    {
      "epoch": 0.2027541292322178,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.7512,
      "step": 646
    },
    {
      "epoch": 0.20306799011338225,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 2.19,
      "step": 647
    },
    {
      "epoch": 0.20338185099454667,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 2.2673,
      "step": 648
    },
    {
      "epoch": 0.2036957118757111,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.8386,
      "step": 649
    },
    {
      "epoch": 0.2040095727568755,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 3.1839,
      "step": 650
    },
    {
      "epoch": 0.20432343363803993,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0002,
      "loss": 1.381,
      "step": 651
    },
    {
      "epoch": 0.20463729451920437,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2011,
      "step": 652
    },
    {
      "epoch": 0.2049511554003688,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.479,
      "step": 653
    },
    {
      "epoch": 0.2052650162815332,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4244,
      "step": 654
    },
    {
      "epoch": 0.20557887716269763,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.5518,
      "step": 655
    },
    {
      "epoch": 0.20589273804386204,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4713,
      "step": 656
    },
    {
      "epoch": 0.2062065989250265,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.537,
      "step": 657
    },
    {
      "epoch": 0.2065204598061909,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.376,
      "step": 658
    },
    {
      "epoch": 0.20683432068735533,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.5974,
      "step": 659
    },
    {
      "epoch": 0.20714818156851975,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.5159,
      "step": 660
    },
    {
      "epoch": 0.20746204244968416,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.4994,
      "step": 661
    },
    {
      "epoch": 0.2077759033308486,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.4817,
      "step": 662
    },
    {
      "epoch": 0.20808976421201303,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4856,
      "step": 663
    },
    {
      "epoch": 0.20840362509317745,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.5026,
      "step": 664
    },
    {
      "epoch": 0.20871748597434187,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.4981,
      "step": 665
    },
    {
      "epoch": 0.20903134685550628,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3784,
      "step": 666
    },
    {
      "epoch": 0.20934520773667073,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2459,
      "step": 667
    },
    {
      "epoch": 0.20965906861783515,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.5568,
      "step": 668
    },
    {
      "epoch": 0.20997292949899957,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.6353,
      "step": 669
    },
    {
      "epoch": 0.21028679038016398,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.5004,
      "step": 670
    },
    {
      "epoch": 0.2106006512613284,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 2.0955,
      "step": 671
    },
    {
      "epoch": 0.21091451214249285,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.6436,
      "step": 672
    },
    {
      "epoch": 0.21122837302365727,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 1.5068,
      "step": 673
    },
    {
      "epoch": 0.21154223390482169,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.7045,
      "step": 674
    },
    {
      "epoch": 0.2118560947859861,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.6256,
      "step": 675
    },
    {
      "epoch": 0.21216995566715052,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.9182,
      "step": 676
    },
    {
      "epoch": 0.21248381654831497,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 2.1139,
      "step": 677
    },
    {
      "epoch": 0.2127976774294794,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.9642,
      "step": 678
    },
    {
      "epoch": 0.2131115383106438,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.7713,
      "step": 679
    },
    {
      "epoch": 0.21342539919180822,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 2.1779,
      "step": 680
    },
    {
      "epoch": 0.21373926007297264,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.6468,
      "step": 681
    },
    {
      "epoch": 0.2140531209541371,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 2.0927,
      "step": 682
    },
    {
      "epoch": 0.2143669818353015,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.1903,
      "step": 683
    },
    {
      "epoch": 0.21468084271646592,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.4789,
      "step": 684
    },
    {
      "epoch": 0.21499470359763034,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 2.1381,
      "step": 685
    },
    {
      "epoch": 0.21530856447879476,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 2.0279,
      "step": 686
    },
    {
      "epoch": 0.2156224253599592,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.5657,
      "step": 687
    },
    {
      "epoch": 0.21593628624112363,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 2.4605,
      "step": 688
    },
    {
      "epoch": 0.21625014712228804,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.2759,
      "step": 689
    },
    {
      "epoch": 0.21656400800345246,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 2.2051,
      "step": 690
    },
    {
      "epoch": 0.21687786888461688,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 2.3526,
      "step": 691
    },
    {
      "epoch": 0.21719172976578133,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.3279,
      "step": 692
    },
    {
      "epoch": 0.21750559064694575,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.2478,
      "step": 693
    },
    {
      "epoch": 0.21781945152811016,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 2.5623,
      "step": 694
    },
    {
      "epoch": 0.21813331240927458,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.5045,
      "step": 695
    },
    {
      "epoch": 0.218447173290439,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 2.1917,
      "step": 696
    },
    {
      "epoch": 0.21876103417160345,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002,
      "loss": 2.473,
      "step": 697
    },
    {
      "epoch": 0.21907489505276787,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 2.4232,
      "step": 698
    },
    {
      "epoch": 0.21938875593393228,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.1088,
      "step": 699
    },
    {
      "epoch": 0.2197026168150967,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 3.1141,
      "step": 700
    },
    {
      "epoch": 0.22001647769626112,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0002,
      "loss": 1.4641,
      "step": 701
    },
    {
      "epoch": 0.22033033857742557,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.5301,
      "step": 702
    },
    {
      "epoch": 0.22064419945858998,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.4995,
      "step": 703
    },
    {
      "epoch": 0.2209580603397544,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.5175,
      "step": 704
    },
    {
      "epoch": 0.22127192122091882,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.5367,
      "step": 705
    },
    {
      "epoch": 0.22158578210208324,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4447,
      "step": 706
    },
    {
      "epoch": 0.22189964298324769,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.3938,
      "step": 707
    },
    {
      "epoch": 0.2222135038644121,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4849,
      "step": 708
    },
    {
      "epoch": 0.22252736474557652,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.4433,
      "step": 709
    },
    {
      "epoch": 0.22284122562674094,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.455,
      "step": 710
    },
    {
      "epoch": 0.22315508650790536,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3936,
      "step": 711
    },
    {
      "epoch": 0.2234689473890698,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.6169,
      "step": 712
    },
    {
      "epoch": 0.22378280827023422,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.526,
      "step": 713
    },
    {
      "epoch": 0.22409666915139864,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.4675,
      "step": 714
    },
    {
      "epoch": 0.22441053003256306,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4285,
      "step": 715
    },
    {
      "epoch": 0.22472439091372748,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3881,
      "step": 716
    },
    {
      "epoch": 0.22503825179489192,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.4186,
      "step": 717
    },
    {
      "epoch": 0.22535211267605634,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4384,
      "step": 718
    },
    {
      "epoch": 0.22566597355722076,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.5125,
      "step": 719
    },
    {
      "epoch": 0.22597983443838518,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4096,
      "step": 720
    },
    {
      "epoch": 0.2262936953195496,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.5214,
      "step": 721
    },
    {
      "epoch": 0.22660755620071404,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.5779,
      "step": 722
    },
    {
      "epoch": 0.22692141708187846,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3637,
      "step": 723
    },
    {
      "epoch": 0.22723527796304288,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.4644,
      "step": 724
    },
    {
      "epoch": 0.2275491388442073,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.8246,
      "step": 725
    },
    {
      "epoch": 0.22786299972537172,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.8298,
      "step": 726
    },
    {
      "epoch": 0.22817686060653616,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 2.0097,
      "step": 727
    },
    {
      "epoch": 0.22849072148770058,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.8344,
      "step": 728
    },
    {
      "epoch": 0.228804582368865,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.8355,
      "step": 729
    },
    {
      "epoch": 0.22911844325002942,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 2.2533,
      "step": 730
    },
    {
      "epoch": 0.22943230413119384,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.7191,
      "step": 731
    },
    {
      "epoch": 0.22974616501235828,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 2.0636,
      "step": 732
    },
    {
      "epoch": 0.2300600258935227,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.1691,
      "step": 733
    },
    {
      "epoch": 0.23037388677468712,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 2.4264,
      "step": 734
    },
    {
      "epoch": 0.23068774765585154,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 2.3442,
      "step": 735
    },
    {
      "epoch": 0.23100160853701596,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 2.2909,
      "step": 736
    },
    {
      "epoch": 0.2313154694181804,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.2776,
      "step": 737
    },
    {
      "epoch": 0.23162933029934482,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 2.1652,
      "step": 738
    },
    {
      "epoch": 0.23194319118050924,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 2.3553,
      "step": 739
    },
    {
      "epoch": 0.23225705206167366,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 2.2077,
      "step": 740
    },
    {
      "epoch": 0.23257091294283808,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.9861,
      "step": 741
    },
    {
      "epoch": 0.23288477382400252,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.9471,
      "step": 742
    },
    {
      "epoch": 0.23319863470516694,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.9327,
      "step": 743
    },
    {
      "epoch": 0.23351249558633136,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 2.4741,
      "step": 744
    },
    {
      "epoch": 0.23382635646749578,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.1698,
      "step": 745
    },
    {
      "epoch": 0.2341402173486602,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 2.3602,
      "step": 746
    },
    {
      "epoch": 0.23445407822982464,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.7317,
      "step": 747
    },
    {
      "epoch": 0.23476793911098906,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.8612,
      "step": 748
    },
    {
      "epoch": 0.23476793911098906,
      "eval_loss": 1.8301838636398315,
      "eval_runtime": 123.7044,
      "eval_samples_per_second": 8.084,
      "eval_steps_per_second": 8.084,
      "step": 748
    },
    {
      "epoch": 0.23476793911098906,
      "mmlu_eval_accuracy": 0.43586306684768467,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "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": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.5,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3448275862068966,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.6166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
      "mmlu_eval_accuracy_human_aging": 0.5652173913043478,
      "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": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.5348837209302325,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.37142857142857144,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_psychology": 0.3333333333333333,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.6818181818181818,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 0.9066392678516819,
      "step": 748
    },
    {
      "epoch": 0.23508179999215348,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 2.4457,
      "step": 749
    },
    {
      "epoch": 0.2353956608733179,
      "grad_norm": 0.84375,
      "learning_rate": 0.0002,
      "loss": 3.3922,
      "step": 750
    },
    {
      "epoch": 0.23570952175448232,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 1.3832,
      "step": 751
    },
    {
      "epoch": 0.23602338263564676,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4088,
      "step": 752
    },
    {
      "epoch": 0.23633724351681118,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4099,
      "step": 753
    },
    {
      "epoch": 0.2366511043979756,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.494,
      "step": 754
    },
    {
      "epoch": 0.23696496527914002,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.465,
      "step": 755
    },
    {
      "epoch": 0.23727882616030443,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.5089,
      "step": 756
    },
    {
      "epoch": 0.23759268704146888,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.5011,
      "step": 757
    },
    {
      "epoch": 0.2379065479226333,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.3608,
      "step": 758
    },
    {
      "epoch": 0.23822040880379772,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.392,
      "step": 759
    },
    {
      "epoch": 0.23853426968496214,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.5432,
      "step": 760
    },
    {
      "epoch": 0.23884813056612655,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.594,
      "step": 761
    },
    {
      "epoch": 0.239161991447291,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.57,
      "step": 762
    },
    {
      "epoch": 0.23947585232845542,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.407,
      "step": 763
    },
    {
      "epoch": 0.23978971320961984,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.3097,
      "step": 764
    },
    {
      "epoch": 0.24010357409078426,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.5793,
      "step": 765
    },
    {
      "epoch": 0.24041743497194867,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.2891,
      "step": 766
    },
    {
      "epoch": 0.24073129585311312,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3422,
      "step": 767
    },
    {
      "epoch": 0.24104515673427754,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.614,
      "step": 768
    },
    {
      "epoch": 0.24135901761544196,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.4728,
      "step": 769
    },
    {
      "epoch": 0.24167287849660637,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.45,
      "step": 770
    },
    {
      "epoch": 0.2419867393777708,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.8436,
      "step": 771
    },
    {
      "epoch": 0.24230060025893524,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.9457,
      "step": 772
    },
    {
      "epoch": 0.24261446114009966,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.6809,
      "step": 773
    },
    {
      "epoch": 0.24292832202126408,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.3838,
      "step": 774
    },
    {
      "epoch": 0.2432421829024285,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.446,
      "step": 775
    },
    {
      "epoch": 0.2435560437835929,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.7502,
      "step": 776
    },
    {
      "epoch": 0.24386990466475736,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 2.0686,
      "step": 777
    },
    {
      "epoch": 0.24418376554592178,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.6674,
      "step": 778
    },
    {
      "epoch": 0.2444976264270862,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 2.1236,
      "step": 779
    },
    {
      "epoch": 0.2448114873082506,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.2482,
      "step": 780
    },
    {
      "epoch": 0.24512534818941503,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.8243,
      "step": 781
    },
    {
      "epoch": 0.24543920907057948,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 2.0005,
      "step": 782
    },
    {
      "epoch": 0.2457530699517439,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.8594,
      "step": 783
    },
    {
      "epoch": 0.24606693083290831,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.9771,
      "step": 784
    },
    {
      "epoch": 0.24638079171407273,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 2.1506,
      "step": 785
    },
    {
      "epoch": 0.24669465259523715,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 2.0608,
      "step": 786
    },
    {
      "epoch": 0.2470085134764016,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.2639,
      "step": 787
    },
    {
      "epoch": 0.24732237435756602,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 2.4051,
      "step": 788
    },
    {
      "epoch": 0.24763623523873043,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 2.3805,
      "step": 789
    },
    {
      "epoch": 0.24795009611989485,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.2338,
      "step": 790
    },
    {
      "epoch": 0.24826395700105927,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.2317,
      "step": 791
    },
    {
      "epoch": 0.24857781788222372,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.3565,
      "step": 792
    },
    {
      "epoch": 0.24889167876338814,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 3.0328,
      "step": 793
    },
    {
      "epoch": 0.24920553964455255,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.253,
      "step": 794
    },
    {
      "epoch": 0.24951940052571697,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 2.3392,
      "step": 795
    },
    {
      "epoch": 0.2498332614068814,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 2.2872,
      "step": 796
    },
    {
      "epoch": 0.2501471222880458,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 2.3179,
      "step": 797
    },
    {
      "epoch": 0.2504609831692102,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 2.6253,
      "step": 798
    },
    {
      "epoch": 0.25077484405037465,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 2.6689,
      "step": 799
    },
    {
      "epoch": 0.2510887049315391,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 2.76,
      "step": 800
    },
    {
      "epoch": 0.25140256581270354,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.4852,
      "step": 801
    },
    {
      "epoch": 0.25171642669386796,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.4893,
      "step": 802
    },
    {
      "epoch": 0.2520302875750324,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.4828,
      "step": 803
    },
    {
      "epoch": 0.2523441484561968,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4697,
      "step": 804
    },
    {
      "epoch": 0.2526580093373612,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.4742,
      "step": 805
    },
    {
      "epoch": 0.25297187021852563,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.4976,
      "step": 806
    },
    {
      "epoch": 0.25328573109969005,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.4181,
      "step": 807
    },
    {
      "epoch": 0.25359959198085447,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.521,
      "step": 808
    },
    {
      "epoch": 0.2539134528620189,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.5012,
      "step": 809
    },
    {
      "epoch": 0.25422731374318336,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.406,
      "step": 810
    },
    {
      "epoch": 0.2545411746243478,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.463,
      "step": 811
    },
    {
      "epoch": 0.2548550355055122,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.5657,
      "step": 812
    },
    {
      "epoch": 0.2551688963866766,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4766,
      "step": 813
    },
    {
      "epoch": 0.25548275726784103,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.5833,
      "step": 814
    },
    {
      "epoch": 0.25579661814900545,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3477,
      "step": 815
    },
    {
      "epoch": 0.25611047903016987,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4316,
      "step": 816
    },
    {
      "epoch": 0.2564243399113343,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.442,
      "step": 817
    },
    {
      "epoch": 0.2567382007924987,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.3093,
      "step": 818
    },
    {
      "epoch": 0.2570520616736631,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.5933,
      "step": 819
    },
    {
      "epoch": 0.2573659225548276,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.439,
      "step": 820
    },
    {
      "epoch": 0.257679783435992,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.4433,
      "step": 821
    },
    {
      "epoch": 0.25799364431715643,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.5624,
      "step": 822
    },
    {
      "epoch": 0.25830750519832085,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.4514,
      "step": 823
    },
    {
      "epoch": 0.25862136607948527,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.4493,
      "step": 824
    },
    {
      "epoch": 0.2589352269606497,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.5928,
      "step": 825
    },
    {
      "epoch": 0.2592490878418141,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.5877,
      "step": 826
    },
    {
      "epoch": 0.2595629487229785,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.7294,
      "step": 827
    },
    {
      "epoch": 0.25987680960414294,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.7468,
      "step": 828
    },
    {
      "epoch": 0.26019067048530736,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 2.0077,
      "step": 829
    },
    {
      "epoch": 0.26050453136647184,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 2.0081,
      "step": 830
    },
    {
      "epoch": 0.26081839224763625,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.8417,
      "step": 831
    },
    {
      "epoch": 0.2611322531288007,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.9381,
      "step": 832
    },
    {
      "epoch": 0.2614461140099651,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 2.2372,
      "step": 833
    },
    {
      "epoch": 0.2617599748911295,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.2899,
      "step": 834
    },
    {
      "epoch": 0.26207383577229393,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 2.0124,
      "step": 835
    },
    {
      "epoch": 0.26238769665345835,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.9612,
      "step": 836
    },
    {
      "epoch": 0.26270155753462276,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 2.2552,
      "step": 837
    },
    {
      "epoch": 0.2630154184157872,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 2.4561,
      "step": 838
    },
    {
      "epoch": 0.2633292792969516,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 2.6507,
      "step": 839
    },
    {
      "epoch": 0.2636431401781161,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 2.3272,
      "step": 840
    },
    {
      "epoch": 0.2639570010592805,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.8104,
      "step": 841
    },
    {
      "epoch": 0.2642708619404449,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002,
      "loss": 2.3128,
      "step": 842
    },
    {
      "epoch": 0.26458472282160933,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.4287,
      "step": 843
    },
    {
      "epoch": 0.26489858370277375,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 2.56,
      "step": 844
    },
    {
      "epoch": 0.26521244458393817,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0002,
      "loss": 2.6142,
      "step": 845
    },
    {
      "epoch": 0.2655263054651026,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.8461,
      "step": 846
    },
    {
      "epoch": 0.265840166346267,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.9298,
      "step": 847
    },
    {
      "epoch": 0.2661540272274314,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.9589,
      "step": 848
    },
    {
      "epoch": 0.26646788810859584,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.3589,
      "step": 849
    },
    {
      "epoch": 0.2667817489897603,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 2.5415,
      "step": 850
    },
    {
      "epoch": 0.26709560987092473,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 1.3115,
      "step": 851
    },
    {
      "epoch": 0.26740947075208915,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.4415,
      "step": 852
    },
    {
      "epoch": 0.26772333163325357,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3668,
      "step": 853
    },
    {
      "epoch": 0.268037192514418,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.5009,
      "step": 854
    },
    {
      "epoch": 0.2683510533955824,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.6886,
      "step": 855
    },
    {
      "epoch": 0.2686649142767468,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.4729,
      "step": 856
    },
    {
      "epoch": 0.26897877515791124,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.4901,
      "step": 857
    },
    {
      "epoch": 0.26929263603907566,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.539,
      "step": 858
    },
    {
      "epoch": 0.2696064969202401,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.4547,
      "step": 859
    },
    {
      "epoch": 0.26992035780140455,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.4896,
      "step": 860
    },
    {
      "epoch": 0.27023421868256897,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3954,
      "step": 861
    },
    {
      "epoch": 0.2705480795637334,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.5198,
      "step": 862
    },
    {
      "epoch": 0.2708619404448978,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.5698,
      "step": 863
    },
    {
      "epoch": 0.2711758013260622,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4155,
      "step": 864
    },
    {
      "epoch": 0.27148966220722665,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.5329,
      "step": 865
    },
    {
      "epoch": 0.27180352308839106,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.4575,
      "step": 866
    },
    {
      "epoch": 0.2721173839695555,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.4322,
      "step": 867
    },
    {
      "epoch": 0.2724312448507199,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3853,
      "step": 868
    },
    {
      "epoch": 0.2727451057318843,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.7375,
      "step": 869
    },
    {
      "epoch": 0.2730589666130488,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.5402,
      "step": 870
    },
    {
      "epoch": 0.2733728274942132,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.6293,
      "step": 871
    },
    {
      "epoch": 0.27368668837537763,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.5219,
      "step": 872
    },
    {
      "epoch": 0.27400054925654205,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.6687,
      "step": 873
    },
    {
      "epoch": 0.27431441013770647,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.7818,
      "step": 874
    },
    {
      "epoch": 0.2746282710188709,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.887,
      "step": 875
    },
    {
      "epoch": 0.2749421319000353,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.7887,
      "step": 876
    },
    {
      "epoch": 0.2752559927811997,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.6122,
      "step": 877
    },
    {
      "epoch": 0.27556985366236414,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.6951,
      "step": 878
    },
    {
      "epoch": 0.27588371454352856,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 2.0917,
      "step": 879
    },
    {
      "epoch": 0.27619757542469303,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 2.3369,
      "step": 880
    },
    {
      "epoch": 0.27651143630585745,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 2.1028,
      "step": 881
    },
    {
      "epoch": 0.27682529718702187,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.8433,
      "step": 882
    },
    {
      "epoch": 0.2771391580681863,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 2.2209,
      "step": 883
    },
    {
      "epoch": 0.2774530189493507,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.4793,
      "step": 884
    },
    {
      "epoch": 0.2777668798305151,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 2.1978,
      "step": 885
    },
    {
      "epoch": 0.27808074071167954,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 2.6263,
      "step": 886
    },
    {
      "epoch": 0.27839460159284396,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 2.1229,
      "step": 887
    },
    {
      "epoch": 0.2787084624740084,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 2.3003,
      "step": 888
    },
    {
      "epoch": 0.2790223233551728,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 2.4488,
      "step": 889
    },
    {
      "epoch": 0.27933618423633727,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 2.2592,
      "step": 890
    },
    {
      "epoch": 0.2796500451175017,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 2.0843,
      "step": 891
    },
    {
      "epoch": 0.2799639059986661,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 2.2622,
      "step": 892
    },
    {
      "epoch": 0.2802777668798305,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 2.2364,
      "step": 893
    },
    {
      "epoch": 0.28059162776099494,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 2.0579,
      "step": 894
    },
    {
      "epoch": 0.28090548864215936,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 2.7902,
      "step": 895
    },
    {
      "epoch": 0.2812193495233238,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.2274,
      "step": 896
    },
    {
      "epoch": 0.2815332104044882,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.9381,
      "step": 897
    },
    {
      "epoch": 0.2818470712856526,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 2.208,
      "step": 898
    },
    {
      "epoch": 0.28216093216681704,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 2.189,
      "step": 899
    },
    {
      "epoch": 0.2824747930479815,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.8205,
      "step": 900
    },
    {
      "epoch": 0.28278865392914593,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 1.3313,
      "step": 901
    },
    {
      "epoch": 0.28310251481031035,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.4316,
      "step": 902
    },
    {
      "epoch": 0.28341637569147476,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.6624,
      "step": 903
    },
    {
      "epoch": 0.2837302365726392,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4856,
      "step": 904
    },
    {
      "epoch": 0.2840440974538036,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3935,
      "step": 905
    },
    {
      "epoch": 0.284357958334968,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.5396,
      "step": 906
    },
    {
      "epoch": 0.28467181921613244,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.51,
      "step": 907
    },
    {
      "epoch": 0.28498568009729686,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.4975,
      "step": 908
    },
    {
      "epoch": 0.2852995409784613,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.6255,
      "step": 909
    },
    {
      "epoch": 0.28561340185962575,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4139,
      "step": 910
    },
    {
      "epoch": 0.28592726274079017,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.4564,
      "step": 911
    },
    {
      "epoch": 0.2862411236219546,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.5092,
      "step": 912
    },
    {
      "epoch": 0.286554984503119,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.3182,
      "step": 913
    },
    {
      "epoch": 0.2868688453842834,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3046,
      "step": 914
    },
    {
      "epoch": 0.28718270626544784,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.34,
      "step": 915
    },
    {
      "epoch": 0.28749656714661226,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4487,
      "step": 916
    },
    {
      "epoch": 0.2878104280277767,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.4902,
      "step": 917
    },
    {
      "epoch": 0.2881242889089411,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4275,
      "step": 918
    },
    {
      "epoch": 0.2884381497901055,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.5213,
      "step": 919
    },
    {
      "epoch": 0.28875201067126993,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1849,
      "step": 920
    },
    {
      "epoch": 0.2890658715524344,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.6107,
      "step": 921
    },
    {
      "epoch": 0.2893797324335988,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.7007,
      "step": 922
    },
    {
      "epoch": 0.28969359331476324,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.6828,
      "step": 923
    },
    {
      "epoch": 0.29000745419592766,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.8089,
      "step": 924
    },
    {
      "epoch": 0.2903213150770921,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.8047,
      "step": 925
    },
    {
      "epoch": 0.2906351759582565,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.6016,
      "step": 926
    },
    {
      "epoch": 0.2909490368394209,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 2.0681,
      "step": 927
    },
    {
      "epoch": 0.29126289772058533,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.8313,
      "step": 928
    },
    {
      "epoch": 0.29157675860174975,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 2.4114,
      "step": 929
    },
    {
      "epoch": 0.29189061948291417,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 2.2211,
      "step": 930
    },
    {
      "epoch": 0.29220448036407864,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.9012,
      "step": 931
    },
    {
      "epoch": 0.29251834124524306,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.1886,
      "step": 932
    },
    {
      "epoch": 0.2928322021264075,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 2.1408,
      "step": 933
    },
    {
      "epoch": 0.2931460630075719,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 2.3237,
      "step": 934
    },
    {
      "epoch": 0.2934599238887363,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 2.4881,
      "step": 935
    },
    {
      "epoch": 0.2934599238887363,
      "eval_loss": 1.8006056547164917,
      "eval_runtime": 149.6622,
      "eval_samples_per_second": 6.682,
      "eval_steps_per_second": 6.682,
      "step": 935
    },
    {
      "epoch": 0.2934599238887363,
      "mmlu_eval_accuracy": 0.42205537825180356,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3103448275862069,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.2692307692307692,
      "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": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3448275862068966,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.6166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.2727272727272727,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5813953488372093,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.48484848484848486,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.3333333333333333,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.0942840913905574,
      "step": 935
    },
    {
      "epoch": 0.29377378476990074,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 2.1141,
      "step": 936
    },
    {
      "epoch": 0.29408764565106515,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 2.6761,
      "step": 937
    },
    {
      "epoch": 0.2944015065322296,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 2.0653,
      "step": 938
    },
    {
      "epoch": 0.294715367413394,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 2.1641,
      "step": 939
    },
    {
      "epoch": 0.2950292282945584,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 2.2414,
      "step": 940
    },
    {
      "epoch": 0.2953430891757229,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.3736,
      "step": 941
    },
    {
      "epoch": 0.2956569500568873,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.3303,
      "step": 942
    },
    {
      "epoch": 0.2959708109380517,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 2.3767,
      "step": 943
    },
    {
      "epoch": 0.29628467181921614,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 2.5385,
      "step": 944
    },
    {
      "epoch": 0.29659853270038056,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.4962,
      "step": 945
    },
    {
      "epoch": 0.296912393581545,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 2.0142,
      "step": 946
    },
    {
      "epoch": 0.2972262544627094,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 2.0577,
      "step": 947
    },
    {
      "epoch": 0.2975401153438738,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 2.6295,
      "step": 948
    },
    {
      "epoch": 0.29785397622503823,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.9394,
      "step": 949
    },
    {
      "epoch": 0.29816783710620265,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 2.9883,
      "step": 950
    },
    {
      "epoch": 0.2984816979873671,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 1.4738,
      "step": 951
    },
    {
      "epoch": 0.29879555886853154,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.43,
      "step": 952
    },
    {
      "epoch": 0.29910941974969596,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.4107,
      "step": 953
    },
    {
      "epoch": 0.2994232806308604,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3789,
      "step": 954
    },
    {
      "epoch": 0.2997371415120248,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4341,
      "step": 955
    },
    {
      "epoch": 0.3000510023931892,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.5085,
      "step": 956
    },
    {
      "epoch": 0.30036486327435363,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3617,
      "step": 957
    },
    {
      "epoch": 0.30067872415551805,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.5131,
      "step": 958
    },
    {
      "epoch": 0.30099258503668247,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.6087,
      "step": 959
    },
    {
      "epoch": 0.3013064459178469,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.3892,
      "step": 960
    },
    {
      "epoch": 0.30162030679901136,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.372,
      "step": 961
    },
    {
      "epoch": 0.3019341676801758,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.3744,
      "step": 962
    },
    {
      "epoch": 0.3022480285613402,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.4449,
      "step": 963
    },
    {
      "epoch": 0.3025618894425046,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4368,
      "step": 964
    },
    {
      "epoch": 0.30287575032366904,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.2957,
      "step": 965
    },
    {
      "epoch": 0.30318961120483345,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.467,
      "step": 966
    },
    {
      "epoch": 0.30350347208599787,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.4731,
      "step": 967
    },
    {
      "epoch": 0.3038173329671623,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.4283,
      "step": 968
    },
    {
      "epoch": 0.3041311938483267,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.664,
      "step": 969
    },
    {
      "epoch": 0.3044450547294911,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.5262,
      "step": 970
    },
    {
      "epoch": 0.3047589156106556,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.361,
      "step": 971
    },
    {
      "epoch": 0.30507277649182,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.559,
      "step": 972
    },
    {
      "epoch": 0.30538663737298444,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.6235,
      "step": 973
    },
    {
      "epoch": 0.30570049825414886,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.7893,
      "step": 974
    },
    {
      "epoch": 0.3060143591353133,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.759,
      "step": 975
    },
    {
      "epoch": 0.3063282200164777,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.7774,
      "step": 976
    },
    {
      "epoch": 0.3066420808976421,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.7255,
      "step": 977
    },
    {
      "epoch": 0.30695594177880653,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.8652,
      "step": 978
    },
    {
      "epoch": 0.30726980265997095,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.5051,
      "step": 979
    },
    {
      "epoch": 0.30758366354113537,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 2.0593,
      "step": 980
    },
    {
      "epoch": 0.30789752442229984,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.689,
      "step": 981
    },
    {
      "epoch": 0.30821138530346426,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.5818,
      "step": 982
    },
    {
      "epoch": 0.3085252461846287,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.8198,
      "step": 983
    },
    {
      "epoch": 0.3088391070657931,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 2.2934,
      "step": 984
    },
    {
      "epoch": 0.3091529679469575,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 2.2926,
      "step": 985
    },
    {
      "epoch": 0.30946682882812193,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.5109,
      "step": 986
    },
    {
      "epoch": 0.30978068970928635,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 2.0915,
      "step": 987
    },
    {
      "epoch": 0.31009455059045077,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 2.1578,
      "step": 988
    },
    {
      "epoch": 0.3104084114716152,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 2.6873,
      "step": 989
    },
    {
      "epoch": 0.3107222723527796,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.9798,
      "step": 990
    },
    {
      "epoch": 0.3110361332339441,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.9699,
      "step": 991
    },
    {
      "epoch": 0.3113499941151085,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 2.0929,
      "step": 992
    },
    {
      "epoch": 0.3116638549962729,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.7644,
      "step": 993
    },
    {
      "epoch": 0.31197771587743733,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 2.3392,
      "step": 994
    },
    {
      "epoch": 0.31229157675860175,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.7899,
      "step": 995
    },
    {
      "epoch": 0.31260543763976617,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 2.5258,
      "step": 996
    },
    {
      "epoch": 0.3129192985209306,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 2.4348,
      "step": 997
    },
    {
      "epoch": 0.313233159402095,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 2.0247,
      "step": 998
    },
    {
      "epoch": 0.3135470202832594,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.4701,
      "step": 999
    },
    {
      "epoch": 0.31386088116442384,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.5099,
      "step": 1000
    },
    {
      "epoch": 0.3141747420455883,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0002,
      "loss": 1.4304,
      "step": 1001
    },
    {
      "epoch": 0.31448860292675274,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3457,
      "step": 1002
    },
    {
      "epoch": 0.31480246380791715,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.5733,
      "step": 1003
    },
    {
      "epoch": 0.3151163246890816,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.4797,
      "step": 1004
    },
    {
      "epoch": 0.315430185570246,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3706,
      "step": 1005
    },
    {
      "epoch": 0.3157440464514104,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.3173,
      "step": 1006
    },
    {
      "epoch": 0.31605790733257483,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.555,
      "step": 1007
    },
    {
      "epoch": 0.31637176821373925,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.3216,
      "step": 1008
    },
    {
      "epoch": 0.31668562909490366,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.4556,
      "step": 1009
    },
    {
      "epoch": 0.3169994899760681,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.197,
      "step": 1010
    },
    {
      "epoch": 0.31731335085723256,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.3905,
      "step": 1011
    },
    {
      "epoch": 0.317627211738397,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.4539,
      "step": 1012
    },
    {
      "epoch": 0.3179410726195614,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3952,
      "step": 1013
    },
    {
      "epoch": 0.3182549335007258,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4926,
      "step": 1014
    },
    {
      "epoch": 0.31856879438189023,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.5527,
      "step": 1015
    },
    {
      "epoch": 0.31888265526305465,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.5856,
      "step": 1016
    },
    {
      "epoch": 0.31919651614421907,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.5755,
      "step": 1017
    },
    {
      "epoch": 0.3195103770253835,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2928,
      "step": 1018
    },
    {
      "epoch": 0.3198242379065479,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3928,
      "step": 1019
    },
    {
      "epoch": 0.3201380987877123,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.5289,
      "step": 1020
    },
    {
      "epoch": 0.3204519596688768,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.5257,
      "step": 1021
    },
    {
      "epoch": 0.3207658205500412,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.662,
      "step": 1022
    },
    {
      "epoch": 0.32107968143120563,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.5929,
      "step": 1023
    },
    {
      "epoch": 0.32139354231237005,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.6586,
      "step": 1024
    },
    {
      "epoch": 0.32170740319353447,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.4007,
      "step": 1025
    },
    {
      "epoch": 0.3220212640746989,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.6509,
      "step": 1026
    },
    {
      "epoch": 0.3223351249558633,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.7426,
      "step": 1027
    },
    {
      "epoch": 0.3226489858370277,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 2.0932,
      "step": 1028
    },
    {
      "epoch": 0.32296284671819214,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.7658,
      "step": 1029
    },
    {
      "epoch": 0.32327670759935656,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.8366,
      "step": 1030
    },
    {
      "epoch": 0.32359056848052103,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.76,
      "step": 1031
    },
    {
      "epoch": 0.32390442936168545,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 2.0031,
      "step": 1032
    },
    {
      "epoch": 0.32421829024284987,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.9684,
      "step": 1033
    },
    {
      "epoch": 0.3245321511240143,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.6309,
      "step": 1034
    },
    {
      "epoch": 0.3248460120051787,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 2.1617,
      "step": 1035
    },
    {
      "epoch": 0.3251598728863431,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 2.5695,
      "step": 1036
    },
    {
      "epoch": 0.32547373376750754,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 2.3536,
      "step": 1037
    },
    {
      "epoch": 0.32578759464867196,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 2.6076,
      "step": 1038
    },
    {
      "epoch": 0.3261014555298364,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.6511,
      "step": 1039
    },
    {
      "epoch": 0.3264153164110008,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 2.3847,
      "step": 1040
    },
    {
      "epoch": 0.3267291772921653,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.8911,
      "step": 1041
    },
    {
      "epoch": 0.3270430381733297,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.8818,
      "step": 1042
    },
    {
      "epoch": 0.3273568990544941,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.3726,
      "step": 1043
    },
    {
      "epoch": 0.32767075993565853,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.7633,
      "step": 1044
    },
    {
      "epoch": 0.32798462081682295,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.7109,
      "step": 1045
    },
    {
      "epoch": 0.32829848169798737,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.6085,
      "step": 1046
    },
    {
      "epoch": 0.3286123425791518,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 2.2521,
      "step": 1047
    },
    {
      "epoch": 0.3289262034603162,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 2.1469,
      "step": 1048
    },
    {
      "epoch": 0.3292400643414806,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 2.7812,
      "step": 1049
    },
    {
      "epoch": 0.32955392522264504,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 2.6907,
      "step": 1050
    },
    {
      "epoch": 0.3298677861038095,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0002,
      "loss": 1.3973,
      "step": 1051
    },
    {
      "epoch": 0.33018164698497393,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0002,
      "loss": 1.2347,
      "step": 1052
    },
    {
      "epoch": 0.33049550786613835,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.38,
      "step": 1053
    },
    {
      "epoch": 0.33080936874730277,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3663,
      "step": 1054
    },
    {
      "epoch": 0.3311232296284672,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4329,
      "step": 1055
    },
    {
      "epoch": 0.3314370905096316,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.396,
      "step": 1056
    },
    {
      "epoch": 0.331750951390796,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.652,
      "step": 1057
    },
    {
      "epoch": 0.33206481227196044,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.644,
      "step": 1058
    },
    {
      "epoch": 0.33237867315312486,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4038,
      "step": 1059
    },
    {
      "epoch": 0.3326925340342893,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.4993,
      "step": 1060
    },
    {
      "epoch": 0.33300639491545375,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.3896,
      "step": 1061
    },
    {
      "epoch": 0.33332025579661817,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.5448,
      "step": 1062
    },
    {
      "epoch": 0.3336341166777826,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.409,
      "step": 1063
    },
    {
      "epoch": 0.333947977558947,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.3842,
      "step": 1064
    },
    {
      "epoch": 0.3342618384401114,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3284,
      "step": 1065
    },
    {
      "epoch": 0.33457569932127584,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.6964,
      "step": 1066
    },
    {
      "epoch": 0.33488956020244026,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.6085,
      "step": 1067
    },
    {
      "epoch": 0.3352034210836047,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.5205,
      "step": 1068
    },
    {
      "epoch": 0.3355172819647691,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.3477,
      "step": 1069
    },
    {
      "epoch": 0.3358311428459335,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.4518,
      "step": 1070
    },
    {
      "epoch": 0.336145003727098,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.7597,
      "step": 1071
    },
    {
      "epoch": 0.3364588646082624,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.4113,
      "step": 1072
    },
    {
      "epoch": 0.3367727254894268,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.4123,
      "step": 1073
    },
    {
      "epoch": 0.33708658637059125,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.471,
      "step": 1074
    },
    {
      "epoch": 0.33740044725175566,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.4666,
      "step": 1075
    },
    {
      "epoch": 0.3377143081329201,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.956,
      "step": 1076
    },
    {
      "epoch": 0.3380281690140845,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.6647,
      "step": 1077
    },
    {
      "epoch": 0.3383420298952489,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.8713,
      "step": 1078
    },
    {
      "epoch": 0.33865589077641334,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.8572,
      "step": 1079
    },
    {
      "epoch": 0.33896975165757776,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.9718,
      "step": 1080
    },
    {
      "epoch": 0.33928361253874223,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.8293,
      "step": 1081
    },
    {
      "epoch": 0.33959747341990665,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.9903,
      "step": 1082
    },
    {
      "epoch": 0.33991133430107107,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 2.1682,
      "step": 1083
    },
    {
      "epoch": 0.3402251951822355,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 2.1994,
      "step": 1084
    },
    {
      "epoch": 0.3405390560633999,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.4478,
      "step": 1085
    },
    {
      "epoch": 0.3408529169445643,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.9303,
      "step": 1086
    },
    {
      "epoch": 0.34116677782572874,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 2.002,
      "step": 1087
    },
    {
      "epoch": 0.34148063870689316,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 2.3037,
      "step": 1088
    },
    {
      "epoch": 0.3417944995880576,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 2.3346,
      "step": 1089
    },
    {
      "epoch": 0.342108360469222,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 2.318,
      "step": 1090
    },
    {
      "epoch": 0.34242222135038647,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.101,
      "step": 1091
    },
    {
      "epoch": 0.3427360822315509,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.2494,
      "step": 1092
    },
    {
      "epoch": 0.3430499431127153,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 2.2361,
      "step": 1093
    },
    {
      "epoch": 0.3433638039938797,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.8695,
      "step": 1094
    },
    {
      "epoch": 0.34367766487504414,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 2.2861,
      "step": 1095
    },
    {
      "epoch": 0.34399152575620856,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.7394,
      "step": 1096
    },
    {
      "epoch": 0.344305386637373,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.768,
      "step": 1097
    },
    {
      "epoch": 0.3446192475185374,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.6265,
      "step": 1098
    },
    {
      "epoch": 0.3449331083997018,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002,
      "loss": 2.6331,
      "step": 1099
    },
    {
      "epoch": 0.34524696928086623,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002,
      "loss": 3.8936,
      "step": 1100
    },
    {
      "epoch": 0.34556083016203065,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0002,
      "loss": 1.3829,
      "step": 1101
    },
    {
      "epoch": 0.3458746910431951,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 1.3983,
      "step": 1102
    },
    {
      "epoch": 0.34618855192435954,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.5221,
      "step": 1103
    },
    {
      "epoch": 0.34650241280552396,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.3643,
      "step": 1104
    },
    {
      "epoch": 0.3468162736866884,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.5967,
      "step": 1105
    },
    {
      "epoch": 0.3471301345678528,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.4927,
      "step": 1106
    },
    {
      "epoch": 0.3474439954490172,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4529,
      "step": 1107
    },
    {
      "epoch": 0.34775785633018164,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.2427,
      "step": 1108
    },
    {
      "epoch": 0.34807171721134605,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3519,
      "step": 1109
    },
    {
      "epoch": 0.3483855780925105,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.4845,
      "step": 1110
    },
    {
      "epoch": 0.3486994389736749,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.4421,
      "step": 1111
    },
    {
      "epoch": 0.34901329985483937,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2866,
      "step": 1112
    },
    {
      "epoch": 0.3493271607360038,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.5621,
      "step": 1113
    },
    {
      "epoch": 0.3496410216171682,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3539,
      "step": 1114
    },
    {
      "epoch": 0.3499548824983326,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.4143,
      "step": 1115
    },
    {
      "epoch": 0.35026874337949704,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.4976,
      "step": 1116
    },
    {
      "epoch": 0.35058260426066146,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3493,
      "step": 1117
    },
    {
      "epoch": 0.3508964651418259,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3277,
      "step": 1118
    },
    {
      "epoch": 0.3512103260229903,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4038,
      "step": 1119
    },
    {
      "epoch": 0.3515241869041547,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.5456,
      "step": 1120
    },
    {
      "epoch": 0.35183804778531913,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.4138,
      "step": 1121
    },
    {
      "epoch": 0.3521519086664836,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.4485,
      "step": 1122
    },
    {
      "epoch": 0.3521519086664836,
      "eval_loss": 1.7991268634796143,
      "eval_runtime": 123.0627,
      "eval_samples_per_second": 8.126,
      "eval_steps_per_second": 8.126,
      "step": 1122
    },
    {
      "epoch": 0.3521519086664836,
      "mmlu_eval_accuracy": 0.4303232791485502,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
      "mmlu_eval_accuracy_conceptual_physics": 0.23076923076923078,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.6166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.5652173913043478,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5930232558139535,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.2857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_psychology": 0.3188405797101449,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.3333333333333333,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.4021799655656733,
      "step": 1122
    },
    {
      "epoch": 0.352465769547648,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.6944,
      "step": 1123
    },
    {
      "epoch": 0.35277963042881244,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.4365,
      "step": 1124
    },
    {
      "epoch": 0.35309349130997686,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.7668,
      "step": 1125
    },
    {
      "epoch": 0.3534073521911413,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.7494,
      "step": 1126
    },
    {
      "epoch": 0.3537212130723057,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.7577,
      "step": 1127
    },
    {
      "epoch": 0.3540350739534701,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.5836,
      "step": 1128
    },
    {
      "epoch": 0.35434893483463453,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.7775,
      "step": 1129
    },
    {
      "epoch": 0.35466279571579895,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.7426,
      "step": 1130
    },
    {
      "epoch": 0.35497665659696337,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.7906,
      "step": 1131
    },
    {
      "epoch": 0.35529051747812784,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.7566,
      "step": 1132
    },
    {
      "epoch": 0.35560437835929226,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 2.2073,
      "step": 1133
    },
    {
      "epoch": 0.3559182392404567,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.885,
      "step": 1134
    },
    {
      "epoch": 0.3562321001216211,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 2.2592,
      "step": 1135
    },
    {
      "epoch": 0.3565459610027855,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 2.0171,
      "step": 1136
    },
    {
      "epoch": 0.35685982188394993,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 2.4336,
      "step": 1137
    },
    {
      "epoch": 0.35717368276511435,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 2.1057,
      "step": 1138
    },
    {
      "epoch": 0.35748754364627877,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.6518,
      "step": 1139
    },
    {
      "epoch": 0.3578014045274432,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.7937,
      "step": 1140
    },
    {
      "epoch": 0.3581152654086076,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 2.2845,
      "step": 1141
    },
    {
      "epoch": 0.3584291262897721,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.41,
      "step": 1142
    },
    {
      "epoch": 0.3587429871709365,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.1395,
      "step": 1143
    },
    {
      "epoch": 0.3590568480521009,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 2.2529,
      "step": 1144
    },
    {
      "epoch": 0.35937070893326534,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.7725,
      "step": 1145
    },
    {
      "epoch": 0.35968456981442976,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 2.3379,
      "step": 1146
    },
    {
      "epoch": 0.3599984306955942,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.8914,
      "step": 1147
    },
    {
      "epoch": 0.3603122915767586,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 2.4866,
      "step": 1148
    },
    {
      "epoch": 0.360626152457923,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.8759,
      "step": 1149
    },
    {
      "epoch": 0.36094001333908743,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 2.9355,
      "step": 1150
    },
    {
      "epoch": 0.36125387422025185,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.5055,
      "step": 1151
    },
    {
      "epoch": 0.3615677351014163,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0002,
      "loss": 1.382,
      "step": 1152
    },
    {
      "epoch": 0.36188159598258074,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.3176,
      "step": 1153
    },
    {
      "epoch": 0.36219545686374516,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.4593,
      "step": 1154
    },
    {
      "epoch": 0.3625093177449096,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.385,
      "step": 1155
    },
    {
      "epoch": 0.362823178626074,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.3072,
      "step": 1156
    },
    {
      "epoch": 0.3631370395072384,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2866,
      "step": 1157
    },
    {
      "epoch": 0.36345090038840283,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.3623,
      "step": 1158
    },
    {
      "epoch": 0.36376476126956725,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3802,
      "step": 1159
    },
    {
      "epoch": 0.36407862215073167,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.4173,
      "step": 1160
    },
    {
      "epoch": 0.3643924830318961,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.4814,
      "step": 1161
    },
    {
      "epoch": 0.36470634391306056,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.4864,
      "step": 1162
    },
    {
      "epoch": 0.365020204794225,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.4863,
      "step": 1163
    },
    {
      "epoch": 0.3653340656753894,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.4433,
      "step": 1164
    },
    {
      "epoch": 0.3656479265565538,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4984,
      "step": 1165
    },
    {
      "epoch": 0.36596178743771823,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4259,
      "step": 1166
    },
    {
      "epoch": 0.36627564831888265,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.566,
      "step": 1167
    },
    {
      "epoch": 0.36658950920004707,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.415,
      "step": 1168
    },
    {
      "epoch": 0.3669033700812115,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.7164,
      "step": 1169
    },
    {
      "epoch": 0.3672172309623759,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.5214,
      "step": 1170
    },
    {
      "epoch": 0.3675310918435403,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.6026,
      "step": 1171
    },
    {
      "epoch": 0.3678449527247048,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.5882,
      "step": 1172
    },
    {
      "epoch": 0.3681588136058692,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.6785,
      "step": 1173
    },
    {
      "epoch": 0.36847267448703364,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.4065,
      "step": 1174
    },
    {
      "epoch": 0.36878653536819805,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.7289,
      "step": 1175
    },
    {
      "epoch": 0.3691003962493625,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 2.0728,
      "step": 1176
    },
    {
      "epoch": 0.3694142571305269,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.5221,
      "step": 1177
    },
    {
      "epoch": 0.3697281180116913,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 2.0874,
      "step": 1178
    },
    {
      "epoch": 0.3700419788928557,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.9565,
      "step": 1179
    },
    {
      "epoch": 0.37035583977402015,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.9323,
      "step": 1180
    },
    {
      "epoch": 0.37066970065518456,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.9522,
      "step": 1181
    },
    {
      "epoch": 0.37098356153634904,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 2.0937,
      "step": 1182
    },
    {
      "epoch": 0.37129742241751346,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.8839,
      "step": 1183
    },
    {
      "epoch": 0.3716112832986779,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.8556,
      "step": 1184
    },
    {
      "epoch": 0.3719251441798423,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 2.0557,
      "step": 1185
    },
    {
      "epoch": 0.3722390050610067,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.3553,
      "step": 1186
    },
    {
      "epoch": 0.37255286594217113,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 2.1199,
      "step": 1187
    },
    {
      "epoch": 0.37286672682333555,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 2.1553,
      "step": 1188
    },
    {
      "epoch": 0.37318058770449997,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.7584,
      "step": 1189
    },
    {
      "epoch": 0.3734944485856644,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.5655,
      "step": 1190
    },
    {
      "epoch": 0.3738083094668288,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 2.7206,
      "step": 1191
    },
    {
      "epoch": 0.3741221703479933,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.2983,
      "step": 1192
    },
    {
      "epoch": 0.3744360312291577,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 2.5373,
      "step": 1193
    },
    {
      "epoch": 0.3747498921103221,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 2.1976,
      "step": 1194
    },
    {
      "epoch": 0.37506375299148653,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.7856,
      "step": 1195
    },
    {
      "epoch": 0.37537761387265095,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 2.1118,
      "step": 1196
    },
    {
      "epoch": 0.37569147475381537,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.2917,
      "step": 1197
    },
    {
      "epoch": 0.3760053356349798,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.2105,
      "step": 1198
    },
    {
      "epoch": 0.3763191965161442,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.2213,
      "step": 1199
    },
    {
      "epoch": 0.3766330573973086,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.4509,
      "step": 1200
    },
    {
      "epoch": 0.37694691827847304,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 1.4907,
      "step": 1201
    },
    {
      "epoch": 0.3772607791596375,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4751,
      "step": 1202
    },
    {
      "epoch": 0.37757464004080193,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.41,
      "step": 1203
    },
    {
      "epoch": 0.37788850092196635,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4708,
      "step": 1204
    },
    {
      "epoch": 0.37820236180313077,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.4557,
      "step": 1205
    },
    {
      "epoch": 0.3785162226842952,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.4991,
      "step": 1206
    },
    {
      "epoch": 0.3788300835654596,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4234,
      "step": 1207
    },
    {
      "epoch": 0.379143944446624,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3855,
      "step": 1208
    },
    {
      "epoch": 0.37945780532778844,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4934,
      "step": 1209
    },
    {
      "epoch": 0.37977166620895286,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4267,
      "step": 1210
    },
    {
      "epoch": 0.3800855270901173,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.443,
      "step": 1211
    },
    {
      "epoch": 0.38039938797128175,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.352,
      "step": 1212
    },
    {
      "epoch": 0.3807132488524462,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3038,
      "step": 1213
    },
    {
      "epoch": 0.3810271097336106,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2391,
      "step": 1214
    },
    {
      "epoch": 0.381340970614775,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4306,
      "step": 1215
    },
    {
      "epoch": 0.38165483149593943,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.588,
      "step": 1216
    },
    {
      "epoch": 0.38196869237710385,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3168,
      "step": 1217
    },
    {
      "epoch": 0.38228255325826827,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.5673,
      "step": 1218
    },
    {
      "epoch": 0.3825964141394327,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.44,
      "step": 1219
    },
    {
      "epoch": 0.3829102750205971,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.3603,
      "step": 1220
    },
    {
      "epoch": 0.3832241359017615,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.5589,
      "step": 1221
    },
    {
      "epoch": 0.383537996782926,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.4931,
      "step": 1222
    },
    {
      "epoch": 0.3838518576640904,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.42,
      "step": 1223
    },
    {
      "epoch": 0.38416571854525483,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.6192,
      "step": 1224
    },
    {
      "epoch": 0.38447957942641925,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.7035,
      "step": 1225
    },
    {
      "epoch": 0.38479344030758367,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.5969,
      "step": 1226
    },
    {
      "epoch": 0.3851073011887481,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.8406,
      "step": 1227
    },
    {
      "epoch": 0.3854211620699125,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 2.1648,
      "step": 1228
    },
    {
      "epoch": 0.3857350229510769,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.9683,
      "step": 1229
    },
    {
      "epoch": 0.38604888383224134,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.817,
      "step": 1230
    },
    {
      "epoch": 0.38636274471340576,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.045,
      "step": 1231
    },
    {
      "epoch": 0.38667660559457023,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 2.0243,
      "step": 1232
    },
    {
      "epoch": 0.38699046647573465,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.7933,
      "step": 1233
    },
    {
      "epoch": 0.38730432735689907,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.1071,
      "step": 1234
    },
    {
      "epoch": 0.3876181882380635,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 2.33,
      "step": 1235
    },
    {
      "epoch": 0.3879320491192279,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.4712,
      "step": 1236
    },
    {
      "epoch": 0.3882459100003923,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.8365,
      "step": 1237
    },
    {
      "epoch": 0.38855977088155674,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 2.4525,
      "step": 1238
    },
    {
      "epoch": 0.38887363176272116,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.8756,
      "step": 1239
    },
    {
      "epoch": 0.3891874926438856,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.762,
      "step": 1240
    },
    {
      "epoch": 0.38950135352505,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 2.0963,
      "step": 1241
    },
    {
      "epoch": 0.38981521440621447,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.136,
      "step": 1242
    },
    {
      "epoch": 0.3901290752873789,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 2.3817,
      "step": 1243
    },
    {
      "epoch": 0.3904429361685433,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 2.2338,
      "step": 1244
    },
    {
      "epoch": 0.3907567970497077,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.9984,
      "step": 1245
    },
    {
      "epoch": 0.39107065793087215,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 2.2326,
      "step": 1246
    },
    {
      "epoch": 0.39138451881203656,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 2.0614,
      "step": 1247
    },
    {
      "epoch": 0.391698379693201,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.408,
      "step": 1248
    },
    {
      "epoch": 0.3920122405743654,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 2.4836,
      "step": 1249
    },
    {
      "epoch": 0.3923261014555298,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 2.8903,
      "step": 1250
    },
    {
      "epoch": 0.39263996233669424,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0002,
      "loss": 1.3626,
      "step": 1251
    },
    {
      "epoch": 0.3929538232178587,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.4351,
      "step": 1252
    },
    {
      "epoch": 0.39326768409902313,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.4794,
      "step": 1253
    },
    {
      "epoch": 0.39358154498018755,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.6374,
      "step": 1254
    },
    {
      "epoch": 0.39389540586135197,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4264,
      "step": 1255
    },
    {
      "epoch": 0.3942092667425164,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3838,
      "step": 1256
    },
    {
      "epoch": 0.3945231276236808,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.4814,
      "step": 1257
    },
    {
      "epoch": 0.3948369885048452,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.4941,
      "step": 1258
    },
    {
      "epoch": 0.39515084938600964,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3982,
      "step": 1259
    },
    {
      "epoch": 0.39546471026717406,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.5024,
      "step": 1260
    },
    {
      "epoch": 0.3957785711483385,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.5613,
      "step": 1261
    },
    {
      "epoch": 0.39609243202950295,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.726,
      "step": 1262
    },
    {
      "epoch": 0.39640629291066737,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.5011,
      "step": 1263
    },
    {
      "epoch": 0.3967201537918318,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.4721,
      "step": 1264
    },
    {
      "epoch": 0.3970340146729962,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.6612,
      "step": 1265
    },
    {
      "epoch": 0.3973478755541606,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.4582,
      "step": 1266
    },
    {
      "epoch": 0.39766173643532504,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.4811,
      "step": 1267
    },
    {
      "epoch": 0.39797559731648946,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.4813,
      "step": 1268
    },
    {
      "epoch": 0.3982894581976539,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.567,
      "step": 1269
    },
    {
      "epoch": 0.3986033190788183,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.5624,
      "step": 1270
    },
    {
      "epoch": 0.3989171799599827,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.3426,
      "step": 1271
    },
    {
      "epoch": 0.3992310408411472,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.701,
      "step": 1272
    },
    {
      "epoch": 0.3995449017223116,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.5935,
      "step": 1273
    },
    {
      "epoch": 0.399858762603476,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.6688,
      "step": 1274
    },
    {
      "epoch": 0.40017262348464044,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.6152,
      "step": 1275
    },
    {
      "epoch": 0.40048648436580486,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.9036,
      "step": 1276
    },
    {
      "epoch": 0.4008003452469693,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.8394,
      "step": 1277
    },
    {
      "epoch": 0.4011142061281337,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.8218,
      "step": 1278
    },
    {
      "epoch": 0.4014280670092981,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.7053,
      "step": 1279
    },
    {
      "epoch": 0.40174192789046254,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.6976,
      "step": 1280
    },
    {
      "epoch": 0.40205578877162695,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.8774,
      "step": 1281
    },
    {
      "epoch": 0.40236964965279143,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 2.3061,
      "step": 1282
    },
    {
      "epoch": 0.40268351053395585,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 2.1891,
      "step": 1283
    },
    {
      "epoch": 0.40299737141512026,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 2.3157,
      "step": 1284
    },
    {
      "epoch": 0.4033112322962847,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 2.4991,
      "step": 1285
    },
    {
      "epoch": 0.4036250931774491,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.5725,
      "step": 1286
    },
    {
      "epoch": 0.4039389540586135,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.971,
      "step": 1287
    },
    {
      "epoch": 0.40425281493977794,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.7505,
      "step": 1288
    },
    {
      "epoch": 0.40456667582094236,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.991,
      "step": 1289
    },
    {
      "epoch": 0.4048805367021068,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.1957,
      "step": 1290
    },
    {
      "epoch": 0.4051943975832712,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.4308,
      "step": 1291
    },
    {
      "epoch": 0.4055082584644356,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.1227,
      "step": 1292
    },
    {
      "epoch": 0.4058221193456001,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 2.5946,
      "step": 1293
    },
    {
      "epoch": 0.4061359802267645,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 2.3162,
      "step": 1294
    },
    {
      "epoch": 0.4064498411079289,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 2.3666,
      "step": 1295
    },
    {
      "epoch": 0.40676370198909334,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.0322,
      "step": 1296
    },
    {
      "epoch": 0.40707756287025776,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 2.0175,
      "step": 1297
    },
    {
      "epoch": 0.4073914237514222,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 2.339,
      "step": 1298
    },
    {
      "epoch": 0.4077052846325866,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 2.1715,
      "step": 1299
    },
    {
      "epoch": 0.408019145513751,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002,
      "loss": 2.9335,
      "step": 1300
    },
    {
      "epoch": 0.40833300639491543,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0002,
      "loss": 1.3224,
      "step": 1301
    },
    {
      "epoch": 0.40864686727607985,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0002,
      "loss": 1.3435,
      "step": 1302
    },
    {
      "epoch": 0.4089607281572443,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.4457,
      "step": 1303
    },
    {
      "epoch": 0.40927458903840874,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.4359,
      "step": 1304
    },
    {
      "epoch": 0.40958844991957316,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.5369,
      "step": 1305
    },
    {
      "epoch": 0.4099023108007376,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.3717,
      "step": 1306
    },
    {
      "epoch": 0.410216171681902,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.6337,
      "step": 1307
    },
    {
      "epoch": 0.4105300325630664,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.5547,
      "step": 1308
    },
    {
      "epoch": 0.41084389344423083,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.5146,
      "step": 1309
    },
    {
      "epoch": 0.41084389344423083,
      "eval_loss": 1.7959522008895874,
      "eval_runtime": 142.4063,
      "eval_samples_per_second": 7.022,
      "eval_steps_per_second": 7.022,
      "step": 1309
    },
    {
      "epoch": 0.41084389344423083,
      "mmlu_eval_accuracy": 0.42657891675533444,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
      "mmlu_eval_accuracy_conceptual_physics": 0.23076923076923078,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
      "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": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.42857142857142855,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.6166666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.5,
      "mmlu_eval_accuracy_human_aging": 0.5652173913043478,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
      "mmlu_eval_accuracy_management": 0.2727272727272727,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.5697674418604651,
      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.5151515151515151,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.3188405797101449,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.200824461015979,
      "step": 1309
    },
    {
      "epoch": 0.41115775432539525,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.3974,
      "step": 1310
    },
    {
      "epoch": 0.41147161520655967,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.3121,
      "step": 1311
    },
    {
      "epoch": 0.4117854760877241,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.3059,
      "step": 1312
    },
    {
      "epoch": 0.41209933696888856,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.4512,
      "step": 1313
    },
    {
      "epoch": 0.412413197850053,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4743,
      "step": 1314
    },
    {
      "epoch": 0.4127270587312174,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.4525,
      "step": 1315
    },
    {
      "epoch": 0.4130409196123818,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.5023,
      "step": 1316
    },
    {
      "epoch": 0.41335478049354624,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3757,
      "step": 1317
    },
    {
      "epoch": 0.41366864137471065,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.5656,
      "step": 1318
    },
    {
      "epoch": 0.4139825022558751,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.6074,
      "step": 1319
    },
    {
      "epoch": 0.4142963631370395,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.5414,
      "step": 1320
    },
    {
      "epoch": 0.4146102240182039,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.5299,
      "step": 1321
    },
    {
      "epoch": 0.41492408489936833,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.6561,
      "step": 1322
    },
    {
      "epoch": 0.4152379457805328,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2308,
      "step": 1323
    },
    {
      "epoch": 0.4155518066616972,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.6029,
      "step": 1324
    },
    {
      "epoch": 0.41586566754286164,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.4665,
      "step": 1325
    },
    {
      "epoch": 0.41617952842402606,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.6009,
      "step": 1326
    },
    {
      "epoch": 0.4164933893051905,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.7153,
      "step": 1327
    },
    {
      "epoch": 0.4168072501863549,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.8695,
      "step": 1328
    },
    {
      "epoch": 0.4171211110675193,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.9682,
      "step": 1329
    },
    {
      "epoch": 0.41743497194868373,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.8397,
      "step": 1330
    },
    {
      "epoch": 0.41774883282984815,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.594,
      "step": 1331
    },
    {
      "epoch": 0.41806269371101257,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.9119,
      "step": 1332
    },
    {
      "epoch": 0.41837655459217704,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 2.1301,
      "step": 1333
    },
    {
      "epoch": 0.41869041547334146,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.4656,
      "step": 1334
    },
    {
      "epoch": 0.4190042763545059,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.9128,
      "step": 1335
    },
    {
      "epoch": 0.4193181372356703,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.8979,
      "step": 1336
    },
    {
      "epoch": 0.4196319981168347,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 2.1664,
      "step": 1337
    },
    {
      "epoch": 0.41994585899799913,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.746,
      "step": 1338
    },
    {
      "epoch": 0.42025971987916355,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 2.2825,
      "step": 1339
    },
    {
      "epoch": 0.42057358076032797,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.7572,
      "step": 1340
    },
    {
      "epoch": 0.4208874416414924,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 2.2997,
      "step": 1341
    },
    {
      "epoch": 0.4212013025226568,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 2.6267,
      "step": 1342
    },
    {
      "epoch": 0.4215151634038213,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.9881,
      "step": 1343
    },
    {
      "epoch": 0.4218290242849857,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 2.358,
      "step": 1344
    },
    {
      "epoch": 0.4221428851661501,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.8243,
      "step": 1345
    },
    {
      "epoch": 0.42245674604731454,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 2.2387,
      "step": 1346
    },
    {
      "epoch": 0.42277060692847895,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 2.2483,
      "step": 1347
    },
    {
      "epoch": 0.42308446780964337,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 2.4793,
      "step": 1348
    },
    {
      "epoch": 0.4233983286908078,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 2.668,
      "step": 1349
    },
    {
      "epoch": 0.4237121895719722,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 2.9841,
      "step": 1350
    },
    {
      "epoch": 0.4240260504531366,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0002,
      "loss": 1.4247,
      "step": 1351
    },
    {
      "epoch": 0.42433991133430105,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0002,
      "loss": 1.2979,
      "step": 1352
    },
    {
      "epoch": 0.4246537722154655,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.4912,
      "step": 1353
    },
    {
      "epoch": 0.42496763309662994,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.382,
      "step": 1354
    },
    {
      "epoch": 0.42528149397779436,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.345,
      "step": 1355
    },
    {
      "epoch": 0.4255953548589588,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.5246,
      "step": 1356
    },
    {
      "epoch": 0.4259092157401232,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3958,
      "step": 1357
    },
    {
      "epoch": 0.4262230766212876,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.5113,
      "step": 1358
    },
    {
      "epoch": 0.42653693750245203,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.3832,
      "step": 1359
    },
    {
      "epoch": 0.42685079838361645,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.3995,
      "step": 1360
    },
    {
      "epoch": 0.42716465926478087,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3415,
      "step": 1361
    },
    {
      "epoch": 0.4274785201459453,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.6801,
      "step": 1362
    },
    {
      "epoch": 0.42779238102710976,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3269,
      "step": 1363
    },
    {
      "epoch": 0.4281062419082742,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4729,
      "step": 1364
    },
    {
      "epoch": 0.4284201027894386,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.553,
      "step": 1365
    },
    {
      "epoch": 0.428733963670603,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.4457,
      "step": 1366
    },
    {
      "epoch": 0.42904782455176743,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.4326,
      "step": 1367
    },
    {
      "epoch": 0.42936168543293185,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.6397,
      "step": 1368
    },
    {
      "epoch": 0.42967554631409627,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2723,
      "step": 1369
    },
    {
      "epoch": 0.4299894071952607,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.4549,
      "step": 1370
    },
    {
      "epoch": 0.4303032680764251,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3559,
      "step": 1371
    },
    {
      "epoch": 0.4306171289575895,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.5031,
      "step": 1372
    },
    {
      "epoch": 0.430930989838754,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.5448,
      "step": 1373
    },
    {
      "epoch": 0.4312448507199184,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.4973,
      "step": 1374
    },
    {
      "epoch": 0.43155871160108283,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.7627,
      "step": 1375
    },
    {
      "epoch": 0.43187257248224725,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.8157,
      "step": 1376
    },
    {
      "epoch": 0.43218643336341167,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.828,
      "step": 1377
    },
    {
      "epoch": 0.4325002942445761,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.8489,
      "step": 1378
    },
    {
      "epoch": 0.4328141551257405,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.9214,
      "step": 1379
    },
    {
      "epoch": 0.4331280160069049,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.9501,
      "step": 1380
    },
    {
      "epoch": 0.43344187688806934,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.7582,
      "step": 1381
    },
    {
      "epoch": 0.43375573776923376,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.9698,
      "step": 1382
    },
    {
      "epoch": 0.43406959865039824,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 2.0568,
      "step": 1383
    },
    {
      "epoch": 0.43438345953156265,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.9915,
      "step": 1384
    },
    {
      "epoch": 0.4346973204127271,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.3223,
      "step": 1385
    },
    {
      "epoch": 0.4350111812938915,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.9109,
      "step": 1386
    },
    {
      "epoch": 0.4353250421750559,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 2.236,
      "step": 1387
    },
    {
      "epoch": 0.43563890305622033,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 2.3323,
      "step": 1388
    },
    {
      "epoch": 0.43595276393738475,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.3034,
      "step": 1389
    },
    {
      "epoch": 0.43626662481854916,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 2.5363,
      "step": 1390
    },
    {
      "epoch": 0.4365804856997136,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 2.1009,
      "step": 1391
    },
    {
      "epoch": 0.436894346580878,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 2.165,
      "step": 1392
    },
    {
      "epoch": 0.4372082074620425,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.0697,
      "step": 1393
    },
    {
      "epoch": 0.4375220683432069,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.7393,
      "step": 1394
    },
    {
      "epoch": 0.4378359292243713,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 2.1264,
      "step": 1395
    },
    {
      "epoch": 0.43814979010553573,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 2.3717,
      "step": 1396
    },
    {
      "epoch": 0.43846365098670015,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.3081,
      "step": 1397
    },
    {
      "epoch": 0.43877751186786457,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.9563,
      "step": 1398
    },
    {
      "epoch": 0.439091372749029,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.0758,
      "step": 1399
    },
    {
      "epoch": 0.4394052336301934,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 3.0342,
      "step": 1400
    },
    {
      "epoch": 0.4397190945113578,
      "grad_norm": 0.07421875,
      "learning_rate": 0.0002,
      "loss": 1.3142,
      "step": 1401
    },
    {
      "epoch": 0.44003295539252224,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0002,
      "loss": 1.1996,
      "step": 1402
    },
    {
      "epoch": 0.4403468162736867,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 1.39,
      "step": 1403
    },
    {
      "epoch": 0.44066067715485113,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.3667,
      "step": 1404
    },
    {
      "epoch": 0.44097453803601555,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.4288,
      "step": 1405
    },
    {
      "epoch": 0.44128839891717997,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.4433,
      "step": 1406
    },
    {
      "epoch": 0.4416022597983444,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.5323,
      "step": 1407
    },
    {
      "epoch": 0.4419161206795088,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.3991,
      "step": 1408
    },
    {
      "epoch": 0.4422299815606732,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.6745,
      "step": 1409
    },
    {
      "epoch": 0.44254384244183764,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.2545,
      "step": 1410
    },
    {
      "epoch": 0.44285770332300206,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.4249,
      "step": 1411
    },
    {
      "epoch": 0.4431715642041665,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.4019,
      "step": 1412
    },
    {
      "epoch": 0.44348542508533095,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.5158,
      "step": 1413
    },
    {
      "epoch": 0.44379928596649537,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.5706,
      "step": 1414
    },
    {
      "epoch": 0.4441131468476598,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3462,
      "step": 1415
    },
    {
      "epoch": 0.4444270077288242,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.354,
      "step": 1416
    },
    {
      "epoch": 0.4447408686099886,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.71,
      "step": 1417
    },
    {
      "epoch": 0.44505472949115304,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3592,
      "step": 1418
    },
    {
      "epoch": 0.44536859037231746,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.5735,
      "step": 1419
    },
    {
      "epoch": 0.4456824512534819,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2627,
      "step": 1420
    },
    {
      "epoch": 0.4459963121346463,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.4576,
      "step": 1421
    },
    {
      "epoch": 0.4463101730158107,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.5926,
      "step": 1422
    },
    {
      "epoch": 0.4466240338969752,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.5425,
      "step": 1423
    },
    {
      "epoch": 0.4469378947781396,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.444,
      "step": 1424
    },
    {
      "epoch": 0.44725175565930403,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.5796,
      "step": 1425
    },
    {
      "epoch": 0.44756561654046845,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.8711,
      "step": 1426
    },
    {
      "epoch": 0.44787947742163287,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.9377,
      "step": 1427
    },
    {
      "epoch": 0.4481933383027973,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 2.199,
      "step": 1428
    },
    {
      "epoch": 0.4485071991839617,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.8875,
      "step": 1429
    },
    {
      "epoch": 0.4488210600651261,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 2.3309,
      "step": 1430
    },
    {
      "epoch": 0.44913492094629054,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.6421,
      "step": 1431
    },
    {
      "epoch": 0.44944878182745496,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 2.215,
      "step": 1432
    },
    {
      "epoch": 0.44976264270861943,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 2.2454,
      "step": 1433
    },
    {
      "epoch": 0.45007650358978385,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.5037,
      "step": 1434
    },
    {
      "epoch": 0.45039036447094827,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.1748,
      "step": 1435
    },
    {
      "epoch": 0.4507042253521127,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.5059,
      "step": 1436
    },
    {
      "epoch": 0.4510180862332771,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 2.2051,
      "step": 1437
    },
    {
      "epoch": 0.4513319471144415,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 2.0578,
      "step": 1438
    },
    {
      "epoch": 0.45164580799560594,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 2.4286,
      "step": 1439
    },
    {
      "epoch": 0.45195966887677036,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.1384,
      "step": 1440
    },
    {
      "epoch": 0.4522735297579348,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 2.4053,
      "step": 1441
    },
    {
      "epoch": 0.4525873906390992,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.3378,
      "step": 1442
    },
    {
      "epoch": 0.45290125152026367,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.9454,
      "step": 1443
    },
    {
      "epoch": 0.4532151124014281,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 2.3694,
      "step": 1444
    },
    {
      "epoch": 0.4535289732825925,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.3593,
      "step": 1445
    },
    {
      "epoch": 0.4538428341637569,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 2.019,
      "step": 1446
    },
    {
      "epoch": 0.45415669504492134,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 2.2726,
      "step": 1447
    },
    {
      "epoch": 0.45447055592608576,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.8556,
      "step": 1448
    },
    {
      "epoch": 0.4547844168072502,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.1626,
      "step": 1449
    },
    {
      "epoch": 0.4550982776884146,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 3.0538,
      "step": 1450
    },
    {
      "epoch": 0.455412138569579,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0002,
      "loss": 1.3931,
      "step": 1451
    },
    {
      "epoch": 0.45572599945074344,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.3111,
      "step": 1452
    },
    {
      "epoch": 0.4560398603319079,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.4333,
      "step": 1453
    },
    {
      "epoch": 0.4563537212130723,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4416,
      "step": 1454
    },
    {
      "epoch": 0.45666758209423675,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.5229,
      "step": 1455
    },
    {
      "epoch": 0.45698144297540116,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.5088,
      "step": 1456
    },
    {
      "epoch": 0.4572953038565656,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.4849,
      "step": 1457
    },
    {
      "epoch": 0.45760916473773,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.5722,
      "step": 1458
    },
    {
      "epoch": 0.4579230256188944,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3394,
      "step": 1459
    },
    {
      "epoch": 0.45823688650005884,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.4405,
      "step": 1460
    },
    {
      "epoch": 0.45855074738122326,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.4214,
      "step": 1461
    },
    {
      "epoch": 0.4588646082623877,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3724,
      "step": 1462
    },
    {
      "epoch": 0.45917846914355215,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.4223,
      "step": 1463
    },
    {
      "epoch": 0.45949233002471657,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4021,
      "step": 1464
    },
    {
      "epoch": 0.459806190905881,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.5362,
      "step": 1465
    },
    {
      "epoch": 0.4601200517870454,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.4489,
      "step": 1466
    },
    {
      "epoch": 0.4604339126682098,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4884,
      "step": 1467
    },
    {
      "epoch": 0.46074777354937424,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.5037,
      "step": 1468
    },
    {
      "epoch": 0.46106163443053866,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.7523,
      "step": 1469
    },
    {
      "epoch": 0.4613754953117031,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.5479,
      "step": 1470
    },
    {
      "epoch": 0.4616893561928675,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.5897,
      "step": 1471
    },
    {
      "epoch": 0.4620032170740319,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.7879,
      "step": 1472
    },
    {
      "epoch": 0.46231707795519633,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.5347,
      "step": 1473
    },
    {
      "epoch": 0.4626309388363608,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.5437,
      "step": 1474
    },
    {
      "epoch": 0.4629447997175252,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.6741,
      "step": 1475
    },
    {
      "epoch": 0.46325866059868964,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.654,
      "step": 1476
    },
    {
      "epoch": 0.46357252147985406,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.951,
      "step": 1477
    },
    {
      "epoch": 0.4638863823610185,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.9001,
      "step": 1478
    },
    {
      "epoch": 0.4642002432421829,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.8554,
      "step": 1479
    },
    {
      "epoch": 0.4645141041233473,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.8504,
      "step": 1480
    },
    {
      "epoch": 0.46482796500451173,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.9588,
      "step": 1481
    },
    {
      "epoch": 0.46514182588567615,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 2.1388,
      "step": 1482
    },
    {
      "epoch": 0.46545568676684057,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.8643,
      "step": 1483
    },
    {
      "epoch": 0.46576954764800504,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 2.3198,
      "step": 1484
    },
    {
      "epoch": 0.46608340852916946,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 2.5174,
      "step": 1485
    },
    {
      "epoch": 0.4663972694103339,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 2.1416,
      "step": 1486
    },
    {
      "epoch": 0.4667111302914983,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 2.2324,
      "step": 1487
    },
    {
      "epoch": 0.4670249911726627,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.2841,
      "step": 1488
    },
    {
      "epoch": 0.46733885205382714,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 2.5772,
      "step": 1489
    },
    {
      "epoch": 0.46765271293499155,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 2.4692,
      "step": 1490
    },
    {
      "epoch": 0.467966573816156,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 2.1528,
      "step": 1491
    },
    {
      "epoch": 0.4682804346973204,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 2.0464,
      "step": 1492
    },
    {
      "epoch": 0.4685942955784848,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.8042,
      "step": 1493
    },
    {
      "epoch": 0.4689081564596493,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 2.0473,
      "step": 1494
    },
    {
      "epoch": 0.4692220173408137,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.8333,
      "step": 1495
    },
    {
      "epoch": 0.4695358782219781,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 2.6745,
      "step": 1496
    },
    {
      "epoch": 0.4695358782219781,
      "eval_loss": 1.7854562997817993,
      "eval_runtime": 123.6184,
      "eval_samples_per_second": 8.089,
      "eval_steps_per_second": 8.089,
      "step": 1496
    },
    {
      "epoch": 0.4695358782219781,
      "mmlu_eval_accuracy": 0.4358320976556104,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3448275862068966,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.2692307692307692,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.5625,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "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": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.42857142857142855,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.6,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.5652173913043478,
      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "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": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.6046511627906976,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_psychology": 0.3188405797101449,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.1939274586352704,
      "step": 1496
    },
    {
      "epoch": 0.46984973910314254,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 2.1995,
      "step": 1497
    },
    {
      "epoch": 0.47016359998430696,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.9518,
      "step": 1498
    },
    {
      "epoch": 0.4704774608654714,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 2.5733,
      "step": 1499
    },
    {
      "epoch": 0.4707913217466358,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 3.3822,
      "step": 1500
    },
    {
      "epoch": 0.4711051826278002,
      "grad_norm": 0.08447265625,
      "learning_rate": 0.0002,
      "loss": 1.3619,
      "step": 1501
    },
    {
      "epoch": 0.47141904350896463,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0002,
      "loss": 1.36,
      "step": 1502
    },
    {
      "epoch": 0.47173290439012905,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.3708,
      "step": 1503
    },
    {
      "epoch": 0.4720467652712935,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.6959,
      "step": 1504
    },
    {
      "epoch": 0.47236062615245794,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.6331,
      "step": 1505
    },
    {
      "epoch": 0.47267448703362236,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3806,
      "step": 1506
    },
    {
      "epoch": 0.4729883479147868,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.5572,
      "step": 1507
    },
    {
      "epoch": 0.4733022087959512,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.7204,
      "step": 1508
    },
    {
      "epoch": 0.4736160696771156,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.2484,
      "step": 1509
    },
    {
      "epoch": 0.47392993055828003,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.4883,
      "step": 1510
    },
    {
      "epoch": 0.47424379143944445,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.4229,
      "step": 1511
    },
    {
      "epoch": 0.47455765232060887,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.5399,
      "step": 1512
    },
    {
      "epoch": 0.4748715132017733,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4752,
      "step": 1513
    },
    {
      "epoch": 0.47518537408293776,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3572,
      "step": 1514
    },
    {
      "epoch": 0.4754992349641022,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3585,
      "step": 1515
    },
    {
      "epoch": 0.4758130958452666,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3959,
      "step": 1516
    },
    {
      "epoch": 0.476126956726431,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.4074,
      "step": 1517
    },
    {
      "epoch": 0.47644081760759543,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.5638,
      "step": 1518
    },
    {
      "epoch": 0.47675467848875985,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3467,
      "step": 1519
    },
    {
      "epoch": 0.47706853936992427,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.4784,
      "step": 1520
    },
    {
      "epoch": 0.4773824002510887,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.716,
      "step": 1521
    },
    {
      "epoch": 0.4776962611322531,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3299,
      "step": 1522
    },
    {
      "epoch": 0.4780101220134175,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.4561,
      "step": 1523
    },
    {
      "epoch": 0.478323982894582,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.4871,
      "step": 1524
    },
    {
      "epoch": 0.4786378437757464,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.7243,
      "step": 1525
    },
    {
      "epoch": 0.47895170465691084,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.6425,
      "step": 1526
    },
    {
      "epoch": 0.47926556553807526,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.5112,
      "step": 1527
    },
    {
      "epoch": 0.4795794264192397,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.6385,
      "step": 1528
    },
    {
      "epoch": 0.4798932873004041,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.7025,
      "step": 1529
    },
    {
      "epoch": 0.4802071481815685,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 2.2003,
      "step": 1530
    },
    {
      "epoch": 0.48052100906273293,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 2.3669,
      "step": 1531
    },
    {
      "epoch": 0.48083486994389735,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 2.0398,
      "step": 1532
    },
    {
      "epoch": 0.48114873082506177,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.8588,
      "step": 1533
    },
    {
      "epoch": 0.48146259170622624,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 2.4884,
      "step": 1534
    },
    {
      "epoch": 0.48177645258739066,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 2.7152,
      "step": 1535
    },
    {
      "epoch": 0.4820903134685551,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.975,
      "step": 1536
    },
    {
      "epoch": 0.4824041743497195,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 2.4158,
      "step": 1537
    },
    {
      "epoch": 0.4827180352308839,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 2.0198,
      "step": 1538
    },
    {
      "epoch": 0.48303189611204833,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.0766,
      "step": 1539
    },
    {
      "epoch": 0.48334575699321275,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 2.4221,
      "step": 1540
    },
    {
      "epoch": 0.48365961787437717,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 2.1022,
      "step": 1541
    },
    {
      "epoch": 0.4839734787555416,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.1633,
      "step": 1542
    },
    {
      "epoch": 0.484287339636706,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 2.2258,
      "step": 1543
    },
    {
      "epoch": 0.4846012005178705,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.8711,
      "step": 1544
    },
    {
      "epoch": 0.4849150613990349,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 2.2828,
      "step": 1545
    },
    {
      "epoch": 0.4852289222801993,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.8074,
      "step": 1546
    },
    {
      "epoch": 0.48554278316136373,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 2.1619,
      "step": 1547
    },
    {
      "epoch": 0.48585664404252815,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 2.0935,
      "step": 1548
    },
    {
      "epoch": 0.48617050492369257,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 2.1691,
      "step": 1549
    },
    {
      "epoch": 0.486484365804857,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 2.8085,
      "step": 1550
    },
    {
      "epoch": 0.4867982266860214,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0002,
      "loss": 1.4212,
      "step": 1551
    },
    {
      "epoch": 0.4871120875671858,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 1.3257,
      "step": 1552
    },
    {
      "epoch": 0.48742594844835024,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.4715,
      "step": 1553
    },
    {
      "epoch": 0.4877398093295147,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.5925,
      "step": 1554
    },
    {
      "epoch": 0.48805367021067914,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.4331,
      "step": 1555
    },
    {
      "epoch": 0.48836753109184355,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.4069,
      "step": 1556
    },
    {
      "epoch": 0.488681391973008,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3879,
      "step": 1557
    },
    {
      "epoch": 0.4889952528541724,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.376,
      "step": 1558
    },
    {
      "epoch": 0.4893091137353368,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.3164,
      "step": 1559
    },
    {
      "epoch": 0.4896229746165012,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.5005,
      "step": 1560
    },
    {
      "epoch": 0.48993683549766565,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.4592,
      "step": 1561
    },
    {
      "epoch": 0.49025069637883006,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.501,
      "step": 1562
    },
    {
      "epoch": 0.4905645572599945,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.4858,
      "step": 1563
    },
    {
      "epoch": 0.49087841814115896,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.4335,
      "step": 1564
    },
    {
      "epoch": 0.4911922790223234,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4227,
      "step": 1565
    },
    {
      "epoch": 0.4915061399034878,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.5895,
      "step": 1566
    },
    {
      "epoch": 0.4918200007846522,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.4231,
      "step": 1567
    },
    {
      "epoch": 0.49213386166581663,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.4445,
      "step": 1568
    },
    {
      "epoch": 0.49244772254698105,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.5881,
      "step": 1569
    },
    {
      "epoch": 0.49276158342814547,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.5042,
      "step": 1570
    },
    {
      "epoch": 0.4930754443093099,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.567,
      "step": 1571
    },
    {
      "epoch": 0.4933893051904743,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.6221,
      "step": 1572
    },
    {
      "epoch": 0.4937031660716387,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.6948,
      "step": 1573
    },
    {
      "epoch": 0.4940170269528032,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.597,
      "step": 1574
    },
    {
      "epoch": 0.4943308878339676,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.6998,
      "step": 1575
    },
    {
      "epoch": 0.49464474871513203,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.5341,
      "step": 1576
    },
    {
      "epoch": 0.49495860959629645,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.7922,
      "step": 1577
    },
    {
      "epoch": 0.49527247047746087,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.7886,
      "step": 1578
    },
    {
      "epoch": 0.4955863313586253,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.7991,
      "step": 1579
    },
    {
      "epoch": 0.4959001922397897,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 2.1349,
      "step": 1580
    },
    {
      "epoch": 0.4962140531209541,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.8186,
      "step": 1581
    },
    {
      "epoch": 0.49652791400211854,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.5956,
      "step": 1582
    },
    {
      "epoch": 0.49684177488328296,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 2.3392,
      "step": 1583
    },
    {
      "epoch": 0.49715563576444743,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.8827,
      "step": 1584
    },
    {
      "epoch": 0.49746949664561185,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 2.1126,
      "step": 1585
    },
    {
      "epoch": 0.49778335752677627,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.847,
      "step": 1586
    },
    {
      "epoch": 0.4980972184079407,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 2.3026,
      "step": 1587
    },
    {
      "epoch": 0.4984110792891051,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 2.4587,
      "step": 1588
    },
    {
      "epoch": 0.4987249401702695,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.9976,
      "step": 1589
    },
    {
      "epoch": 0.49903880105143394,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.1905,
      "step": 1590
    },
    {
      "epoch": 0.49935266193259836,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 2.089,
      "step": 1591
    },
    {
      "epoch": 0.4996665228137628,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 2.9884,
      "step": 1592
    },
    {
      "epoch": 0.4999803836949272,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 2.4244,
      "step": 1593
    },
    {
      "epoch": 0.5002942445760916,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 2.0699,
      "step": 1594
    },
    {
      "epoch": 0.5006081054572561,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.6804,
      "step": 1595
    },
    {
      "epoch": 0.5009219663384205,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.8598,
      "step": 1596
    },
    {
      "epoch": 0.5012358272195849,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.8533,
      "step": 1597
    },
    {
      "epoch": 0.5015496881007493,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.03,
      "step": 1598
    },
    {
      "epoch": 0.5018635489819138,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 2.8499,
      "step": 1599
    },
    {
      "epoch": 0.5021774098630782,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 2.8263,
      "step": 1600
    },
    {
      "epoch": 0.5024912707442426,
      "grad_norm": 0.07275390625,
      "learning_rate": 0.0002,
      "loss": 1.3193,
      "step": 1601
    },
    {
      "epoch": 0.5028051316254071,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.4002,
      "step": 1602
    },
    {
      "epoch": 0.5031189925065714,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.3441,
      "step": 1603
    },
    {
      "epoch": 0.5034328533877359,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.4197,
      "step": 1604
    },
    {
      "epoch": 0.5037467142689003,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3579,
      "step": 1605
    },
    {
      "epoch": 0.5040605751500647,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.3139,
      "step": 1606
    },
    {
      "epoch": 0.5043744360312291,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3955,
      "step": 1607
    },
    {
      "epoch": 0.5046882969123936,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.6143,
      "step": 1608
    },
    {
      "epoch": 0.5050021577935581,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.3207,
      "step": 1609
    },
    {
      "epoch": 0.5053160186747224,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.3933,
      "step": 1610
    },
    {
      "epoch": 0.5056298795558869,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.449,
      "step": 1611
    },
    {
      "epoch": 0.5059437404370513,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.342,
      "step": 1612
    },
    {
      "epoch": 0.5062576013182157,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.457,
      "step": 1613
    },
    {
      "epoch": 0.5065714621993801,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3511,
      "step": 1614
    },
    {
      "epoch": 0.5068853230805446,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.3069,
      "step": 1615
    },
    {
      "epoch": 0.5071991839617089,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.4269,
      "step": 1616
    },
    {
      "epoch": 0.5075130448428734,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.5465,
      "step": 1617
    },
    {
      "epoch": 0.5078269057240378,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.6049,
      "step": 1618
    },
    {
      "epoch": 0.5081407666052022,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.411,
      "step": 1619
    },
    {
      "epoch": 0.5084546274863667,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.4713,
      "step": 1620
    },
    {
      "epoch": 0.5087684883675311,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.5034,
      "step": 1621
    },
    {
      "epoch": 0.5090823492486956,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.6373,
      "step": 1622
    },
    {
      "epoch": 0.5093962101298599,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.7171,
      "step": 1623
    },
    {
      "epoch": 0.5097100710110244,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.415,
      "step": 1624
    },
    {
      "epoch": 0.5100239318921888,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.8343,
      "step": 1625
    },
    {
      "epoch": 0.5103377927733532,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.869,
      "step": 1626
    },
    {
      "epoch": 0.5106516536545176,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.7027,
      "step": 1627
    },
    {
      "epoch": 0.5109655145356821,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.7334,
      "step": 1628
    },
    {
      "epoch": 0.5112793754168465,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.8781,
      "step": 1629
    },
    {
      "epoch": 0.5115932362980109,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.7577,
      "step": 1630
    },
    {
      "epoch": 0.5119070971791754,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 2.245,
      "step": 1631
    },
    {
      "epoch": 0.5122209580603397,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.851,
      "step": 1632
    },
    {
      "epoch": 0.5125348189415042,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.8517,
      "step": 1633
    },
    {
      "epoch": 0.5128486798226686,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 2.6865,
      "step": 1634
    },
    {
      "epoch": 0.513162540703833,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 2.5321,
      "step": 1635
    },
    {
      "epoch": 0.5134764015849974,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.2039,
      "step": 1636
    },
    {
      "epoch": 0.5137902624661619,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 2.0997,
      "step": 1637
    },
    {
      "epoch": 0.5141041233473262,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 2.5026,
      "step": 1638
    },
    {
      "epoch": 0.5144179842284907,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 2.027,
      "step": 1639
    },
    {
      "epoch": 0.5147318451096552,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.9668,
      "step": 1640
    },
    {
      "epoch": 0.5150457059908196,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.2957,
      "step": 1641
    },
    {
      "epoch": 0.515359566871984,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 2.3475,
      "step": 1642
    },
    {
      "epoch": 0.5156734277531484,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 2.2628,
      "step": 1643
    },
    {
      "epoch": 0.5159872886343129,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 2.5057,
      "step": 1644
    },
    {
      "epoch": 0.5163011495154772,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.9935,
      "step": 1645
    },
    {
      "epoch": 0.5166150103966417,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 2.4644,
      "step": 1646
    },
    {
      "epoch": 0.5169288712778061,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 2.177,
      "step": 1647
    },
    {
      "epoch": 0.5172427321589705,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.3228,
      "step": 1648
    },
    {
      "epoch": 0.517556593040135,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.629,
      "step": 1649
    },
    {
      "epoch": 0.5178704539212994,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 2.8592,
      "step": 1650
    },
    {
      "epoch": 0.5181843148024639,
      "grad_norm": 0.07763671875,
      "learning_rate": 0.0002,
      "loss": 1.34,
      "step": 1651
    },
    {
      "epoch": 0.5184981756836282,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.4626,
      "step": 1652
    },
    {
      "epoch": 0.5188120365647927,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.5062,
      "step": 1653
    },
    {
      "epoch": 0.519125897445957,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.4513,
      "step": 1654
    },
    {
      "epoch": 0.5194397583271215,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.2568,
      "step": 1655
    },
    {
      "epoch": 0.5197536192082859,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.408,
      "step": 1656
    },
    {
      "epoch": 0.5200674800894504,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2524,
      "step": 1657
    },
    {
      "epoch": 0.5203813409706147,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.3407,
      "step": 1658
    },
    {
      "epoch": 0.5206952018517792,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.4142,
      "step": 1659
    },
    {
      "epoch": 0.5210090627329437,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.4058,
      "step": 1660
    },
    {
      "epoch": 0.521322923614108,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.3903,
      "step": 1661
    },
    {
      "epoch": 0.5216367844952725,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.3566,
      "step": 1662
    },
    {
      "epoch": 0.5219506453764369,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.4786,
      "step": 1663
    },
    {
      "epoch": 0.5222645062576013,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.5583,
      "step": 1664
    },
    {
      "epoch": 0.5225783671387657,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4332,
      "step": 1665
    },
    {
      "epoch": 0.5228922280199302,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.5549,
      "step": 1666
    },
    {
      "epoch": 0.5232060889010945,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.544,
      "step": 1667
    },
    {
      "epoch": 0.523519949782259,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.5544,
      "step": 1668
    },
    {
      "epoch": 0.5238338106634234,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.8305,
      "step": 1669
    },
    {
      "epoch": 0.5241476715445879,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.4007,
      "step": 1670
    },
    {
      "epoch": 0.5244615324257523,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.6236,
      "step": 1671
    },
    {
      "epoch": 0.5247753933069167,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.7004,
      "step": 1672
    },
    {
      "epoch": 0.5250892541880812,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.5039,
      "step": 1673
    },
    {
      "epoch": 0.5254031150692455,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.5396,
      "step": 1674
    },
    {
      "epoch": 0.52571697595041,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.6841,
      "step": 1675
    },
    {
      "epoch": 0.5260308368315744,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.3799,
      "step": 1676
    },
    {
      "epoch": 0.5263446977127388,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.3486,
      "step": 1677
    },
    {
      "epoch": 0.5266585585939032,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.7302,
      "step": 1678
    },
    {
      "epoch": 0.5269724194750677,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.9683,
      "step": 1679
    },
    {
      "epoch": 0.5272862803562322,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 2.2732,
      "step": 1680
    },
    {
      "epoch": 0.5276001412373965,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 2.1772,
      "step": 1681
    },
    {
      "epoch": 0.527914002118561,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.9295,
      "step": 1682
    },
    {
      "epoch": 0.5282278629997254,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.9274,
      "step": 1683
    },
    {
      "epoch": 0.5282278629997254,
      "eval_loss": 1.7762224674224854,
      "eval_runtime": 123.1506,
      "eval_samples_per_second": 8.12,
      "eval_steps_per_second": 8.12,
      "step": 1683
    },
    {
      "epoch": 0.5282278629997254,
      "mmlu_eval_accuracy": 0.438688172318441,
      "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": 0.3448275862068966,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.5625,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.3125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.65,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.5652173913043478,
      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.68,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.5930232558139535,
      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.5151515151515151,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.37058823529411766,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.3188405797101449,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.4215357886623667,
      "step": 1683
    },
    {
      "epoch": 0.5285417238808898,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 2.0894,
      "step": 1684
    },
    {
      "epoch": 0.5288555847620542,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 2.2266,
      "step": 1685
    },
    {
      "epoch": 0.5291694456432187,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 2.0832,
      "step": 1686
    },
    {
      "epoch": 0.529483306524383,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 2.0019,
      "step": 1687
    },
    {
      "epoch": 0.5297971674055475,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.0644,
      "step": 1688
    },
    {
      "epoch": 0.5301110282867119,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 2.1444,
      "step": 1689
    },
    {
      "epoch": 0.5304248891678763,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 2.3479,
      "step": 1690
    },
    {
      "epoch": 0.5307387500490408,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.8993,
      "step": 1691
    },
    {
      "epoch": 0.5310526109302052,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 2.1527,
      "step": 1692
    },
    {
      "epoch": 0.5313664718113696,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 2.3942,
      "step": 1693
    },
    {
      "epoch": 0.531680332692534,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 2.4652,
      "step": 1694
    },
    {
      "epoch": 0.5319941935736985,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.5987,
      "step": 1695
    },
    {
      "epoch": 0.5323080544548628,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.9468,
      "step": 1696
    },
    {
      "epoch": 0.5326219153360273,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.8692,
      "step": 1697
    },
    {
      "epoch": 0.5329357762171917,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 2.0162,
      "step": 1698
    },
    {
      "epoch": 0.5332496370983562,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 2.8707,
      "step": 1699
    },
    {
      "epoch": 0.5335634979795206,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.8278,
      "step": 1700
    },
    {
      "epoch": 0.533877358860685,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0002,
      "loss": 1.4067,
      "step": 1701
    },
    {
      "epoch": 0.5341912197418495,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.4477,
      "step": 1702
    },
    {
      "epoch": 0.5345050806230138,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.4612,
      "step": 1703
    },
    {
      "epoch": 0.5348189415041783,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.5085,
      "step": 1704
    },
    {
      "epoch": 0.5351328023853427,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3591,
      "step": 1705
    },
    {
      "epoch": 0.5354466632665071,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.3802,
      "step": 1706
    },
    {
      "epoch": 0.5357605241476715,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.468,
      "step": 1707
    },
    {
      "epoch": 0.536074385028836,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.4686,
      "step": 1708
    },
    {
      "epoch": 0.5363882459100003,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3237,
      "step": 1709
    },
    {
      "epoch": 0.5367021067911648,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.4139,
      "step": 1710
    },
    {
      "epoch": 0.5370159676723293,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3979,
      "step": 1711
    },
    {
      "epoch": 0.5373298285534936,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3369,
      "step": 1712
    },
    {
      "epoch": 0.5376436894346581,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.289,
      "step": 1713
    },
    {
      "epoch": 0.5379575503158225,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.361,
      "step": 1714
    },
    {
      "epoch": 0.538271411196987,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.3253,
      "step": 1715
    },
    {
      "epoch": 0.5385852720781513,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.5354,
      "step": 1716
    },
    {
      "epoch": 0.5388991329593158,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.5638,
      "step": 1717
    },
    {
      "epoch": 0.5392129938404802,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.481,
      "step": 1718
    },
    {
      "epoch": 0.5395268547216446,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3881,
      "step": 1719
    },
    {
      "epoch": 0.5398407156028091,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.6546,
      "step": 1720
    },
    {
      "epoch": 0.5401545764839735,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3408,
      "step": 1721
    },
    {
      "epoch": 0.5404684373651379,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.5049,
      "step": 1722
    },
    {
      "epoch": 0.5407822982463023,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4163,
      "step": 1723
    },
    {
      "epoch": 0.5410961591274668,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.7485,
      "step": 1724
    },
    {
      "epoch": 0.5414100200086311,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.4358,
      "step": 1725
    },
    {
      "epoch": 0.5417238808897956,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.7268,
      "step": 1726
    },
    {
      "epoch": 0.54203774177096,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.4654,
      "step": 1727
    },
    {
      "epoch": 0.5423516026521245,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 2.0668,
      "step": 1728
    },
    {
      "epoch": 0.5426654635332888,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 2.0296,
      "step": 1729
    },
    {
      "epoch": 0.5429793244144533,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.8383,
      "step": 1730
    },
    {
      "epoch": 0.5432931852956178,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.6799,
      "step": 1731
    },
    {
      "epoch": 0.5436070461767821,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.9518,
      "step": 1732
    },
    {
      "epoch": 0.5439209070579466,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 2.1723,
      "step": 1733
    },
    {
      "epoch": 0.544234767939111,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.9347,
      "step": 1734
    },
    {
      "epoch": 0.5445486288202754,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 2.0049,
      "step": 1735
    },
    {
      "epoch": 0.5448624897014398,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 2.1533,
      "step": 1736
    },
    {
      "epoch": 0.5451763505826043,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.9575,
      "step": 1737
    },
    {
      "epoch": 0.5454902114637686,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 2.0899,
      "step": 1738
    },
    {
      "epoch": 0.5458040723449331,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 2.5263,
      "step": 1739
    },
    {
      "epoch": 0.5461179332260976,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 2.1424,
      "step": 1740
    },
    {
      "epoch": 0.546431794107262,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 2.0427,
      "step": 1741
    },
    {
      "epoch": 0.5467456549884264,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.9787,
      "step": 1742
    },
    {
      "epoch": 0.5470595158695908,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.9796,
      "step": 1743
    },
    {
      "epoch": 0.5473733767507553,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.8059,
      "step": 1744
    },
    {
      "epoch": 0.5476872376319196,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 2.2813,
      "step": 1745
    },
    {
      "epoch": 0.5480010985130841,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.2681,
      "step": 1746
    },
    {
      "epoch": 0.5483149593942485,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 2.3991,
      "step": 1747
    },
    {
      "epoch": 0.5486288202754129,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 2.5695,
      "step": 1748
    },
    {
      "epoch": 0.5489426811565773,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.3254,
      "step": 1749
    },
    {
      "epoch": 0.5492565420377418,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 2.7664,
      "step": 1750
    },
    {
      "epoch": 0.5495704029189062,
      "grad_norm": 0.078125,
      "learning_rate": 0.0002,
      "loss": 1.4485,
      "step": 1751
    },
    {
      "epoch": 0.5498842638000706,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.4486,
      "step": 1752
    },
    {
      "epoch": 0.5501981246812351,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.4205,
      "step": 1753
    },
    {
      "epoch": 0.5505119855623994,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.4582,
      "step": 1754
    },
    {
      "epoch": 0.5508258464435639,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.503,
      "step": 1755
    },
    {
      "epoch": 0.5511397073247283,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.4567,
      "step": 1756
    },
    {
      "epoch": 0.5514535682058928,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.3373,
      "step": 1757
    },
    {
      "epoch": 0.5517674290870571,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4963,
      "step": 1758
    },
    {
      "epoch": 0.5520812899682216,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.3178,
      "step": 1759
    },
    {
      "epoch": 0.5523951508493861,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4644,
      "step": 1760
    },
    {
      "epoch": 0.5527090117305504,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.5167,
      "step": 1761
    },
    {
      "epoch": 0.5530228726117149,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3299,
      "step": 1762
    },
    {
      "epoch": 0.5533367334928793,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2503,
      "step": 1763
    },
    {
      "epoch": 0.5536505943740437,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4346,
      "step": 1764
    },
    {
      "epoch": 0.5539644552552081,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3315,
      "step": 1765
    },
    {
      "epoch": 0.5542783161363726,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3975,
      "step": 1766
    },
    {
      "epoch": 0.5545921770175369,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1707,
      "step": 1767
    },
    {
      "epoch": 0.5549060378987014,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.6337,
      "step": 1768
    },
    {
      "epoch": 0.5552198987798658,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.4859,
      "step": 1769
    },
    {
      "epoch": 0.5555337596610302,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.521,
      "step": 1770
    },
    {
      "epoch": 0.5558476205421947,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.4928,
      "step": 1771
    },
    {
      "epoch": 0.5561614814233591,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.5976,
      "step": 1772
    },
    {
      "epoch": 0.5564753423045236,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.4979,
      "step": 1773
    },
    {
      "epoch": 0.5567892031856879,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.5024,
      "step": 1774
    },
    {
      "epoch": 0.5571030640668524,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.412,
      "step": 1775
    },
    {
      "epoch": 0.5574169249480168,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.5321,
      "step": 1776
    },
    {
      "epoch": 0.5577307858291812,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.7385,
      "step": 1777
    },
    {
      "epoch": 0.5580446467103456,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.7201,
      "step": 1778
    },
    {
      "epoch": 0.5583585075915101,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.9139,
      "step": 1779
    },
    {
      "epoch": 0.5586723684726745,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 2.0861,
      "step": 1780
    },
    {
      "epoch": 0.5589862293538389,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 2.155,
      "step": 1781
    },
    {
      "epoch": 0.5593000902350034,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 2.086,
      "step": 1782
    },
    {
      "epoch": 0.5596139511161677,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.937,
      "step": 1783
    },
    {
      "epoch": 0.5599278119973322,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 2.3259,
      "step": 1784
    },
    {
      "epoch": 0.5602416728784966,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 2.1166,
      "step": 1785
    },
    {
      "epoch": 0.560555533759661,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 2.0513,
      "step": 1786
    },
    {
      "epoch": 0.5608693946408254,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 2.4344,
      "step": 1787
    },
    {
      "epoch": 0.5611832555219899,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.3386,
      "step": 1788
    },
    {
      "epoch": 0.5614971164031543,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.6012,
      "step": 1789
    },
    {
      "epoch": 0.5618109772843187,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 2.0929,
      "step": 1790
    },
    {
      "epoch": 0.5621248381654832,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 2.3641,
      "step": 1791
    },
    {
      "epoch": 0.5624386990466476,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 2.5071,
      "step": 1792
    },
    {
      "epoch": 0.562752559927812,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 2.3544,
      "step": 1793
    },
    {
      "epoch": 0.5630664208089764,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.0547,
      "step": 1794
    },
    {
      "epoch": 0.5633802816901409,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 2.2086,
      "step": 1795
    },
    {
      "epoch": 0.5636941425713052,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.958,
      "step": 1796
    },
    {
      "epoch": 0.5640080034524697,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.3892,
      "step": 1797
    },
    {
      "epoch": 0.5643218643336341,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 2.1997,
      "step": 1798
    },
    {
      "epoch": 0.5646357252147985,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 2.2381,
      "step": 1799
    },
    {
      "epoch": 0.564949586095963,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 2.6629,
      "step": 1800
    },
    {
      "epoch": 0.5652634469771274,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0002,
      "loss": 1.3331,
      "step": 1801
    },
    {
      "epoch": 0.5655773078582919,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 1.4385,
      "step": 1802
    },
    {
      "epoch": 0.5658911687394562,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.4304,
      "step": 1803
    },
    {
      "epoch": 0.5662050296206207,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4488,
      "step": 1804
    },
    {
      "epoch": 0.566518890501785,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.4353,
      "step": 1805
    },
    {
      "epoch": 0.5668327513829495,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.4794,
      "step": 1806
    },
    {
      "epoch": 0.5671466122641139,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.4254,
      "step": 1807
    },
    {
      "epoch": 0.5674604731452784,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.5094,
      "step": 1808
    },
    {
      "epoch": 0.5677743340264427,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.4609,
      "step": 1809
    },
    {
      "epoch": 0.5680881949076072,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.1764,
      "step": 1810
    },
    {
      "epoch": 0.5684020557887717,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.3245,
      "step": 1811
    },
    {
      "epoch": 0.568715916669936,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4325,
      "step": 1812
    },
    {
      "epoch": 0.5690297775511005,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.4363,
      "step": 1813
    },
    {
      "epoch": 0.5693436384322649,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.4602,
      "step": 1814
    },
    {
      "epoch": 0.5696574993134293,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2947,
      "step": 1815
    },
    {
      "epoch": 0.5699713601945937,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.4927,
      "step": 1816
    },
    {
      "epoch": 0.5702852210757582,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.4334,
      "step": 1817
    },
    {
      "epoch": 0.5705990819569225,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.3099,
      "step": 1818
    },
    {
      "epoch": 0.570912942838087,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.409,
      "step": 1819
    },
    {
      "epoch": 0.5712268037192515,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.5078,
      "step": 1820
    },
    {
      "epoch": 0.5715406646004159,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.6973,
      "step": 1821
    },
    {
      "epoch": 0.5718545254815803,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.5313,
      "step": 1822
    },
    {
      "epoch": 0.5721683863627447,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.4193,
      "step": 1823
    },
    {
      "epoch": 0.5724822472439092,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.6857,
      "step": 1824
    },
    {
      "epoch": 0.5727961081250735,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.7664,
      "step": 1825
    },
    {
      "epoch": 0.573109969006238,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.6576,
      "step": 1826
    },
    {
      "epoch": 0.5734238298874024,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.1437,
      "step": 1827
    },
    {
      "epoch": 0.5737376907685668,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 2.103,
      "step": 1828
    },
    {
      "epoch": 0.5740515516497312,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.8686,
      "step": 1829
    },
    {
      "epoch": 0.5743654125308957,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.706,
      "step": 1830
    },
    {
      "epoch": 0.5746792734120602,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.7805,
      "step": 1831
    },
    {
      "epoch": 0.5749931342932245,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 2.52,
      "step": 1832
    },
    {
      "epoch": 0.575306995174389,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.8605,
      "step": 1833
    },
    {
      "epoch": 0.5756208560555534,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 2.0258,
      "step": 1834
    },
    {
      "epoch": 0.5759347169367178,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 2.0296,
      "step": 1835
    },
    {
      "epoch": 0.5762485778178822,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.0942,
      "step": 1836
    },
    {
      "epoch": 0.5765624386990467,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 2.1924,
      "step": 1837
    },
    {
      "epoch": 0.576876299580211,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 2.4209,
      "step": 1838
    },
    {
      "epoch": 0.5771901604613755,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 1.9007,
      "step": 1839
    },
    {
      "epoch": 0.5775040213425399,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 3.0011,
      "step": 1840
    },
    {
      "epoch": 0.5778178822237043,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.9979,
      "step": 1841
    },
    {
      "epoch": 0.5781317431048688,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 2.0207,
      "step": 1842
    },
    {
      "epoch": 0.5784456039860332,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 1.9656,
      "step": 1843
    },
    {
      "epoch": 0.5787594648671976,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 2.6595,
      "step": 1844
    },
    {
      "epoch": 0.579073325748362,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 2.6086,
      "step": 1845
    },
    {
      "epoch": 0.5793871866295265,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 2.3001,
      "step": 1846
    },
    {
      "epoch": 0.5797010475106908,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.8254,
      "step": 1847
    },
    {
      "epoch": 0.5800149083918553,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 2.1071,
      "step": 1848
    },
    {
      "epoch": 0.5803287692730197,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 2.1997,
      "step": 1849
    },
    {
      "epoch": 0.5806426301541842,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.7012,
      "step": 1850
    },
    {
      "epoch": 0.5809564910353486,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.3707,
      "step": 1851
    },
    {
      "epoch": 0.581270351916513,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.3206,
      "step": 1852
    },
    {
      "epoch": 0.5815842127976775,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.5333,
      "step": 1853
    },
    {
      "epoch": 0.5818980736788418,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.6303,
      "step": 1854
    },
    {
      "epoch": 0.5822119345600063,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.4587,
      "step": 1855
    },
    {
      "epoch": 0.5825257954411707,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.4127,
      "step": 1856
    },
    {
      "epoch": 0.5828396563223351,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.3319,
      "step": 1857
    },
    {
      "epoch": 0.5831535172034995,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4086,
      "step": 1858
    },
    {
      "epoch": 0.583467378084664,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4039,
      "step": 1859
    },
    {
      "epoch": 0.5837812389658283,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.4437,
      "step": 1860
    },
    {
      "epoch": 0.5840950998469928,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4793,
      "step": 1861
    },
    {
      "epoch": 0.5844089607281573,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.5648,
      "step": 1862
    },
    {
      "epoch": 0.5847228216093217,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.5329,
      "step": 1863
    },
    {
      "epoch": 0.5850366824904861,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.5164,
      "step": 1864
    },
    {
      "epoch": 0.5853505433716505,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.592,
      "step": 1865
    },
    {
      "epoch": 0.585664404252815,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4296,
      "step": 1866
    },
    {
      "epoch": 0.5859782651339793,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.4401,
      "step": 1867
    },
    {
      "epoch": 0.5862921260151438,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.4334,
      "step": 1868
    },
    {
      "epoch": 0.5866059868963082,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.3723,
      "step": 1869
    },
    {
      "epoch": 0.5869198477774726,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.513,
      "step": 1870
    },
    {
      "epoch": 0.5869198477774726,
      "eval_loss": 1.7979594469070435,
      "eval_runtime": 123.3498,
      "eval_samples_per_second": 8.107,
      "eval_steps_per_second": 8.107,
      "step": 1870
    },
    {
      "epoch": 0.5869198477774726,
      "mmlu_eval_accuracy": 0.4160179039656003,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3448275862068966,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.0,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.22727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.3125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.3333333333333333,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.5833333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
      "mmlu_eval_accuracy_miscellaneous": 0.5930232558139535,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.42424242424242425,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.4,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.36470588235294116,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.2989568939635678,
      "step": 1870
    },
    {
      "epoch": 0.5872337086586371,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.4714,
      "step": 1871
    },
    {
      "epoch": 0.5875475695398015,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.6044,
      "step": 1872
    },
    {
      "epoch": 0.587861430420966,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.7849,
      "step": 1873
    },
    {
      "epoch": 0.5881752913021303,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.8635,
      "step": 1874
    },
    {
      "epoch": 0.5884891521832948,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.7972,
      "step": 1875
    },
    {
      "epoch": 0.5888030130644591,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.7615,
      "step": 1876
    },
    {
      "epoch": 0.5891168739456236,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.6072,
      "step": 1877
    },
    {
      "epoch": 0.589430734826788,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.8632,
      "step": 1878
    },
    {
      "epoch": 0.5897445957079525,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 2.0262,
      "step": 1879
    },
    {
      "epoch": 0.5900584565891168,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002,
      "loss": 2.0907,
      "step": 1880
    },
    {
      "epoch": 0.5903723174702813,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.9217,
      "step": 1881
    },
    {
      "epoch": 0.5906861783514458,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.2323,
      "step": 1882
    },
    {
      "epoch": 0.5910000392326101,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 2.0756,
      "step": 1883
    },
    {
      "epoch": 0.5913139001137746,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 2.1437,
      "step": 1884
    },
    {
      "epoch": 0.591627760994939,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 2.1764,
      "step": 1885
    },
    {
      "epoch": 0.5919416218761034,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.9309,
      "step": 1886
    },
    {
      "epoch": 0.5922554827572678,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 2.2414,
      "step": 1887
    },
    {
      "epoch": 0.5925693436384323,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 2.5843,
      "step": 1888
    },
    {
      "epoch": 0.5928832045195966,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 1.8418,
      "step": 1889
    },
    {
      "epoch": 0.5931970654007611,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.9504,
      "step": 1890
    },
    {
      "epoch": 0.5935109262819256,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 2.0956,
      "step": 1891
    },
    {
      "epoch": 0.59382478716309,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.7819,
      "step": 1892
    },
    {
      "epoch": 0.5941386480442544,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 2.29,
      "step": 1893
    },
    {
      "epoch": 0.5944525089254188,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.695,
      "step": 1894
    },
    {
      "epoch": 0.5947663698065833,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 2.5746,
      "step": 1895
    },
    {
      "epoch": 0.5950802306877476,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.748,
      "step": 1896
    },
    {
      "epoch": 0.5953940915689121,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.7555,
      "step": 1897
    },
    {
      "epoch": 0.5957079524500765,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 2.1987,
      "step": 1898
    },
    {
      "epoch": 0.5960218133312409,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 2.6141,
      "step": 1899
    },
    {
      "epoch": 0.5963356742124053,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 3.0325,
      "step": 1900
    },
    {
      "epoch": 0.5966495350935698,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 1.353,
      "step": 1901
    },
    {
      "epoch": 0.5969633959747342,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.4564,
      "step": 1902
    },
    {
      "epoch": 0.5972772568558986,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.3223,
      "step": 1903
    },
    {
      "epoch": 0.5975911177370631,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3212,
      "step": 1904
    },
    {
      "epoch": 0.5979049786182274,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4801,
      "step": 1905
    },
    {
      "epoch": 0.5982188394993919,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.2425,
      "step": 1906
    },
    {
      "epoch": 0.5985327003805563,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.5301,
      "step": 1907
    },
    {
      "epoch": 0.5988465612617208,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.6526,
      "step": 1908
    },
    {
      "epoch": 0.5991604221428851,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.359,
      "step": 1909
    },
    {
      "epoch": 0.5994742830240496,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3382,
      "step": 1910
    },
    {
      "epoch": 0.5997881439052141,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.5714,
      "step": 1911
    },
    {
      "epoch": 0.6001020047863784,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4322,
      "step": 1912
    },
    {
      "epoch": 0.6004158656675429,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.5023,
      "step": 1913
    },
    {
      "epoch": 0.6007297265487073,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4193,
      "step": 1914
    },
    {
      "epoch": 0.6010435874298717,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.4606,
      "step": 1915
    },
    {
      "epoch": 0.6013574483110361,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4069,
      "step": 1916
    },
    {
      "epoch": 0.6016713091922006,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.635,
      "step": 1917
    },
    {
      "epoch": 0.6019851700733649,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.4834,
      "step": 1918
    },
    {
      "epoch": 0.6022990309545294,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.4178,
      "step": 1919
    },
    {
      "epoch": 0.6026128918356938,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.4411,
      "step": 1920
    },
    {
      "epoch": 0.6029267527168582,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.5469,
      "step": 1921
    },
    {
      "epoch": 0.6032406135980227,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.3598,
      "step": 1922
    },
    {
      "epoch": 0.6035544744791871,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.4062,
      "step": 1923
    },
    {
      "epoch": 0.6038683353603516,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.469,
      "step": 1924
    },
    {
      "epoch": 0.6041821962415159,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.6936,
      "step": 1925
    },
    {
      "epoch": 0.6044960571226804,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002,
      "loss": 1.9837,
      "step": 1926
    },
    {
      "epoch": 0.6048099180038448,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.9,
      "step": 1927
    },
    {
      "epoch": 0.6051237788850092,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.4733,
      "step": 1928
    },
    {
      "epoch": 0.6054376397661736,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.5585,
      "step": 1929
    },
    {
      "epoch": 0.6057515006473381,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.9682,
      "step": 1930
    },
    {
      "epoch": 0.6060653615285025,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.7815,
      "step": 1931
    },
    {
      "epoch": 0.6063792224096669,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.9774,
      "step": 1932
    },
    {
      "epoch": 0.6066930832908314,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 2.0198,
      "step": 1933
    },
    {
      "epoch": 0.6070069441719957,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.9991,
      "step": 1934
    },
    {
      "epoch": 0.6073208050531602,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 2.0222,
      "step": 1935
    },
    {
      "epoch": 0.6076346659343246,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 2.1085,
      "step": 1936
    },
    {
      "epoch": 0.607948526815489,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.104,
      "step": 1937
    },
    {
      "epoch": 0.6082623876966534,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 2.2331,
      "step": 1938
    },
    {
      "epoch": 0.6085762485778179,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 2.3148,
      "step": 1939
    },
    {
      "epoch": 0.6088901094589823,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.8777,
      "step": 1940
    },
    {
      "epoch": 0.6092039703401467,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 2.0675,
      "step": 1941
    },
    {
      "epoch": 0.6095178312213112,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 2.0153,
      "step": 1942
    },
    {
      "epoch": 0.6098316921024756,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.746,
      "step": 1943
    },
    {
      "epoch": 0.61014555298364,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.3711,
      "step": 1944
    },
    {
      "epoch": 0.6104594138648044,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.7351,
      "step": 1945
    },
    {
      "epoch": 0.6107732747459689,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.8861,
      "step": 1946
    },
    {
      "epoch": 0.6110871356271332,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.8354,
      "step": 1947
    },
    {
      "epoch": 0.6114009965082977,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 2.2304,
      "step": 1948
    },
    {
      "epoch": 0.6117148573894621,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.6094,
      "step": 1949
    },
    {
      "epoch": 0.6120287182706265,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002,
      "loss": 2.7485,
      "step": 1950
    },
    {
      "epoch": 0.612342579151791,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.5545,
      "step": 1951
    },
    {
      "epoch": 0.6126564400329554,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.4116,
      "step": 1952
    },
    {
      "epoch": 0.6129703009141199,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3595,
      "step": 1953
    },
    {
      "epoch": 0.6132841617952842,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.4558,
      "step": 1954
    },
    {
      "epoch": 0.6135980226764487,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.6147,
      "step": 1955
    },
    {
      "epoch": 0.6139118835576131,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3496,
      "step": 1956
    },
    {
      "epoch": 0.6142257444387775,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.4669,
      "step": 1957
    },
    {
      "epoch": 0.6145396053199419,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.4654,
      "step": 1958
    },
    {
      "epoch": 0.6148534662011064,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.4288,
      "step": 1959
    },
    {
      "epoch": 0.6151673270822707,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3818,
      "step": 1960
    },
    {
      "epoch": 0.6154811879634352,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.4344,
      "step": 1961
    },
    {
      "epoch": 0.6157950488445997,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.3964,
      "step": 1962
    },
    {
      "epoch": 0.616108909725764,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.5546,
      "step": 1963
    },
    {
      "epoch": 0.6164227706069285,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.5167,
      "step": 1964
    },
    {
      "epoch": 0.6167366314880929,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3906,
      "step": 1965
    },
    {
      "epoch": 0.6170504923692574,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.5305,
      "step": 1966
    },
    {
      "epoch": 0.6173643532504217,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.4886,
      "step": 1967
    },
    {
      "epoch": 0.6176782141315862,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.3195,
      "step": 1968
    },
    {
      "epoch": 0.6179920750127506,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.6338,
      "step": 1969
    },
    {
      "epoch": 0.618305935893915,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.1924,
      "step": 1970
    },
    {
      "epoch": 0.6186197967750795,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0002,
      "loss": 1.8635,
      "step": 1971
    },
    {
      "epoch": 0.6189336576562439,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0002,
      "loss": 1.705,
      "step": 1972
    },
    {
      "epoch": 0.6192475185374083,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 1.4895,
      "step": 1973
    },
    {
      "epoch": 0.6195613794185727,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.4939,
      "step": 1974
    },
    {
      "epoch": 0.6198752402997372,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.551,
      "step": 1975
    },
    {
      "epoch": 0.6201891011809015,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.6042,
      "step": 1976
    },
    {
      "epoch": 0.620502962062066,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.6444,
      "step": 1977
    },
    {
      "epoch": 0.6208168229432304,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.2018,
      "step": 1978
    },
    {
      "epoch": 0.6211306838243948,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.7853,
      "step": 1979
    },
    {
      "epoch": 0.6214445447055592,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.7374,
      "step": 1980
    },
    {
      "epoch": 0.6217584055867237,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.6137,
      "step": 1981
    },
    {
      "epoch": 0.6220722664678882,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.2201,
      "step": 1982
    },
    {
      "epoch": 0.6223861273490525,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.8339,
      "step": 1983
    },
    {
      "epoch": 0.622699988230217,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002,
      "loss": 2.3659,
      "step": 1984
    },
    {
      "epoch": 0.6230138491113814,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 2.1963,
      "step": 1985
    },
    {
      "epoch": 0.6233277099925458,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.8056,
      "step": 1986
    },
    {
      "epoch": 0.6236415708737102,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 2.1867,
      "step": 1987
    },
    {
      "epoch": 0.6239554317548747,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.9651,
      "step": 1988
    },
    {
      "epoch": 0.624269292636039,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 2.3103,
      "step": 1989
    },
    {
      "epoch": 0.6245831535172035,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 2.2022,
      "step": 1990
    },
    {
      "epoch": 0.624897014398368,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 2.1621,
      "step": 1991
    },
    {
      "epoch": 0.6252108752795323,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.5333,
      "step": 1992
    },
    {
      "epoch": 0.6255247361606968,
      "grad_norm": 0.953125,
      "learning_rate": 0.0002,
      "loss": 2.9862,
      "step": 1993
    },
    {
      "epoch": 0.6258385970418612,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.0522,
      "step": 1994
    },
    {
      "epoch": 0.6261524579230257,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.937,
      "step": 1995
    },
    {
      "epoch": 0.62646631880419,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 2.407,
      "step": 1996
    },
    {
      "epoch": 0.6267801796853545,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 2.128,
      "step": 1997
    },
    {
      "epoch": 0.6270940405665189,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 2.4289,
      "step": 1998
    },
    {
      "epoch": 0.6274079014476833,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 2.9464,
      "step": 1999
    },
    {
      "epoch": 0.6277217623288477,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 2.3799,
      "step": 2000
    },
    {
      "epoch": 0.6280356232100122,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0002,
      "loss": 1.3815,
      "step": 2001
    },
    {
      "epoch": 0.6283494840911766,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.4014,
      "step": 2002
    },
    {
      "epoch": 0.628663344972341,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.5265,
      "step": 2003
    },
    {
      "epoch": 0.6289772058535055,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.524,
      "step": 2004
    },
    {
      "epoch": 0.6292910667346698,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.5699,
      "step": 2005
    },
    {
      "epoch": 0.6296049276158343,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3892,
      "step": 2006
    },
    {
      "epoch": 0.6299187884969987,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4151,
      "step": 2007
    },
    {
      "epoch": 0.6302326493781631,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.4622,
      "step": 2008
    },
    {
      "epoch": 0.6305465102593275,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4662,
      "step": 2009
    },
    {
      "epoch": 0.630860371140492,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.4608,
      "step": 2010
    },
    {
      "epoch": 0.6311742320216565,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.4231,
      "step": 2011
    },
    {
      "epoch": 0.6314880929028208,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4065,
      "step": 2012
    },
    {
      "epoch": 0.6318019537839853,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.683,
      "step": 2013
    },
    {
      "epoch": 0.6321158146651497,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.4582,
      "step": 2014
    },
    {
      "epoch": 0.6324296755463141,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.389,
      "step": 2015
    },
    {
      "epoch": 0.6327435364274785,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.2963,
      "step": 2016
    },
    {
      "epoch": 0.633057397308643,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.6131,
      "step": 2017
    },
    {
      "epoch": 0.6333712581898073,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.4931,
      "step": 2018
    },
    {
      "epoch": 0.6336851190709718,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.6715,
      "step": 2019
    },
    {
      "epoch": 0.6339989799521362,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.4375,
      "step": 2020
    },
    {
      "epoch": 0.6343128408333006,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.6019,
      "step": 2021
    },
    {
      "epoch": 0.6346267017144651,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.37,
      "step": 2022
    },
    {
      "epoch": 0.6349405625956295,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.5931,
      "step": 2023
    },
    {
      "epoch": 0.635254423476794,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.7223,
      "step": 2024
    },
    {
      "epoch": 0.6355682843579583,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.6626,
      "step": 2025
    },
    {
      "epoch": 0.6358821452391228,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.4276,
      "step": 2026
    },
    {
      "epoch": 0.6361960061202871,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.7796,
      "step": 2027
    },
    {
      "epoch": 0.6365098670014516,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.6912,
      "step": 2028
    },
    {
      "epoch": 0.636823727882616,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.8183,
      "step": 2029
    },
    {
      "epoch": 0.6371375887637805,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.7979,
      "step": 2030
    },
    {
      "epoch": 0.6374514496449448,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002,
      "loss": 1.9475,
      "step": 2031
    },
    {
      "epoch": 0.6377653105261093,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.9001,
      "step": 2032
    },
    {
      "epoch": 0.6380791714072738,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.8966,
      "step": 2033
    },
    {
      "epoch": 0.6383930322884381,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 2.4315,
      "step": 2034
    },
    {
      "epoch": 0.6387068931696026,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 2.07,
      "step": 2035
    },
    {
      "epoch": 0.639020754050767,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 2.5365,
      "step": 2036
    },
    {
      "epoch": 0.6393346149319314,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.3692,
      "step": 2037
    },
    {
      "epoch": 0.6396484758130958,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.9882,
      "step": 2038
    },
    {
      "epoch": 0.6399623366942603,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.08,
      "step": 2039
    },
    {
      "epoch": 0.6402761975754246,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 2.1133,
      "step": 2040
    },
    {
      "epoch": 0.6405900584565891,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 2.3706,
      "step": 2041
    },
    {
      "epoch": 0.6409039193377536,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 2.2254,
      "step": 2042
    },
    {
      "epoch": 0.641217780218918,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 2.3317,
      "step": 2043
    },
    {
      "epoch": 0.6415316411000824,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 2.5359,
      "step": 2044
    },
    {
      "epoch": 0.6418455019812468,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 2.541,
      "step": 2045
    },
    {
      "epoch": 0.6421593628624113,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 2.2715,
      "step": 2046
    },
    {
      "epoch": 0.6424732237435756,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.9943,
      "step": 2047
    },
    {
      "epoch": 0.6427870846247401,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 2.1426,
      "step": 2048
    },
    {
      "epoch": 0.6431009455059045,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 2.2063,
      "step": 2049
    },
    {
      "epoch": 0.6434148063870689,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.5465,
      "step": 2050
    },
    {
      "epoch": 0.6437286672682333,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.5447,
      "step": 2051
    },
    {
      "epoch": 0.6440425281493978,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.4158,
      "step": 2052
    },
    {
      "epoch": 0.6443563890305622,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.5656,
      "step": 2053
    },
    {
      "epoch": 0.6446702499117266,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.5103,
      "step": 2054
    },
    {
      "epoch": 0.6449841107928911,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3915,
      "step": 2055
    },
    {
      "epoch": 0.6452979716740554,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3889,
      "step": 2056
    },
    {
      "epoch": 0.6456118325552199,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.5592,
      "step": 2057
    },
    {
      "epoch": 0.6456118325552199,
      "eval_loss": 1.7869147062301636,
      "eval_runtime": 123.5793,
      "eval_samples_per_second": 8.092,
      "eval_steps_per_second": 8.092,
      "step": 2057
    },
    {
      "epoch": 0.6456118325552199,
      "mmlu_eval_accuracy": 0.4170033028018425,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.35714285714285715,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.7272727272727273,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.22727272727272727,
      "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": 0.5,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.5333333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.5813953488372093,
      "mmlu_eval_accuracy_moral_disputes": 0.34210526315789475,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.5151515151515151,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3588235294117647,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
      "mmlu_eval_accuracy_virology": 0.5555555555555556,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 0.9896279288167505,
      "step": 2057
    },
    {
      "epoch": 0.6459256934363843,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.5043,
      "step": 2058
    },
    {
      "epoch": 0.6462395543175488,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4262,
      "step": 2059
    },
    {
      "epoch": 0.6465534151987131,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.5209,
      "step": 2060
    },
    {
      "epoch": 0.6468672760798776,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.3202,
      "step": 2061
    },
    {
      "epoch": 0.6471811369610421,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.2827,
      "step": 2062
    },
    {
      "epoch": 0.6474949978422064,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4121,
      "step": 2063
    },
    {
      "epoch": 0.6478088587233709,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.4375,
      "step": 2064
    },
    {
      "epoch": 0.6481227196045353,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2448,
      "step": 2065
    },
    {
      "epoch": 0.6484365804856997,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4004,
      "step": 2066
    },
    {
      "epoch": 0.6487504413668641,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3909,
      "step": 2067
    },
    {
      "epoch": 0.6490643022480286,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.3046,
      "step": 2068
    },
    {
      "epoch": 0.6493781631291929,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3599,
      "step": 2069
    },
    {
      "epoch": 0.6496920240103574,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.3598,
      "step": 2070
    },
    {
      "epoch": 0.6500058848915218,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.8061,
      "step": 2071
    },
    {
      "epoch": 0.6503197457726863,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.3645,
      "step": 2072
    },
    {
      "epoch": 0.6506336066538507,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.607,
      "step": 2073
    },
    {
      "epoch": 0.6509474675350151,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.7126,
      "step": 2074
    },
    {
      "epoch": 0.6512613284161796,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.4475,
      "step": 2075
    },
    {
      "epoch": 0.6515751892973439,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.778,
      "step": 2076
    },
    {
      "epoch": 0.6518890501785084,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.6442,
      "step": 2077
    },
    {
      "epoch": 0.6522029110596728,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.5434,
      "step": 2078
    },
    {
      "epoch": 0.6525167719408372,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.7534,
      "step": 2079
    },
    {
      "epoch": 0.6528306328220016,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.7837,
      "step": 2080
    },
    {
      "epoch": 0.6531444937031661,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.125,
      "step": 2081
    },
    {
      "epoch": 0.6534583545843305,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 2.1823,
      "step": 2082
    },
    {
      "epoch": 0.6537722154654949,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 2.1828,
      "step": 2083
    },
    {
      "epoch": 0.6540860763466594,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.9816,
      "step": 2084
    },
    {
      "epoch": 0.6543999372278237,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.2713,
      "step": 2085
    },
    {
      "epoch": 0.6547137981089882,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.7811,
      "step": 2086
    },
    {
      "epoch": 0.6550276589901526,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 2.0191,
      "step": 2087
    },
    {
      "epoch": 0.6553415198713171,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 2.0255,
      "step": 2088
    },
    {
      "epoch": 0.6556553807524814,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.7514,
      "step": 2089
    },
    {
      "epoch": 0.6559692416336459,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 2.3259,
      "step": 2090
    },
    {
      "epoch": 0.6562831025148103,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.3979,
      "step": 2091
    },
    {
      "epoch": 0.6565969633959747,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.8578,
      "step": 2092
    },
    {
      "epoch": 0.6569108242771392,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.9541,
      "step": 2093
    },
    {
      "epoch": 0.6572246851583036,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.5271,
      "step": 2094
    },
    {
      "epoch": 0.657538546039468,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.8429,
      "step": 2095
    },
    {
      "epoch": 0.6578524069206324,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.8997,
      "step": 2096
    },
    {
      "epoch": 0.6581662678017969,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.293,
      "step": 2097
    },
    {
      "epoch": 0.6584801286829612,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 2.1048,
      "step": 2098
    },
    {
      "epoch": 0.6587939895641257,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 2.4086,
      "step": 2099
    },
    {
      "epoch": 0.6591078504452901,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 3.0144,
      "step": 2100
    },
    {
      "epoch": 0.6594217113264546,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0002,
      "loss": 1.3451,
      "step": 2101
    },
    {
      "epoch": 0.659735572207619,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.347,
      "step": 2102
    },
    {
      "epoch": 0.6600494330887834,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.3807,
      "step": 2103
    },
    {
      "epoch": 0.6603632939699479,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.4438,
      "step": 2104
    },
    {
      "epoch": 0.6606771548511122,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.404,
      "step": 2105
    },
    {
      "epoch": 0.6609910157322767,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4676,
      "step": 2106
    },
    {
      "epoch": 0.6613048766134411,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2827,
      "step": 2107
    },
    {
      "epoch": 0.6616187374946055,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.5957,
      "step": 2108
    },
    {
      "epoch": 0.6619325983757699,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.487,
      "step": 2109
    },
    {
      "epoch": 0.6622464592569344,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4979,
      "step": 2110
    },
    {
      "epoch": 0.6625603201380987,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.4709,
      "step": 2111
    },
    {
      "epoch": 0.6628741810192632,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.5848,
      "step": 2112
    },
    {
      "epoch": 0.6631880419004277,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.5552,
      "step": 2113
    },
    {
      "epoch": 0.663501902781592,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.4407,
      "step": 2114
    },
    {
      "epoch": 0.6638157636627565,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.3967,
      "step": 2115
    },
    {
      "epoch": 0.6641296245439209,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.6367,
      "step": 2116
    },
    {
      "epoch": 0.6644434854250854,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.2283,
      "step": 2117
    },
    {
      "epoch": 0.6647573463062497,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.7342,
      "step": 2118
    },
    {
      "epoch": 0.6650712071874142,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.4274,
      "step": 2119
    },
    {
      "epoch": 0.6653850680685786,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.466,
      "step": 2120
    },
    {
      "epoch": 0.665698928949743,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.4551,
      "step": 2121
    },
    {
      "epoch": 0.6660127898309075,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.6061,
      "step": 2122
    },
    {
      "epoch": 0.6663266507120719,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.4578,
      "step": 2123
    },
    {
      "epoch": 0.6666405115932363,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.7235,
      "step": 2124
    },
    {
      "epoch": 0.6669543724744007,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.6891,
      "step": 2125
    },
    {
      "epoch": 0.6672682333555652,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.894,
      "step": 2126
    },
    {
      "epoch": 0.6675820942367295,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.91,
      "step": 2127
    },
    {
      "epoch": 0.667895955117894,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 2.1388,
      "step": 2128
    },
    {
      "epoch": 0.6682098159990584,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.6798,
      "step": 2129
    },
    {
      "epoch": 0.6685236768802229,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4976,
      "step": 2130
    },
    {
      "epoch": 0.6688375377613872,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.9372,
      "step": 2131
    },
    {
      "epoch": 0.6691513986425517,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 2.1857,
      "step": 2132
    },
    {
      "epoch": 0.6694652595237162,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.9852,
      "step": 2133
    },
    {
      "epoch": 0.6697791204048805,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 2.1061,
      "step": 2134
    },
    {
      "epoch": 0.670092981286045,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.8907,
      "step": 2135
    },
    {
      "epoch": 0.6704068421672094,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 2.0113,
      "step": 2136
    },
    {
      "epoch": 0.6707207030483738,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.0771,
      "step": 2137
    },
    {
      "epoch": 0.6710345639295382,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 2.6552,
      "step": 2138
    },
    {
      "epoch": 0.6713484248107027,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 2.2316,
      "step": 2139
    },
    {
      "epoch": 0.671662285691867,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 2.2031,
      "step": 2140
    },
    {
      "epoch": 0.6719761465730315,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.4282,
      "step": 2141
    },
    {
      "epoch": 0.672290007454196,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 2.2798,
      "step": 2142
    },
    {
      "epoch": 0.6726038683353603,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.9225,
      "step": 2143
    },
    {
      "epoch": 0.6729177292165248,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 2.0801,
      "step": 2144
    },
    {
      "epoch": 0.6732315900976892,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 2.1691,
      "step": 2145
    },
    {
      "epoch": 0.6735454509788537,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 2.5013,
      "step": 2146
    },
    {
      "epoch": 0.673859311860018,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 2.3714,
      "step": 2147
    },
    {
      "epoch": 0.6741731727411825,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 2.117,
      "step": 2148
    },
    {
      "epoch": 0.6744870336223469,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.5755,
      "step": 2149
    },
    {
      "epoch": 0.6748008945035113,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 3.2677,
      "step": 2150
    },
    {
      "epoch": 0.6751147553846757,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.1853,
      "step": 2151
    },
    {
      "epoch": 0.6754286162658402,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0002,
      "loss": 1.3131,
      "step": 2152
    },
    {
      "epoch": 0.6757424771470046,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 1.2502,
      "step": 2153
    },
    {
      "epoch": 0.676056338028169,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.3046,
      "step": 2154
    },
    {
      "epoch": 0.6763701989093335,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.4713,
      "step": 2155
    },
    {
      "epoch": 0.6766840597904978,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.4688,
      "step": 2156
    },
    {
      "epoch": 0.6769979206716623,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3407,
      "step": 2157
    },
    {
      "epoch": 0.6773117815528267,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.426,
      "step": 2158
    },
    {
      "epoch": 0.6776256424339911,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.3701,
      "step": 2159
    },
    {
      "epoch": 0.6779395033151555,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3456,
      "step": 2160
    },
    {
      "epoch": 0.67825336419632,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2171,
      "step": 2161
    },
    {
      "epoch": 0.6785672250774845,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.545,
      "step": 2162
    },
    {
      "epoch": 0.6788810859586488,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3447,
      "step": 2163
    },
    {
      "epoch": 0.6791949468398133,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.4775,
      "step": 2164
    },
    {
      "epoch": 0.6795088077209777,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.4556,
      "step": 2165
    },
    {
      "epoch": 0.6798226686021421,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.5541,
      "step": 2166
    },
    {
      "epoch": 0.6801365294833065,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.4576,
      "step": 2167
    },
    {
      "epoch": 0.680450390364471,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2738,
      "step": 2168
    },
    {
      "epoch": 0.6807642512456353,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.6087,
      "step": 2169
    },
    {
      "epoch": 0.6810781121267998,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.3709,
      "step": 2170
    },
    {
      "epoch": 0.6813919730079642,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.4467,
      "step": 2171
    },
    {
      "epoch": 0.6817058338891286,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.5383,
      "step": 2172
    },
    {
      "epoch": 0.6820196947702931,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.6602,
      "step": 2173
    },
    {
      "epoch": 0.6823335556514575,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.5281,
      "step": 2174
    },
    {
      "epoch": 0.682647416532622,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.5583,
      "step": 2175
    },
    {
      "epoch": 0.6829612774137863,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.7739,
      "step": 2176
    },
    {
      "epoch": 0.6832751382949508,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.5948,
      "step": 2177
    },
    {
      "epoch": 0.6835889991761152,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.7775,
      "step": 2178
    },
    {
      "epoch": 0.6839028600572796,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.4536,
      "step": 2179
    },
    {
      "epoch": 0.684216720938444,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.951,
      "step": 2180
    },
    {
      "epoch": 0.6845305818196085,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.9433,
      "step": 2181
    },
    {
      "epoch": 0.6848444427007729,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 2.1355,
      "step": 2182
    },
    {
      "epoch": 0.6851583035819373,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 2.168,
      "step": 2183
    },
    {
      "epoch": 0.6854721644631018,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 2.1087,
      "step": 2184
    },
    {
      "epoch": 0.6857860253442661,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 2.5491,
      "step": 2185
    },
    {
      "epoch": 0.6860998862254306,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 2.0621,
      "step": 2186
    },
    {
      "epoch": 0.686413747106595,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.376,
      "step": 2187
    },
    {
      "epoch": 0.6867276079877594,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 2.5973,
      "step": 2188
    },
    {
      "epoch": 0.6870414688689238,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.8221,
      "step": 2189
    },
    {
      "epoch": 0.6873553297500883,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.0147,
      "step": 2190
    },
    {
      "epoch": 0.6876691906312526,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 2.1272,
      "step": 2191
    },
    {
      "epoch": 0.6879830515124171,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 2.0319,
      "step": 2192
    },
    {
      "epoch": 0.6882969123935816,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.3523,
      "step": 2193
    },
    {
      "epoch": 0.688610773274746,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.8954,
      "step": 2194
    },
    {
      "epoch": 0.6889246341559104,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.0146,
      "step": 2195
    },
    {
      "epoch": 0.6892384950370748,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.8074,
      "step": 2196
    },
    {
      "epoch": 0.6895523559182393,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.0586,
      "step": 2197
    },
    {
      "epoch": 0.6898662167994036,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.9138,
      "step": 2198
    },
    {
      "epoch": 0.6901800776805681,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 2.2715,
      "step": 2199
    },
    {
      "epoch": 0.6904939385617325,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002,
      "loss": 3.2919,
      "step": 2200
    },
    {
      "epoch": 0.6908077994428969,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0002,
      "loss": 1.353,
      "step": 2201
    },
    {
      "epoch": 0.6911216603240613,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0002,
      "loss": 1.2476,
      "step": 2202
    },
    {
      "epoch": 0.6914355212052258,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.2903,
      "step": 2203
    },
    {
      "epoch": 0.6917493820863903,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.4513,
      "step": 2204
    },
    {
      "epoch": 0.6920632429675546,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3309,
      "step": 2205
    },
    {
      "epoch": 0.6923771038487191,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.5383,
      "step": 2206
    },
    {
      "epoch": 0.6926909647298835,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3742,
      "step": 2207
    },
    {
      "epoch": 0.6930048256110479,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.4576,
      "step": 2208
    },
    {
      "epoch": 0.6933186864922123,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3171,
      "step": 2209
    },
    {
      "epoch": 0.6936325473733768,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3857,
      "step": 2210
    },
    {
      "epoch": 0.6939464082545411,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.4446,
      "step": 2211
    },
    {
      "epoch": 0.6942602691357056,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.471,
      "step": 2212
    },
    {
      "epoch": 0.6945741300168701,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3299,
      "step": 2213
    },
    {
      "epoch": 0.6948879908980344,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.4443,
      "step": 2214
    },
    {
      "epoch": 0.6952018517791989,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.371,
      "step": 2215
    },
    {
      "epoch": 0.6955157126603633,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3477,
      "step": 2216
    },
    {
      "epoch": 0.6958295735415277,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.3967,
      "step": 2217
    },
    {
      "epoch": 0.6961434344226921,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.5423,
      "step": 2218
    },
    {
      "epoch": 0.6964572953038566,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.5446,
      "step": 2219
    },
    {
      "epoch": 0.696771156185021,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.5536,
      "step": 2220
    },
    {
      "epoch": 0.6970850170661854,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3579,
      "step": 2221
    },
    {
      "epoch": 0.6973988779473498,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2884,
      "step": 2222
    },
    {
      "epoch": 0.6977127388285143,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.9048,
      "step": 2223
    },
    {
      "epoch": 0.6980265997096787,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.5609,
      "step": 2224
    },
    {
      "epoch": 0.6983404605908431,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.4689,
      "step": 2225
    },
    {
      "epoch": 0.6986543214720076,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.5317,
      "step": 2226
    },
    {
      "epoch": 0.6989681823531719,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.7305,
      "step": 2227
    },
    {
      "epoch": 0.6992820432343364,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.7872,
      "step": 2228
    },
    {
      "epoch": 0.6995959041155008,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.4927,
      "step": 2229
    },
    {
      "epoch": 0.6999097649966652,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 2.1224,
      "step": 2230
    },
    {
      "epoch": 0.7002236258778296,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.929,
      "step": 2231
    },
    {
      "epoch": 0.7005374867589941,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 2.4381,
      "step": 2232
    },
    {
      "epoch": 0.7008513476401586,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 2.2564,
      "step": 2233
    },
    {
      "epoch": 0.7011652085213229,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 2.7457,
      "step": 2234
    },
    {
      "epoch": 0.7014790694024874,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 2.2323,
      "step": 2235
    },
    {
      "epoch": 0.7017929302836518,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.4091,
      "step": 2236
    },
    {
      "epoch": 0.7021067911648162,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 2.3004,
      "step": 2237
    },
    {
      "epoch": 0.7024206520459806,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 2.1265,
      "step": 2238
    },
    {
      "epoch": 0.7027345129271451,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 2.4745,
      "step": 2239
    },
    {
      "epoch": 0.7030483738083094,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.9056,
      "step": 2240
    },
    {
      "epoch": 0.7033622346894739,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.9232,
      "step": 2241
    },
    {
      "epoch": 0.7036760955706383,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 2.3736,
      "step": 2242
    },
    {
      "epoch": 0.7039899564518027,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.5,
      "step": 2243
    },
    {
      "epoch": 0.7043038173329672,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 2.1852,
      "step": 2244
    },
    {
      "epoch": 0.7043038173329672,
      "eval_loss": 1.7769441604614258,
      "eval_runtime": 123.5225,
      "eval_samples_per_second": 8.096,
      "eval_steps_per_second": 8.096,
      "step": 2244
    },
    {
      "epoch": 0.7043038173329672,
      "mmlu_eval_accuracy": 0.4180599978588911,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.2692307692307692,
      "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": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.6,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.5930232558139535,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.3939393939393939,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.1384575194116513,
      "step": 2244
    },
    {
      "epoch": 0.7046176782141316,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.0939,
      "step": 2245
    },
    {
      "epoch": 0.704931539095296,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.8281,
      "step": 2246
    },
    {
      "epoch": 0.7052453999764604,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 2.0069,
      "step": 2247
    },
    {
      "epoch": 0.7055592608576249,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 2.0889,
      "step": 2248
    },
    {
      "epoch": 0.7058731217387892,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002,
      "loss": 2.6987,
      "step": 2249
    },
    {
      "epoch": 0.7061869826199537,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 2.7403,
      "step": 2250
    },
    {
      "epoch": 0.7065008435011181,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.457,
      "step": 2251
    },
    {
      "epoch": 0.7068147043822826,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 1.3027,
      "step": 2252
    },
    {
      "epoch": 0.707128565263447,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.4681,
      "step": 2253
    },
    {
      "epoch": 0.7074424261446114,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3744,
      "step": 2254
    },
    {
      "epoch": 0.7077562870257759,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.6237,
      "step": 2255
    },
    {
      "epoch": 0.7080701479069402,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.4545,
      "step": 2256
    },
    {
      "epoch": 0.7083840087881047,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4803,
      "step": 2257
    },
    {
      "epoch": 0.7086978696692691,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.4537,
      "step": 2258
    },
    {
      "epoch": 0.7090117305504335,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 1.708,
      "step": 2259
    },
    {
      "epoch": 0.7093255914315979,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.5124,
      "step": 2260
    },
    {
      "epoch": 0.7096394523127624,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.4564,
      "step": 2261
    },
    {
      "epoch": 0.7099533131939267,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3397,
      "step": 2262
    },
    {
      "epoch": 0.7102671740750912,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.3918,
      "step": 2263
    },
    {
      "epoch": 0.7105810349562557,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2112,
      "step": 2264
    },
    {
      "epoch": 0.71089489583742,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2941,
      "step": 2265
    },
    {
      "epoch": 0.7112087567185845,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.5815,
      "step": 2266
    },
    {
      "epoch": 0.7115226175997489,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.5319,
      "step": 2267
    },
    {
      "epoch": 0.7118364784809134,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.4616,
      "step": 2268
    },
    {
      "epoch": 0.7121503393620777,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.6004,
      "step": 2269
    },
    {
      "epoch": 0.7124642002432422,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.3365,
      "step": 2270
    },
    {
      "epoch": 0.7127780611244066,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.4398,
      "step": 2271
    },
    {
      "epoch": 0.713091922005571,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.4959,
      "step": 2272
    },
    {
      "epoch": 0.7134057828867355,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.8303,
      "step": 2273
    },
    {
      "epoch": 0.7137196437678999,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.8074,
      "step": 2274
    },
    {
      "epoch": 0.7140335046490643,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.5667,
      "step": 2275
    },
    {
      "epoch": 0.7143473655302287,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.8409,
      "step": 2276
    },
    {
      "epoch": 0.7146612264113932,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.9792,
      "step": 2277
    },
    {
      "epoch": 0.7149750872925575,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.9356,
      "step": 2278
    },
    {
      "epoch": 0.715288948173722,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.7637,
      "step": 2279
    },
    {
      "epoch": 0.7156028090548864,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.5164,
      "step": 2280
    },
    {
      "epoch": 0.7159166699360509,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.7661,
      "step": 2281
    },
    {
      "epoch": 0.7162305308172152,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 2.1665,
      "step": 2282
    },
    {
      "epoch": 0.7165443916983797,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.0457,
      "step": 2283
    },
    {
      "epoch": 0.7168582525795442,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 2.7448,
      "step": 2284
    },
    {
      "epoch": 0.7171721134607085,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.9213,
      "step": 2285
    },
    {
      "epoch": 0.717485974341873,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 2.2519,
      "step": 2286
    },
    {
      "epoch": 0.7177998352230374,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.6954,
      "step": 2287
    },
    {
      "epoch": 0.7181136961042018,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 2.1847,
      "step": 2288
    },
    {
      "epoch": 0.7184275569853662,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.3177,
      "step": 2289
    },
    {
      "epoch": 0.7187414178665307,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 2.7715,
      "step": 2290
    },
    {
      "epoch": 0.719055278747695,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.1505,
      "step": 2291
    },
    {
      "epoch": 0.7193691396288595,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 2.0164,
      "step": 2292
    },
    {
      "epoch": 0.719683000510024,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 2.3602,
      "step": 2293
    },
    {
      "epoch": 0.7199968613911883,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 2.0161,
      "step": 2294
    },
    {
      "epoch": 0.7203107222723528,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 2.4021,
      "step": 2295
    },
    {
      "epoch": 0.7206245831535172,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 2.5295,
      "step": 2296
    },
    {
      "epoch": 0.7209384440346817,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 2.341,
      "step": 2297
    },
    {
      "epoch": 0.721252304915846,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.7983,
      "step": 2298
    },
    {
      "epoch": 0.7215661657970105,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 2.3877,
      "step": 2299
    },
    {
      "epoch": 0.7218800266781749,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 2.8016,
      "step": 2300
    },
    {
      "epoch": 0.7221938875593393,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0002,
      "loss": 1.2162,
      "step": 2301
    },
    {
      "epoch": 0.7225077484405037,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3024,
      "step": 2302
    },
    {
      "epoch": 0.7228216093216682,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.4018,
      "step": 2303
    },
    {
      "epoch": 0.7231354702028326,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 1.3134,
      "step": 2304
    },
    {
      "epoch": 0.723449331083997,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.431,
      "step": 2305
    },
    {
      "epoch": 0.7237631919651615,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3435,
      "step": 2306
    },
    {
      "epoch": 0.7240770528463258,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.5706,
      "step": 2307
    },
    {
      "epoch": 0.7243909137274903,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.4139,
      "step": 2308
    },
    {
      "epoch": 0.7247047746086547,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3822,
      "step": 2309
    },
    {
      "epoch": 0.7250186354898192,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.5015,
      "step": 2310
    },
    {
      "epoch": 0.7253324963709835,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3385,
      "step": 2311
    },
    {
      "epoch": 0.725646357252148,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.4255,
      "step": 2312
    },
    {
      "epoch": 0.7259602181333125,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3187,
      "step": 2313
    },
    {
      "epoch": 0.7262740790144768,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.2845,
      "step": 2314
    },
    {
      "epoch": 0.7265879398956413,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3354,
      "step": 2315
    },
    {
      "epoch": 0.7269018007768057,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3538,
      "step": 2316
    },
    {
      "epoch": 0.7272156616579701,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.624,
      "step": 2317
    },
    {
      "epoch": 0.7275295225391345,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.4957,
      "step": 2318
    },
    {
      "epoch": 0.727843383420299,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.4016,
      "step": 2319
    },
    {
      "epoch": 0.7281572443014633,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.5854,
      "step": 2320
    },
    {
      "epoch": 0.7284711051826278,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.6723,
      "step": 2321
    },
    {
      "epoch": 0.7287849660637922,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.6414,
      "step": 2322
    },
    {
      "epoch": 0.7290988269449566,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.6206,
      "step": 2323
    },
    {
      "epoch": 0.7294126878261211,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.8653,
      "step": 2324
    },
    {
      "epoch": 0.7297265487072855,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.9866,
      "step": 2325
    },
    {
      "epoch": 0.73004040958845,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.774,
      "step": 2326
    },
    {
      "epoch": 0.7303542704696143,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.601,
      "step": 2327
    },
    {
      "epoch": 0.7306681313507788,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.6226,
      "step": 2328
    },
    {
      "epoch": 0.7309819922319432,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.9242,
      "step": 2329
    },
    {
      "epoch": 0.7312958531131076,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.6306,
      "step": 2330
    },
    {
      "epoch": 0.731609713994272,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.917,
      "step": 2331
    },
    {
      "epoch": 0.7319235748754365,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.7724,
      "step": 2332
    },
    {
      "epoch": 0.7322374357566009,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.172,
      "step": 2333
    },
    {
      "epoch": 0.7325512966377653,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 2.1364,
      "step": 2334
    },
    {
      "epoch": 0.7328651575189298,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 2.1089,
      "step": 2335
    },
    {
      "epoch": 0.7331790184000941,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 2.3062,
      "step": 2336
    },
    {
      "epoch": 0.7334928792812586,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 2.3192,
      "step": 2337
    },
    {
      "epoch": 0.733806740162423,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 2.1375,
      "step": 2338
    },
    {
      "epoch": 0.7341206010435875,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.6586,
      "step": 2339
    },
    {
      "epoch": 0.7344344619247518,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 2.358,
      "step": 2340
    },
    {
      "epoch": 0.7347483228059163,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.8156,
      "step": 2341
    },
    {
      "epoch": 0.7350621836870807,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 2.6493,
      "step": 2342
    },
    {
      "epoch": 0.7353760445682451,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.01,
      "step": 2343
    },
    {
      "epoch": 0.7356899054494096,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.1826,
      "step": 2344
    },
    {
      "epoch": 0.736003766330574,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 2.1244,
      "step": 2345
    },
    {
      "epoch": 0.7363176272117384,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.8929,
      "step": 2346
    },
    {
      "epoch": 0.7366314880929028,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.9275,
      "step": 2347
    },
    {
      "epoch": 0.7369453489740673,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.1545,
      "step": 2348
    },
    {
      "epoch": 0.7372592098552316,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 2.342,
      "step": 2349
    },
    {
      "epoch": 0.7375730707363961,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002,
      "loss": 2.5309,
      "step": 2350
    },
    {
      "epoch": 0.7378869316175605,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0002,
      "loss": 1.2762,
      "step": 2351
    },
    {
      "epoch": 0.738200792498725,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3485,
      "step": 2352
    },
    {
      "epoch": 0.7385146533798894,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3308,
      "step": 2353
    },
    {
      "epoch": 0.7388285142610538,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4661,
      "step": 2354
    },
    {
      "epoch": 0.7391423751422183,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.5322,
      "step": 2355
    },
    {
      "epoch": 0.7394562360233826,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.5151,
      "step": 2356
    },
    {
      "epoch": 0.7397700969045471,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.567,
      "step": 2357
    },
    {
      "epoch": 0.7400839577857115,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.5931,
      "step": 2358
    },
    {
      "epoch": 0.7403978186668759,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4611,
      "step": 2359
    },
    {
      "epoch": 0.7407116795480403,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.3765,
      "step": 2360
    },
    {
      "epoch": 0.7410255404292048,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3375,
      "step": 2361
    },
    {
      "epoch": 0.7413394013103691,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4999,
      "step": 2362
    },
    {
      "epoch": 0.7416532621915336,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.4183,
      "step": 2363
    },
    {
      "epoch": 0.7419671230726981,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3219,
      "step": 2364
    },
    {
      "epoch": 0.7422809839538624,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2917,
      "step": 2365
    },
    {
      "epoch": 0.7425948448350269,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.4742,
      "step": 2366
    },
    {
      "epoch": 0.7429087057161913,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.6195,
      "step": 2367
    },
    {
      "epoch": 0.7432225665973557,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3573,
      "step": 2368
    },
    {
      "epoch": 0.7435364274785201,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.3846,
      "step": 2369
    },
    {
      "epoch": 0.7438502883596846,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.4372,
      "step": 2370
    },
    {
      "epoch": 0.744164149240849,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.4481,
      "step": 2371
    },
    {
      "epoch": 0.7444780101220134,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.5518,
      "step": 2372
    },
    {
      "epoch": 0.7447918710031779,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.2905,
      "step": 2373
    },
    {
      "epoch": 0.7451057318843423,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.5086,
      "step": 2374
    },
    {
      "epoch": 0.7454195927655067,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.8025,
      "step": 2375
    },
    {
      "epoch": 0.7457334536466711,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.5683,
      "step": 2376
    },
    {
      "epoch": 0.7460473145278356,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 2.157,
      "step": 2377
    },
    {
      "epoch": 0.7463611754089999,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.8266,
      "step": 2378
    },
    {
      "epoch": 0.7466750362901644,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.8815,
      "step": 2379
    },
    {
      "epoch": 0.7469888971713288,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.8668,
      "step": 2380
    },
    {
      "epoch": 0.7473027580524932,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.9803,
      "step": 2381
    },
    {
      "epoch": 0.7476166189336576,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.9043,
      "step": 2382
    },
    {
      "epoch": 0.7479304798148221,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.9333,
      "step": 2383
    },
    {
      "epoch": 0.7482443406959866,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.9732,
      "step": 2384
    },
    {
      "epoch": 0.7485582015771509,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.9008,
      "step": 2385
    },
    {
      "epoch": 0.7488720624583154,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 2.2057,
      "step": 2386
    },
    {
      "epoch": 0.7491859233394798,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 2.0298,
      "step": 2387
    },
    {
      "epoch": 0.7494997842206442,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 2.1335,
      "step": 2388
    },
    {
      "epoch": 0.7498136451018086,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 2.3354,
      "step": 2389
    },
    {
      "epoch": 0.7501275059829731,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 2.044,
      "step": 2390
    },
    {
      "epoch": 0.7504413668641374,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 2.3769,
      "step": 2391
    },
    {
      "epoch": 0.7507552277453019,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.2939,
      "step": 2392
    },
    {
      "epoch": 0.7510690886264663,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.0112,
      "step": 2393
    },
    {
      "epoch": 0.7513829495076307,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.8882,
      "step": 2394
    },
    {
      "epoch": 0.7516968103887952,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.8986,
      "step": 2395
    },
    {
      "epoch": 0.7520106712699596,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.9775,
      "step": 2396
    },
    {
      "epoch": 0.752324532151124,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.8405,
      "step": 2397
    },
    {
      "epoch": 0.7526383930322884,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 2.4113,
      "step": 2398
    },
    {
      "epoch": 0.7529522539134529,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 2.4571,
      "step": 2399
    },
    {
      "epoch": 0.7532661147946172,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 3.0139,
      "step": 2400
    },
    {
      "epoch": 0.7535799756757817,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2474,
      "step": 2401
    },
    {
      "epoch": 0.7538938365569461,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.5637,
      "step": 2402
    },
    {
      "epoch": 0.7542076974381106,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4162,
      "step": 2403
    },
    {
      "epoch": 0.754521558319275,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.4733,
      "step": 2404
    },
    {
      "epoch": 0.7548354192004394,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.4128,
      "step": 2405
    },
    {
      "epoch": 0.7551492800816039,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2692,
      "step": 2406
    },
    {
      "epoch": 0.7554631409627682,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3874,
      "step": 2407
    },
    {
      "epoch": 0.7557770018439327,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4692,
      "step": 2408
    },
    {
      "epoch": 0.7560908627250971,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.528,
      "step": 2409
    },
    {
      "epoch": 0.7564047236062615,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.5061,
      "step": 2410
    },
    {
      "epoch": 0.7567185844874259,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.5193,
      "step": 2411
    },
    {
      "epoch": 0.7570324453685904,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.5211,
      "step": 2412
    },
    {
      "epoch": 0.7573463062497547,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4203,
      "step": 2413
    },
    {
      "epoch": 0.7576601671309192,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.6103,
      "step": 2414
    },
    {
      "epoch": 0.7579740280120837,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.5367,
      "step": 2415
    },
    {
      "epoch": 0.758287888893248,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.3625,
      "step": 2416
    },
    {
      "epoch": 0.7586017497744125,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.4703,
      "step": 2417
    },
    {
      "epoch": 0.7589156106555769,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.362,
      "step": 2418
    },
    {
      "epoch": 0.7592294715367414,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.5297,
      "step": 2419
    },
    {
      "epoch": 0.7595433324179057,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.6779,
      "step": 2420
    },
    {
      "epoch": 0.7598571932990702,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.5044,
      "step": 2421
    },
    {
      "epoch": 0.7601710541802346,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.7196,
      "step": 2422
    },
    {
      "epoch": 0.760484915061399,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.6114,
      "step": 2423
    },
    {
      "epoch": 0.7607987759425635,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.4659,
      "step": 2424
    },
    {
      "epoch": 0.7611126368237279,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.7074,
      "step": 2425
    },
    {
      "epoch": 0.7614264977048923,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.4459,
      "step": 2426
    },
    {
      "epoch": 0.7617403585860567,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.8313,
      "step": 2427
    },
    {
      "epoch": 0.7620542194672212,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.8035,
      "step": 2428
    },
    {
      "epoch": 0.7623680803483855,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 2.2824,
      "step": 2429
    },
    {
      "epoch": 0.76268194122955,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.9295,
      "step": 2430
    },
    {
      "epoch": 0.7629958021107144,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 2.6543,
      "step": 2431
    },
    {
      "epoch": 0.7629958021107144,
      "eval_loss": 1.7652301788330078,
      "eval_runtime": 122.9161,
      "eval_samples_per_second": 8.136,
      "eval_steps_per_second": 8.136,
      "step": 2431
    },
    {
      "epoch": 0.7629958021107144,
      "mmlu_eval_accuracy": 0.42014877718164845,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.5833333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.4230769230769231,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.88,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5930232558139535,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.3939393939393939,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_psychology": 0.36231884057971014,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5,
      "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.2213488024173416,
      "step": 2431
    },
    {
      "epoch": 0.7633096629918789,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.2962,
      "step": 2432
    },
    {
      "epoch": 0.7636235238730432,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.0499,
      "step": 2433
    },
    {
      "epoch": 0.7639373847542077,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.2559,
      "step": 2434
    },
    {
      "epoch": 0.7642512456353722,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 2.0198,
      "step": 2435
    },
    {
      "epoch": 0.7645651065165365,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.8019,
      "step": 2436
    },
    {
      "epoch": 0.764878967397701,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.8726,
      "step": 2437
    },
    {
      "epoch": 0.7651928282788654,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 2.0747,
      "step": 2438
    },
    {
      "epoch": 0.7655066891600298,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 2.5202,
      "step": 2439
    },
    {
      "epoch": 0.7658205500411942,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 2.3145,
      "step": 2440
    },
    {
      "epoch": 0.7661344109223587,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.0447,
      "step": 2441
    },
    {
      "epoch": 0.766448271803523,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 2.6057,
      "step": 2442
    },
    {
      "epoch": 0.7667621326846875,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0002,
      "loss": 2.1988,
      "step": 2443
    },
    {
      "epoch": 0.767075993565852,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 2.3199,
      "step": 2444
    },
    {
      "epoch": 0.7673898544470164,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 2.1278,
      "step": 2445
    },
    {
      "epoch": 0.7677037153281808,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 2.4532,
      "step": 2446
    },
    {
      "epoch": 0.7680175762093452,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 2.0989,
      "step": 2447
    },
    {
      "epoch": 0.7683314370905097,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.963,
      "step": 2448
    },
    {
      "epoch": 0.768645297971674,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 2.337,
      "step": 2449
    },
    {
      "epoch": 0.7689591588528385,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 2.7283,
      "step": 2450
    },
    {
      "epoch": 0.7692730197340029,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.3662,
      "step": 2451
    },
    {
      "epoch": 0.7695868806151673,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.3258,
      "step": 2452
    },
    {
      "epoch": 0.7699007414963317,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3579,
      "step": 2453
    },
    {
      "epoch": 0.7702146023774962,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.5144,
      "step": 2454
    },
    {
      "epoch": 0.7705284632586606,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2933,
      "step": 2455
    },
    {
      "epoch": 0.770842324139825,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4465,
      "step": 2456
    },
    {
      "epoch": 0.7711561850209895,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2679,
      "step": 2457
    },
    {
      "epoch": 0.7714700459021538,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.2474,
      "step": 2458
    },
    {
      "epoch": 0.7717839067833183,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.358,
      "step": 2459
    },
    {
      "epoch": 0.7720977676644827,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.2669,
      "step": 2460
    },
    {
      "epoch": 0.7724116285456472,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.5163,
      "step": 2461
    },
    {
      "epoch": 0.7727254894268115,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2793,
      "step": 2462
    },
    {
      "epoch": 0.773039350307976,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.6406,
      "step": 2463
    },
    {
      "epoch": 0.7733532111891405,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.357,
      "step": 2464
    },
    {
      "epoch": 0.7736670720703048,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.5362,
      "step": 2465
    },
    {
      "epoch": 0.7739809329514693,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.4644,
      "step": 2466
    },
    {
      "epoch": 0.7742947938326337,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4558,
      "step": 2467
    },
    {
      "epoch": 0.7746086547137981,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.3939,
      "step": 2468
    },
    {
      "epoch": 0.7749225155949625,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.576,
      "step": 2469
    },
    {
      "epoch": 0.775236376476127,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.6347,
      "step": 2470
    },
    {
      "epoch": 0.7755502373572913,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3226,
      "step": 2471
    },
    {
      "epoch": 0.7758640982384558,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.3418,
      "step": 2472
    },
    {
      "epoch": 0.7761779591196202,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.723,
      "step": 2473
    },
    {
      "epoch": 0.7764918200007846,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.5778,
      "step": 2474
    },
    {
      "epoch": 0.7768056808819491,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.7445,
      "step": 2475
    },
    {
      "epoch": 0.7771195417631135,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.6642,
      "step": 2476
    },
    {
      "epoch": 0.777433402644278,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.7149,
      "step": 2477
    },
    {
      "epoch": 0.7777472635254423,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.974,
      "step": 2478
    },
    {
      "epoch": 0.7780611244066068,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.8387,
      "step": 2479
    },
    {
      "epoch": 0.7783749852877712,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 2.1841,
      "step": 2480
    },
    {
      "epoch": 0.7786888461689356,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.6258,
      "step": 2481
    },
    {
      "epoch": 0.7790027070501,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 2.3414,
      "step": 2482
    },
    {
      "epoch": 0.7793165679312645,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 2.6989,
      "step": 2483
    },
    {
      "epoch": 0.7796304288124289,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.2343,
      "step": 2484
    },
    {
      "epoch": 0.7799442896935933,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 2.5767,
      "step": 2485
    },
    {
      "epoch": 0.7802581505747578,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.1286,
      "step": 2486
    },
    {
      "epoch": 0.7805720114559221,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 2.1996,
      "step": 2487
    },
    {
      "epoch": 0.7808858723370866,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.9517,
      "step": 2488
    },
    {
      "epoch": 0.781199733218251,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.0916,
      "step": 2489
    },
    {
      "epoch": 0.7815135940994155,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 2.1751,
      "step": 2490
    },
    {
      "epoch": 0.7818274549805798,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.9032,
      "step": 2491
    },
    {
      "epoch": 0.7821413158617443,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 2.132,
      "step": 2492
    },
    {
      "epoch": 0.7824551767429087,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 2.3776,
      "step": 2493
    },
    {
      "epoch": 0.7827690376240731,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 2.0207,
      "step": 2494
    },
    {
      "epoch": 0.7830828985052376,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 2.0179,
      "step": 2495
    },
    {
      "epoch": 0.783396759386402,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 2.2601,
      "step": 2496
    },
    {
      "epoch": 0.7837106202675664,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 2.2889,
      "step": 2497
    },
    {
      "epoch": 0.7840244811487308,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 1.9265,
      "step": 2498
    },
    {
      "epoch": 0.7843383420298953,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.3361,
      "step": 2499
    },
    {
      "epoch": 0.7846522029110596,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 2.7706,
      "step": 2500
    },
    {
      "epoch": 0.7849660637922241,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.386,
      "step": 2501
    },
    {
      "epoch": 0.7852799246733885,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.3792,
      "step": 2502
    },
    {
      "epoch": 0.785593785554553,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.3294,
      "step": 2503
    },
    {
      "epoch": 0.7859076464357174,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4478,
      "step": 2504
    },
    {
      "epoch": 0.7862215073168818,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.5299,
      "step": 2505
    },
    {
      "epoch": 0.7865353681980463,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.4589,
      "step": 2506
    },
    {
      "epoch": 0.7868492290792106,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.5182,
      "step": 2507
    },
    {
      "epoch": 0.7871630899603751,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.4176,
      "step": 2508
    },
    {
      "epoch": 0.7874769508415395,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4129,
      "step": 2509
    },
    {
      "epoch": 0.7877908117227039,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.5108,
      "step": 2510
    },
    {
      "epoch": 0.7881046726038683,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.4665,
      "step": 2511
    },
    {
      "epoch": 0.7884185334850328,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.5009,
      "step": 2512
    },
    {
      "epoch": 0.7887323943661971,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.4032,
      "step": 2513
    },
    {
      "epoch": 0.7890462552473616,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3496,
      "step": 2514
    },
    {
      "epoch": 0.7893601161285261,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.4753,
      "step": 2515
    },
    {
      "epoch": 0.7896739770096904,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.5218,
      "step": 2516
    },
    {
      "epoch": 0.7899878378908549,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4646,
      "step": 2517
    },
    {
      "epoch": 0.7903016987720193,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.3674,
      "step": 2518
    },
    {
      "epoch": 0.7906155596531838,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.4775,
      "step": 2519
    },
    {
      "epoch": 0.7909294205343481,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3576,
      "step": 2520
    },
    {
      "epoch": 0.7912432814155126,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.457,
      "step": 2521
    },
    {
      "epoch": 0.791557142296677,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.973,
      "step": 2522
    },
    {
      "epoch": 0.7918710031778414,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.5673,
      "step": 2523
    },
    {
      "epoch": 0.7921848640590059,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.8701,
      "step": 2524
    },
    {
      "epoch": 0.7924987249401703,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.5442,
      "step": 2525
    },
    {
      "epoch": 0.7928125858213347,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.6074,
      "step": 2526
    },
    {
      "epoch": 0.7931264467024991,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.6917,
      "step": 2527
    },
    {
      "epoch": 0.7934403075836636,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.9244,
      "step": 2528
    },
    {
      "epoch": 0.7937541684648279,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.7359,
      "step": 2529
    },
    {
      "epoch": 0.7940680293459924,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 2.3441,
      "step": 2530
    },
    {
      "epoch": 0.7943818902271568,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 2.0712,
      "step": 2531
    },
    {
      "epoch": 0.7946957511083212,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 2.2452,
      "step": 2532
    },
    {
      "epoch": 0.7950096119894856,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 2.1489,
      "step": 2533
    },
    {
      "epoch": 0.7953234728706501,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.1836,
      "step": 2534
    },
    {
      "epoch": 0.7956373337518146,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.7811,
      "step": 2535
    },
    {
      "epoch": 0.7959511946329789,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.8184,
      "step": 2536
    },
    {
      "epoch": 0.7962650555141434,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.8137,
      "step": 2537
    },
    {
      "epoch": 0.7965789163953078,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 2.1316,
      "step": 2538
    },
    {
      "epoch": 0.7968927772764722,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.976,
      "step": 2539
    },
    {
      "epoch": 0.7972066381576366,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.4813,
      "step": 2540
    },
    {
      "epoch": 0.7975204990388011,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 2.05,
      "step": 2541
    },
    {
      "epoch": 0.7978343599199654,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 2.0406,
      "step": 2542
    },
    {
      "epoch": 0.7981482208011299,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.1095,
      "step": 2543
    },
    {
      "epoch": 0.7984620816822944,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 2.3052,
      "step": 2544
    },
    {
      "epoch": 0.7987759425634587,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 2.2199,
      "step": 2545
    },
    {
      "epoch": 0.7990898034446232,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 2.0261,
      "step": 2546
    },
    {
      "epoch": 0.7994036643257876,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.8794,
      "step": 2547
    },
    {
      "epoch": 0.799717525206952,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.9501,
      "step": 2548
    },
    {
      "epoch": 0.8000313860881164,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 2.1907,
      "step": 2549
    },
    {
      "epoch": 0.8003452469692809,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0002,
      "loss": 3.1421,
      "step": 2550
    },
    {
      "epoch": 0.8006591078504453,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 1.3507,
      "step": 2551
    },
    {
      "epoch": 0.8009729687316097,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.426,
      "step": 2552
    },
    {
      "epoch": 0.8012868296127741,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.312,
      "step": 2553
    },
    {
      "epoch": 0.8016006904939386,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.5689,
      "step": 2554
    },
    {
      "epoch": 0.801914551375103,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.3539,
      "step": 2555
    },
    {
      "epoch": 0.8022284122562674,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.282,
      "step": 2556
    },
    {
      "epoch": 0.8025422731374319,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3984,
      "step": 2557
    },
    {
      "epoch": 0.8028561340185962,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.6132,
      "step": 2558
    },
    {
      "epoch": 0.8031699948997607,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.3977,
      "step": 2559
    },
    {
      "epoch": 0.8034838557809251,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4431,
      "step": 2560
    },
    {
      "epoch": 0.8037977166620895,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.4175,
      "step": 2561
    },
    {
      "epoch": 0.8041115775432539,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.5591,
      "step": 2562
    },
    {
      "epoch": 0.8044254384244184,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.4017,
      "step": 2563
    },
    {
      "epoch": 0.8047392993055829,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.5094,
      "step": 2564
    },
    {
      "epoch": 0.8050531601867472,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.4971,
      "step": 2565
    },
    {
      "epoch": 0.8053670210679117,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.4216,
      "step": 2566
    },
    {
      "epoch": 0.805680881949076,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.4232,
      "step": 2567
    },
    {
      "epoch": 0.8059947428302405,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.4799,
      "step": 2568
    },
    {
      "epoch": 0.8063086037114049,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3335,
      "step": 2569
    },
    {
      "epoch": 0.8066224645925694,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.6929,
      "step": 2570
    },
    {
      "epoch": 0.8069363254737337,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.4864,
      "step": 2571
    },
    {
      "epoch": 0.8072501863548982,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.6533,
      "step": 2572
    },
    {
      "epoch": 0.8075640472360626,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.6576,
      "step": 2573
    },
    {
      "epoch": 0.807877908117227,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.8208,
      "step": 2574
    },
    {
      "epoch": 0.8081917689983915,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.7821,
      "step": 2575
    },
    {
      "epoch": 0.8085056298795559,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.7887,
      "step": 2576
    },
    {
      "epoch": 0.8088194907607203,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.761,
      "step": 2577
    },
    {
      "epoch": 0.8091333516418847,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.7492,
      "step": 2578
    },
    {
      "epoch": 0.8094472125230492,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.9048,
      "step": 2579
    },
    {
      "epoch": 0.8097610734042135,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.9207,
      "step": 2580
    },
    {
      "epoch": 0.810074934285378,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 2.1126,
      "step": 2581
    },
    {
      "epoch": 0.8103887951665424,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.9245,
      "step": 2582
    },
    {
      "epoch": 0.8107026560477069,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.1493,
      "step": 2583
    },
    {
      "epoch": 0.8110165169288712,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.2426,
      "step": 2584
    },
    {
      "epoch": 0.8113303778100357,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.4175,
      "step": 2585
    },
    {
      "epoch": 0.8116442386912002,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.9238,
      "step": 2586
    },
    {
      "epoch": 0.8119580995723645,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.9419,
      "step": 2587
    },
    {
      "epoch": 0.812271960453529,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.2005,
      "step": 2588
    },
    {
      "epoch": 0.8125858213346934,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.8211,
      "step": 2589
    },
    {
      "epoch": 0.8128996822158578,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.8112,
      "step": 2590
    },
    {
      "epoch": 0.8132135430970222,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 2.8194,
      "step": 2591
    },
    {
      "epoch": 0.8135274039781867,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.9225,
      "step": 2592
    },
    {
      "epoch": 0.813841264859351,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 2.1797,
      "step": 2593
    },
    {
      "epoch": 0.8141551257405155,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.9405,
      "step": 2594
    },
    {
      "epoch": 0.81446898662168,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.6221,
      "step": 2595
    },
    {
      "epoch": 0.8147828475028444,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.7334,
      "step": 2596
    },
    {
      "epoch": 0.8150967083840088,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.3404,
      "step": 2597
    },
    {
      "epoch": 0.8154105692651732,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.3539,
      "step": 2598
    },
    {
      "epoch": 0.8157244301463377,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 3.0796,
      "step": 2599
    },
    {
      "epoch": 0.816038291027502,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 2.986,
      "step": 2600
    },
    {
      "epoch": 0.8163521519086665,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0002,
      "loss": 1.1332,
      "step": 2601
    },
    {
      "epoch": 0.8166660127898309,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4435,
      "step": 2602
    },
    {
      "epoch": 0.8169798736709953,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.3176,
      "step": 2603
    },
    {
      "epoch": 0.8172937345521597,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.4673,
      "step": 2604
    },
    {
      "epoch": 0.8176075954333242,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.3753,
      "step": 2605
    },
    {
      "epoch": 0.8179214563144886,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.7531,
      "step": 2606
    },
    {
      "epoch": 0.818235317195653,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.5415,
      "step": 2607
    },
    {
      "epoch": 0.8185491780768175,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3644,
      "step": 2608
    },
    {
      "epoch": 0.8188630389579818,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.5165,
      "step": 2609
    },
    {
      "epoch": 0.8191768998391463,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3583,
      "step": 2610
    },
    {
      "epoch": 0.8194907607203107,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2584,
      "step": 2611
    },
    {
      "epoch": 0.8198046216014752,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.2903,
      "step": 2612
    },
    {
      "epoch": 0.8201184824826395,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3581,
      "step": 2613
    },
    {
      "epoch": 0.820432343363804,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2519,
      "step": 2614
    },
    {
      "epoch": 0.8207462042449685,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2875,
      "step": 2615
    },
    {
      "epoch": 0.8210600651261328,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4079,
      "step": 2616
    },
    {
      "epoch": 0.8213739260072973,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.6645,
      "step": 2617
    },
    {
      "epoch": 0.8216877868884617,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.5979,
      "step": 2618
    },
    {
      "epoch": 0.8216877868884617,
      "eval_loss": 1.7699179649353027,
      "eval_runtime": 123.3214,
      "eval_samples_per_second": 8.109,
      "eval_steps_per_second": 8.109,
      "step": 2618
    },
    {
      "epoch": 0.8216877868884617,
      "mmlu_eval_accuracy": 0.4268902783865228,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.2692307692307692,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.3125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.19230769230769232,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.6333333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5697674418604651,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.42424242424242425,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.2396918582534728,
      "step": 2618
    },
    {
      "epoch": 0.8220016477696261,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.4718,
      "step": 2619
    },
    {
      "epoch": 0.8223155086507905,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.6423,
      "step": 2620
    },
    {
      "epoch": 0.822629369531955,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.4218,
      "step": 2621
    },
    {
      "epoch": 0.8229432304131193,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.5827,
      "step": 2622
    },
    {
      "epoch": 0.8232570912942838,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 2.0448,
      "step": 2623
    },
    {
      "epoch": 0.8235709521754482,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.4869,
      "step": 2624
    },
    {
      "epoch": 0.8238848130566127,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.5805,
      "step": 2625
    },
    {
      "epoch": 0.8241986739377771,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 1.8846,
      "step": 2626
    },
    {
      "epoch": 0.8245125348189415,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.9054,
      "step": 2627
    },
    {
      "epoch": 0.824826395700106,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.1049,
      "step": 2628
    },
    {
      "epoch": 0.8251402565812703,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.9548,
      "step": 2629
    },
    {
      "epoch": 0.8254541174624348,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 2.3645,
      "step": 2630
    },
    {
      "epoch": 0.8257679783435992,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 2.0293,
      "step": 2631
    },
    {
      "epoch": 0.8260818392247636,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 2.3448,
      "step": 2632
    },
    {
      "epoch": 0.826395700105928,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 2.1027,
      "step": 2633
    },
    {
      "epoch": 0.8267095609870925,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.8683,
      "step": 2634
    },
    {
      "epoch": 0.827023421868257,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.9694,
      "step": 2635
    },
    {
      "epoch": 0.8273372827494213,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.9533,
      "step": 2636
    },
    {
      "epoch": 0.8276511436305858,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 2.2496,
      "step": 2637
    },
    {
      "epoch": 0.8279650045117501,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 2.3666,
      "step": 2638
    },
    {
      "epoch": 0.8282788653929146,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.9966,
      "step": 2639
    },
    {
      "epoch": 0.828592726274079,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 2.1991,
      "step": 2640
    },
    {
      "epoch": 0.8289065871552435,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.2321,
      "step": 2641
    },
    {
      "epoch": 0.8292204480364078,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.1258,
      "step": 2642
    },
    {
      "epoch": 0.8295343089175723,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.6097,
      "step": 2643
    },
    {
      "epoch": 0.8298481697987367,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 2.277,
      "step": 2644
    },
    {
      "epoch": 0.8301620306799011,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 2.0178,
      "step": 2645
    },
    {
      "epoch": 0.8304758915610656,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 2.0701,
      "step": 2646
    },
    {
      "epoch": 0.83078975244223,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.6007,
      "step": 2647
    },
    {
      "epoch": 0.8311036133233944,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 2.1087,
      "step": 2648
    },
    {
      "epoch": 0.8314174742045588,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.9444,
      "step": 2649
    },
    {
      "epoch": 0.8317313350857233,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.9424,
      "step": 2650
    },
    {
      "epoch": 0.8320451959668876,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0002,
      "loss": 1.3177,
      "step": 2651
    },
    {
      "epoch": 0.8323590568480521,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.2663,
      "step": 2652
    },
    {
      "epoch": 0.8326729177292165,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.423,
      "step": 2653
    },
    {
      "epoch": 0.832986778610381,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3817,
      "step": 2654
    },
    {
      "epoch": 0.8333006394915454,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.3837,
      "step": 2655
    },
    {
      "epoch": 0.8336145003727098,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.471,
      "step": 2656
    },
    {
      "epoch": 0.8339283612538743,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3721,
      "step": 2657
    },
    {
      "epoch": 0.8342422221350386,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.5549,
      "step": 2658
    },
    {
      "epoch": 0.8345560830162031,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.398,
      "step": 2659
    },
    {
      "epoch": 0.8348699438973675,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2919,
      "step": 2660
    },
    {
      "epoch": 0.8351838047785319,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.5027,
      "step": 2661
    },
    {
      "epoch": 0.8354976656596963,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.4469,
      "step": 2662
    },
    {
      "epoch": 0.8358115265408608,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.4467,
      "step": 2663
    },
    {
      "epoch": 0.8361253874220251,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.5144,
      "step": 2664
    },
    {
      "epoch": 0.8364392483031896,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.5085,
      "step": 2665
    },
    {
      "epoch": 0.8367531091843541,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3534,
      "step": 2666
    },
    {
      "epoch": 0.8370669700655184,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3309,
      "step": 2667
    },
    {
      "epoch": 0.8373808309466829,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3582,
      "step": 2668
    },
    {
      "epoch": 0.8376946918278473,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.4316,
      "step": 2669
    },
    {
      "epoch": 0.8380085527090118,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.5175,
      "step": 2670
    },
    {
      "epoch": 0.8383224135901761,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.5308,
      "step": 2671
    },
    {
      "epoch": 0.8386362744713406,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.677,
      "step": 2672
    },
    {
      "epoch": 0.838950135352505,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.466,
      "step": 2673
    },
    {
      "epoch": 0.8392639962336694,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.4851,
      "step": 2674
    },
    {
      "epoch": 0.8395778571148339,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.6542,
      "step": 2675
    },
    {
      "epoch": 0.8398917179959983,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.6008,
      "step": 2676
    },
    {
      "epoch": 0.8402055788771627,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.5995,
      "step": 2677
    },
    {
      "epoch": 0.8405194397583271,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.6109,
      "step": 2678
    },
    {
      "epoch": 0.8408333006394916,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.9894,
      "step": 2679
    },
    {
      "epoch": 0.8411471615206559,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.9258,
      "step": 2680
    },
    {
      "epoch": 0.8414610224018204,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 2.4025,
      "step": 2681
    },
    {
      "epoch": 0.8417748832829848,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 2.4065,
      "step": 2682
    },
    {
      "epoch": 0.8420887441641492,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 2.0412,
      "step": 2683
    },
    {
      "epoch": 0.8424026050453136,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.6364,
      "step": 2684
    },
    {
      "epoch": 0.8427164659264781,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.8805,
      "step": 2685
    },
    {
      "epoch": 0.8430303268076426,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 2.0839,
      "step": 2686
    },
    {
      "epoch": 0.8433441876888069,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 2.1611,
      "step": 2687
    },
    {
      "epoch": 0.8436580485699714,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 2.2773,
      "step": 2688
    },
    {
      "epoch": 0.8439719094511358,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 2.18,
      "step": 2689
    },
    {
      "epoch": 0.8442857703323002,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 2.009,
      "step": 2690
    },
    {
      "epoch": 0.8445996312134646,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 2.1408,
      "step": 2691
    },
    {
      "epoch": 0.8449134920946291,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 2.4323,
      "step": 2692
    },
    {
      "epoch": 0.8452273529757934,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.6591,
      "step": 2693
    },
    {
      "epoch": 0.8455412138569579,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.1148,
      "step": 2694
    },
    {
      "epoch": 0.8458550747381224,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 2.1269,
      "step": 2695
    },
    {
      "epoch": 0.8461689356192867,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 2.169,
      "step": 2696
    },
    {
      "epoch": 0.8464827965004512,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 2.8229,
      "step": 2697
    },
    {
      "epoch": 0.8467966573816156,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.8605,
      "step": 2698
    },
    {
      "epoch": 0.84711051826278,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 2.0483,
      "step": 2699
    },
    {
      "epoch": 0.8474243791439444,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.4855,
      "step": 2700
    },
    {
      "epoch": 0.8477382400251089,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.4351,
      "step": 2701
    },
    {
      "epoch": 0.8480521009062733,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.3687,
      "step": 2702
    },
    {
      "epoch": 0.8483659617874377,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3651,
      "step": 2703
    },
    {
      "epoch": 0.8486798226686021,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.5124,
      "step": 2704
    },
    {
      "epoch": 0.8489936835497666,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.4039,
      "step": 2705
    },
    {
      "epoch": 0.849307544430931,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.5525,
      "step": 2706
    },
    {
      "epoch": 0.8496214053120954,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2542,
      "step": 2707
    },
    {
      "epoch": 0.8499352661932599,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4129,
      "step": 2708
    },
    {
      "epoch": 0.8502491270744242,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.3412,
      "step": 2709
    },
    {
      "epoch": 0.8505629879555887,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.3434,
      "step": 2710
    },
    {
      "epoch": 0.8508768488367531,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2987,
      "step": 2711
    },
    {
      "epoch": 0.8511907097179175,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4644,
      "step": 2712
    },
    {
      "epoch": 0.8515045705990819,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.4776,
      "step": 2713
    },
    {
      "epoch": 0.8518184314802464,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.4467,
      "step": 2714
    },
    {
      "epoch": 0.8521322923614109,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.5172,
      "step": 2715
    },
    {
      "epoch": 0.8524461532425752,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.4333,
      "step": 2716
    },
    {
      "epoch": 0.8527600141237397,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4989,
      "step": 2717
    },
    {
      "epoch": 0.8530738750049041,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.4445,
      "step": 2718
    },
    {
      "epoch": 0.8533877358860685,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.503,
      "step": 2719
    },
    {
      "epoch": 0.8537015967672329,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3651,
      "step": 2720
    },
    {
      "epoch": 0.8540154576483974,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.43,
      "step": 2721
    },
    {
      "epoch": 0.8543293185295617,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.2472,
      "step": 2722
    },
    {
      "epoch": 0.8546431794107262,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4451,
      "step": 2723
    },
    {
      "epoch": 0.8549570402918906,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.5654,
      "step": 2724
    },
    {
      "epoch": 0.855270901173055,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.8145,
      "step": 2725
    },
    {
      "epoch": 0.8555847620542195,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.5999,
      "step": 2726
    },
    {
      "epoch": 0.8558986229353839,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.6916,
      "step": 2727
    },
    {
      "epoch": 0.8562124838165484,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.8034,
      "step": 2728
    },
    {
      "epoch": 0.8565263446977127,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.5139,
      "step": 2729
    },
    {
      "epoch": 0.8568402055788772,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.6104,
      "step": 2730
    },
    {
      "epoch": 0.8571540664600416,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.9277,
      "step": 2731
    },
    {
      "epoch": 0.857467927341206,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.5512,
      "step": 2732
    },
    {
      "epoch": 0.8577817882223704,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 2.2699,
      "step": 2733
    },
    {
      "epoch": 0.8580956491035349,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.7613,
      "step": 2734
    },
    {
      "epoch": 0.8584095099846993,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 2.4218,
      "step": 2735
    },
    {
      "epoch": 0.8587233708658637,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 2.3356,
      "step": 2736
    },
    {
      "epoch": 0.8590372317470282,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 2.4203,
      "step": 2737
    },
    {
      "epoch": 0.8593510926281925,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 2.055,
      "step": 2738
    },
    {
      "epoch": 0.859664953509357,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.8217,
      "step": 2739
    },
    {
      "epoch": 0.8599788143905214,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 2.0541,
      "step": 2740
    },
    {
      "epoch": 0.8602926752716858,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.9936,
      "step": 2741
    },
    {
      "epoch": 0.8606065361528502,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 2.1029,
      "step": 2742
    },
    {
      "epoch": 0.8609203970340147,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 2.2554,
      "step": 2743
    },
    {
      "epoch": 0.861234257915179,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.9239,
      "step": 2744
    },
    {
      "epoch": 0.8615481187963435,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.6727,
      "step": 2745
    },
    {
      "epoch": 0.861861979677508,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.6823,
      "step": 2746
    },
    {
      "epoch": 0.8621758405586724,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 2.1984,
      "step": 2747
    },
    {
      "epoch": 0.8624897014398368,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.7844,
      "step": 2748
    },
    {
      "epoch": 0.8628035623210012,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 2.5199,
      "step": 2749
    },
    {
      "epoch": 0.8631174232021657,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.9072,
      "step": 2750
    },
    {
      "epoch": 0.86343128408333,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 1.4048,
      "step": 2751
    },
    {
      "epoch": 0.8637451449644945,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.4475,
      "step": 2752
    },
    {
      "epoch": 0.8640590058456589,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.3802,
      "step": 2753
    },
    {
      "epoch": 0.8643728667268233,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.4085,
      "step": 2754
    },
    {
      "epoch": 0.8646867276079877,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.4118,
      "step": 2755
    },
    {
      "epoch": 0.8650005884891522,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.3409,
      "step": 2756
    },
    {
      "epoch": 0.8653144493703167,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.442,
      "step": 2757
    },
    {
      "epoch": 0.865628310251481,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.4154,
      "step": 2758
    },
    {
      "epoch": 0.8659421711326455,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.431,
      "step": 2759
    },
    {
      "epoch": 0.8662560320138099,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4166,
      "step": 2760
    },
    {
      "epoch": 0.8665698928949743,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.3694,
      "step": 2761
    },
    {
      "epoch": 0.8668837537761387,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.3844,
      "step": 2762
    },
    {
      "epoch": 0.8671976146573032,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2504,
      "step": 2763
    },
    {
      "epoch": 0.8675114755384675,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.3266,
      "step": 2764
    },
    {
      "epoch": 0.867825336419632,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.5628,
      "step": 2765
    },
    {
      "epoch": 0.8681391973007965,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.5032,
      "step": 2766
    },
    {
      "epoch": 0.8684530581819608,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.5167,
      "step": 2767
    },
    {
      "epoch": 0.8687669190631253,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.6064,
      "step": 2768
    },
    {
      "epoch": 0.8690807799442897,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.5063,
      "step": 2769
    },
    {
      "epoch": 0.8693946408254541,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.5111,
      "step": 2770
    },
    {
      "epoch": 0.8697085017066185,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.5674,
      "step": 2771
    },
    {
      "epoch": 0.870022362587783,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.564,
      "step": 2772
    },
    {
      "epoch": 0.8703362234689473,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.4562,
      "step": 2773
    },
    {
      "epoch": 0.8706500843501118,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.8018,
      "step": 2774
    },
    {
      "epoch": 0.8709639452312762,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.7119,
      "step": 2775
    },
    {
      "epoch": 0.8712778061124407,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.6327,
      "step": 2776
    },
    {
      "epoch": 0.8715916669936051,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.8718,
      "step": 2777
    },
    {
      "epoch": 0.8719055278747695,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.7342,
      "step": 2778
    },
    {
      "epoch": 0.872219388755934,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.9018,
      "step": 2779
    },
    {
      "epoch": 0.8725332496370983,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.6986,
      "step": 2780
    },
    {
      "epoch": 0.8728471105182628,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.9193,
      "step": 2781
    },
    {
      "epoch": 0.8731609713994272,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.5808,
      "step": 2782
    },
    {
      "epoch": 0.8734748322805916,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.88,
      "step": 2783
    },
    {
      "epoch": 0.873788693161756,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.2908,
      "step": 2784
    },
    {
      "epoch": 0.8741025540429205,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.2287,
      "step": 2785
    },
    {
      "epoch": 0.874416414924085,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.5983,
      "step": 2786
    },
    {
      "epoch": 0.8747302758052493,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 2.2024,
      "step": 2787
    },
    {
      "epoch": 0.8750441366864138,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.9481,
      "step": 2788
    },
    {
      "epoch": 0.8753579975675781,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 2.2747,
      "step": 2789
    },
    {
      "epoch": 0.8756718584487426,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.1885,
      "step": 2790
    },
    {
      "epoch": 0.875985719329907,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.2836,
      "step": 2791
    },
    {
      "epoch": 0.8762995802110715,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 2.3704,
      "step": 2792
    },
    {
      "epoch": 0.8766134410922358,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.0737,
      "step": 2793
    },
    {
      "epoch": 0.8769273019734003,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 2.1677,
      "step": 2794
    },
    {
      "epoch": 0.8772411628545647,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.699,
      "step": 2795
    },
    {
      "epoch": 0.8775550237357291,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 2.4544,
      "step": 2796
    },
    {
      "epoch": 0.8778688846168936,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.804,
      "step": 2797
    },
    {
      "epoch": 0.878182745498058,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.8516,
      "step": 2798
    },
    {
      "epoch": 0.8784966063792224,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.4355,
      "step": 2799
    },
    {
      "epoch": 0.8788104672603868,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.7897,
      "step": 2800
    },
    {
      "epoch": 0.8791243281415513,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3403,
      "step": 2801
    },
    {
      "epoch": 0.8794381890227156,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3172,
      "step": 2802
    },
    {
      "epoch": 0.8797520499038801,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.5083,
      "step": 2803
    },
    {
      "epoch": 0.8800659107850445,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.4518,
      "step": 2804
    },
    {
      "epoch": 0.880379771666209,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.4931,
      "step": 2805
    },
    {
      "epoch": 0.880379771666209,
      "eval_loss": 1.7788866758346558,
      "eval_runtime": 123.2568,
      "eval_samples_per_second": 8.113,
      "eval_steps_per_second": 8.113,
      "step": 2805
    },
    {
      "epoch": 0.880379771666209,
      "mmlu_eval_accuracy": 0.4050652279837273,
      "mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.19230769230769232,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.5833333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.4230769230769231,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5930232558139535,
      "mmlu_eval_accuracy_moral_disputes": 0.34210526315789475,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.3939393939393939,
      "mmlu_eval_accuracy_philosophy": 0.4117647058823529,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.0547769741829822,
      "step": 2805
    },
    {
      "epoch": 0.8806936325473734,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4426,
      "step": 2806
    },
    {
      "epoch": 0.8810074934285378,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.4516,
      "step": 2807
    },
    {
      "epoch": 0.8813213543097023,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4083,
      "step": 2808
    },
    {
      "epoch": 0.8816352151908666,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.2974,
      "step": 2809
    },
    {
      "epoch": 0.8819490760720311,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3967,
      "step": 2810
    },
    {
      "epoch": 0.8822629369531955,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.405,
      "step": 2811
    },
    {
      "epoch": 0.8825767978343599,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3899,
      "step": 2812
    },
    {
      "epoch": 0.8828906587155243,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2886,
      "step": 2813
    },
    {
      "epoch": 0.8832045195966888,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.4096,
      "step": 2814
    },
    {
      "epoch": 0.8835183804778531,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.3102,
      "step": 2815
    },
    {
      "epoch": 0.8838322413590176,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2996,
      "step": 2816
    },
    {
      "epoch": 0.8841461022401821,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4241,
      "step": 2817
    },
    {
      "epoch": 0.8844599631213464,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4082,
      "step": 2818
    },
    {
      "epoch": 0.8847738240025109,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.3717,
      "step": 2819
    },
    {
      "epoch": 0.8850876848836753,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.4103,
      "step": 2820
    },
    {
      "epoch": 0.8854015457648398,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.5948,
      "step": 2821
    },
    {
      "epoch": 0.8857154066460041,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.4344,
      "step": 2822
    },
    {
      "epoch": 0.8860292675271686,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.5446,
      "step": 2823
    },
    {
      "epoch": 0.886343128408333,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.6508,
      "step": 2824
    },
    {
      "epoch": 0.8866569892894974,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.4263,
      "step": 2825
    },
    {
      "epoch": 0.8869708501706619,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.5295,
      "step": 2826
    },
    {
      "epoch": 0.8872847110518263,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.4279,
      "step": 2827
    },
    {
      "epoch": 0.8875985719329907,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.6395,
      "step": 2828
    },
    {
      "epoch": 0.8879124328141551,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.6521,
      "step": 2829
    },
    {
      "epoch": 0.8882262936953196,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.7628,
      "step": 2830
    },
    {
      "epoch": 0.8885401545764839,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.9453,
      "step": 2831
    },
    {
      "epoch": 0.8888540154576484,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 2.0855,
      "step": 2832
    },
    {
      "epoch": 0.8891678763388128,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 2.3188,
      "step": 2833
    },
    {
      "epoch": 0.8894817372199773,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.9983,
      "step": 2834
    },
    {
      "epoch": 0.8897955981011416,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 2.6919,
      "step": 2835
    },
    {
      "epoch": 0.8901094589823061,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 2.2802,
      "step": 2836
    },
    {
      "epoch": 0.8904233198634706,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.7236,
      "step": 2837
    },
    {
      "epoch": 0.8907371807446349,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 2.1315,
      "step": 2838
    },
    {
      "epoch": 0.8910510416257994,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 2.3496,
      "step": 2839
    },
    {
      "epoch": 0.8913649025069638,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.5181,
      "step": 2840
    },
    {
      "epoch": 0.8916787633881282,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 2.4695,
      "step": 2841
    },
    {
      "epoch": 0.8919926242692926,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 2.0334,
      "step": 2842
    },
    {
      "epoch": 0.8923064851504571,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.2868,
      "step": 2843
    },
    {
      "epoch": 0.8926203460316214,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.9286,
      "step": 2844
    },
    {
      "epoch": 0.8929342069127859,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.8997,
      "step": 2845
    },
    {
      "epoch": 0.8932480677939504,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.7818,
      "step": 2846
    },
    {
      "epoch": 0.8935619286751147,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 2.3554,
      "step": 2847
    },
    {
      "epoch": 0.8938757895562792,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.9271,
      "step": 2848
    },
    {
      "epoch": 0.8941896504374436,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.0679,
      "step": 2849
    },
    {
      "epoch": 0.8945035113186081,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 2.6269,
      "step": 2850
    },
    {
      "epoch": 0.8948173721997724,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3056,
      "step": 2851
    },
    {
      "epoch": 0.8951312330809369,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2266,
      "step": 2852
    },
    {
      "epoch": 0.8954450939621013,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4075,
      "step": 2853
    },
    {
      "epoch": 0.8957589548432657,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.4304,
      "step": 2854
    },
    {
      "epoch": 0.8960728157244301,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.4618,
      "step": 2855
    },
    {
      "epoch": 0.8963866766055946,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.4074,
      "step": 2856
    },
    {
      "epoch": 0.896700537486759,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.5807,
      "step": 2857
    },
    {
      "epoch": 0.8970143983679234,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.232,
      "step": 2858
    },
    {
      "epoch": 0.8973282592490879,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.4562,
      "step": 2859
    },
    {
      "epoch": 0.8976421201302522,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.393,
      "step": 2860
    },
    {
      "epoch": 0.8979559810114167,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.5262,
      "step": 2861
    },
    {
      "epoch": 0.8982698418925811,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4298,
      "step": 2862
    },
    {
      "epoch": 0.8985837027737456,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4761,
      "step": 2863
    },
    {
      "epoch": 0.8988975636549099,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.4818,
      "step": 2864
    },
    {
      "epoch": 0.8992114245360744,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.5365,
      "step": 2865
    },
    {
      "epoch": 0.8995252854172389,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2675,
      "step": 2866
    },
    {
      "epoch": 0.8998391462984032,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.3837,
      "step": 2867
    },
    {
      "epoch": 0.9001530071795677,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.4809,
      "step": 2868
    },
    {
      "epoch": 0.9004668680607321,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.6493,
      "step": 2869
    },
    {
      "epoch": 0.9007807289418965,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.5541,
      "step": 2870
    },
    {
      "epoch": 0.9010945898230609,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.638,
      "step": 2871
    },
    {
      "epoch": 0.9014084507042254,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.6448,
      "step": 2872
    },
    {
      "epoch": 0.9017223115853897,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.6104,
      "step": 2873
    },
    {
      "epoch": 0.9020361724665542,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.3193,
      "step": 2874
    },
    {
      "epoch": 0.9023500333477186,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.5697,
      "step": 2875
    },
    {
      "epoch": 0.902663894228883,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.5283,
      "step": 2876
    },
    {
      "epoch": 0.9029777551100475,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.7855,
      "step": 2877
    },
    {
      "epoch": 0.9032916159912119,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.8568,
      "step": 2878
    },
    {
      "epoch": 0.9036054768723764,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 2.3182,
      "step": 2879
    },
    {
      "epoch": 0.9039193377535407,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.2042,
      "step": 2880
    },
    {
      "epoch": 0.9042331986347052,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.8882,
      "step": 2881
    },
    {
      "epoch": 0.9045470595158696,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.2443,
      "step": 2882
    },
    {
      "epoch": 0.904860920397034,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.9831,
      "step": 2883
    },
    {
      "epoch": 0.9051747812781984,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.7611,
      "step": 2884
    },
    {
      "epoch": 0.9054886421593629,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 2.2978,
      "step": 2885
    },
    {
      "epoch": 0.9058025030405273,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 2.4272,
      "step": 2886
    },
    {
      "epoch": 0.9061163639216917,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 2.3842,
      "step": 2887
    },
    {
      "epoch": 0.9064302248028562,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.2893,
      "step": 2888
    },
    {
      "epoch": 0.9067440856840205,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.8636,
      "step": 2889
    },
    {
      "epoch": 0.907057946565185,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 2.3717,
      "step": 2890
    },
    {
      "epoch": 0.9073718074463494,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 2.5191,
      "step": 2891
    },
    {
      "epoch": 0.9076856683275139,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.3966,
      "step": 2892
    },
    {
      "epoch": 0.9079995292086782,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 2.0948,
      "step": 2893
    },
    {
      "epoch": 0.9083133900898427,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.9643,
      "step": 2894
    },
    {
      "epoch": 0.908627250971007,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.7443,
      "step": 2895
    },
    {
      "epoch": 0.9089411118521715,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 2.1531,
      "step": 2896
    },
    {
      "epoch": 0.909254972733336,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 2.0995,
      "step": 2897
    },
    {
      "epoch": 0.9095688336145004,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 2.1549,
      "step": 2898
    },
    {
      "epoch": 0.9098826944956648,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.6941,
      "step": 2899
    },
    {
      "epoch": 0.9101965553768292,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 2.8947,
      "step": 2900
    },
    {
      "epoch": 0.9105104162579937,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0002,
      "loss": 1.2146,
      "step": 2901
    },
    {
      "epoch": 0.910824277139158,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.3977,
      "step": 2902
    },
    {
      "epoch": 0.9111381380203225,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.3151,
      "step": 2903
    },
    {
      "epoch": 0.9114519989014869,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.5023,
      "step": 2904
    },
    {
      "epoch": 0.9117658597826513,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2814,
      "step": 2905
    },
    {
      "epoch": 0.9120797206638158,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.487,
      "step": 2906
    },
    {
      "epoch": 0.9123935815449802,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.4147,
      "step": 2907
    },
    {
      "epoch": 0.9127074424261447,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.6928,
      "step": 2908
    },
    {
      "epoch": 0.913021303307309,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4939,
      "step": 2909
    },
    {
      "epoch": 0.9133351641884735,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.4947,
      "step": 2910
    },
    {
      "epoch": 0.9136490250696379,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.444,
      "step": 2911
    },
    {
      "epoch": 0.9139628859508023,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.4336,
      "step": 2912
    },
    {
      "epoch": 0.9142767468319667,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3835,
      "step": 2913
    },
    {
      "epoch": 0.9145906077131312,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.4232,
      "step": 2914
    },
    {
      "epoch": 0.9149044685942955,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4444,
      "step": 2915
    },
    {
      "epoch": 0.91521832947546,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.5787,
      "step": 2916
    },
    {
      "epoch": 0.9155321903566245,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2689,
      "step": 2917
    },
    {
      "epoch": 0.9158460512377888,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3692,
      "step": 2918
    },
    {
      "epoch": 0.9161599121189533,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.5798,
      "step": 2919
    },
    {
      "epoch": 0.9164737730001177,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.6355,
      "step": 2920
    },
    {
      "epoch": 0.9167876338812821,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.5006,
      "step": 2921
    },
    {
      "epoch": 0.9171014947624465,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.448,
      "step": 2922
    },
    {
      "epoch": 0.917415355643611,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.6705,
      "step": 2923
    },
    {
      "epoch": 0.9177292165247753,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.6719,
      "step": 2924
    },
    {
      "epoch": 0.9180430774059398,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.5722,
      "step": 2925
    },
    {
      "epoch": 0.9183569382871043,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 1.9757,
      "step": 2926
    },
    {
      "epoch": 0.9186707991682687,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.6939,
      "step": 2927
    },
    {
      "epoch": 0.9189846600494331,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.8072,
      "step": 2928
    },
    {
      "epoch": 0.9192985209305975,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.9631,
      "step": 2929
    },
    {
      "epoch": 0.919612381811762,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 2.1459,
      "step": 2930
    },
    {
      "epoch": 0.9199262426929263,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.8304,
      "step": 2931
    },
    {
      "epoch": 0.9202401035740908,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 2.2797,
      "step": 2932
    },
    {
      "epoch": 0.9205539644552552,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 2.0257,
      "step": 2933
    },
    {
      "epoch": 0.9208678253364196,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 2.0304,
      "step": 2934
    },
    {
      "epoch": 0.921181686217584,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.7163,
      "step": 2935
    },
    {
      "epoch": 0.9214955470987485,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 2.1801,
      "step": 2936
    },
    {
      "epoch": 0.921809407979913,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.1863,
      "step": 2937
    },
    {
      "epoch": 0.9221232688610773,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.7344,
      "step": 2938
    },
    {
      "epoch": 0.9224371297422418,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 2.1153,
      "step": 2939
    },
    {
      "epoch": 0.9227509906234062,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 2.1959,
      "step": 2940
    },
    {
      "epoch": 0.9230648515045706,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 2.5984,
      "step": 2941
    },
    {
      "epoch": 0.923378712385735,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 2.3594,
      "step": 2942
    },
    {
      "epoch": 0.9236925732668995,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 2.1415,
      "step": 2943
    },
    {
      "epoch": 0.9240064341480638,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 2.2361,
      "step": 2944
    },
    {
      "epoch": 0.9243202950292283,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.9417,
      "step": 2945
    },
    {
      "epoch": 0.9246341559103927,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.7594,
      "step": 2946
    },
    {
      "epoch": 0.9249480167915571,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.1687,
      "step": 2947
    },
    {
      "epoch": 0.9252618776727216,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 2.1085,
      "step": 2948
    },
    {
      "epoch": 0.925575738553886,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.9088,
      "step": 2949
    },
    {
      "epoch": 0.9258895994350504,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 2.7049,
      "step": 2950
    },
    {
      "epoch": 0.9262034603162148,
      "grad_norm": 0.0732421875,
      "learning_rate": 0.0002,
      "loss": 1.3151,
      "step": 2951
    },
    {
      "epoch": 0.9265173211973793,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0002,
      "loss": 1.2594,
      "step": 2952
    },
    {
      "epoch": 0.9268311820785436,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3481,
      "step": 2953
    },
    {
      "epoch": 0.9271450429597081,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3905,
      "step": 2954
    },
    {
      "epoch": 0.9274589038408725,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.6019,
      "step": 2955
    },
    {
      "epoch": 0.927772764722037,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.452,
      "step": 2956
    },
    {
      "epoch": 0.9280866256032014,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4319,
      "step": 2957
    },
    {
      "epoch": 0.9284004864843658,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.509,
      "step": 2958
    },
    {
      "epoch": 0.9287143473655303,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.3817,
      "step": 2959
    },
    {
      "epoch": 0.9290282082466946,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.476,
      "step": 2960
    },
    {
      "epoch": 0.9293420691278591,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.5592,
      "step": 2961
    },
    {
      "epoch": 0.9296559300090235,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.4542,
      "step": 2962
    },
    {
      "epoch": 0.9299697908901879,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.5691,
      "step": 2963
    },
    {
      "epoch": 0.9302836517713523,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.4798,
      "step": 2964
    },
    {
      "epoch": 0.9305975126525168,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.436,
      "step": 2965
    },
    {
      "epoch": 0.9309113735336811,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.4208,
      "step": 2966
    },
    {
      "epoch": 0.9312252344148456,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3743,
      "step": 2967
    },
    {
      "epoch": 0.9315390952960101,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3202,
      "step": 2968
    },
    {
      "epoch": 0.9318529561771745,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.3094,
      "step": 2969
    },
    {
      "epoch": 0.9321668170583389,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.4133,
      "step": 2970
    },
    {
      "epoch": 0.9324806779395033,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.4211,
      "step": 2971
    },
    {
      "epoch": 0.9327945388206678,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.3897,
      "step": 2972
    },
    {
      "epoch": 0.9331083997018321,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.5081,
      "step": 2973
    },
    {
      "epoch": 0.9334222605829966,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.4796,
      "step": 2974
    },
    {
      "epoch": 0.933736121464161,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.4619,
      "step": 2975
    },
    {
      "epoch": 0.9340499823453254,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.7486,
      "step": 2976
    },
    {
      "epoch": 0.9343638432264899,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.5938,
      "step": 2977
    },
    {
      "epoch": 0.9346777041076543,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.7117,
      "step": 2978
    },
    {
      "epoch": 0.9349915649888187,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 2.0723,
      "step": 2979
    },
    {
      "epoch": 0.9353054258699831,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.6665,
      "step": 2980
    },
    {
      "epoch": 0.9356192867511476,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.8341,
      "step": 2981
    },
    {
      "epoch": 0.935933147632312,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.6164,
      "step": 2982
    },
    {
      "epoch": 0.9362470085134764,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.1117,
      "step": 2983
    },
    {
      "epoch": 0.9365608693946408,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 2.2635,
      "step": 2984
    },
    {
      "epoch": 0.9368747302758053,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.1706,
      "step": 2985
    },
    {
      "epoch": 0.9371885911569696,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.7411,
      "step": 2986
    },
    {
      "epoch": 0.9375024520381341,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 2.133,
      "step": 2987
    },
    {
      "epoch": 0.9378163129192986,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 2.2699,
      "step": 2988
    },
    {
      "epoch": 0.9381301738004629,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 2.3278,
      "step": 2989
    },
    {
      "epoch": 0.9384440346816274,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.8978,
      "step": 2990
    },
    {
      "epoch": 0.9387578955627918,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.4348,
      "step": 2991
    },
    {
      "epoch": 0.9390717564439562,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 2.2424,
      "step": 2992
    },
    {
      "epoch": 0.9390717564439562,
      "eval_loss": 1.764998435974121,
      "eval_runtime": 123.5438,
      "eval_samples_per_second": 8.094,
      "eval_steps_per_second": 8.094,
      "step": 2992
    },
    {
      "epoch": 0.9390717564439562,
      "mmlu_eval_accuracy": 0.4093842500541707,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.22727272727272727,
      "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": 0.5,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.6,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.18181818181818182,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5930232558139535,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.42424242424242425,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.3333333333333333,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.226706854507551,
      "step": 2992
    },
    {
      "epoch": 0.9393856173251206,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.8017,
      "step": 2993
    },
    {
      "epoch": 0.9396994782062851,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 2.5615,
      "step": 2994
    },
    {
      "epoch": 0.9400133390874494,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 2.2336,
      "step": 2995
    },
    {
      "epoch": 0.9403271999686139,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 2.3521,
      "step": 2996
    },
    {
      "epoch": 0.9406410608497784,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.755,
      "step": 2997
    },
    {
      "epoch": 0.9409549217309428,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.2811,
      "step": 2998
    },
    {
      "epoch": 0.9412687826121072,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 2.4054,
      "step": 2999
    },
    {
      "epoch": 0.9415826434932716,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002,
      "loss": 3.1927,
      "step": 3000
    },
    {
      "epoch": 0.9418965043744361,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0002,
      "loss": 1.4615,
      "step": 3001
    },
    {
      "epoch": 0.9422103652556004,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.4297,
      "step": 3002
    },
    {
      "epoch": 0.9425242261367649,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0002,
      "loss": 1.3411,
      "step": 3003
    },
    {
      "epoch": 0.9428380870179293,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 1.3809,
      "step": 3004
    },
    {
      "epoch": 0.9431519478990937,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.5347,
      "step": 3005
    },
    {
      "epoch": 0.9434658087802581,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.256,
      "step": 3006
    },
    {
      "epoch": 0.9437796696614226,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.4369,
      "step": 3007
    },
    {
      "epoch": 0.944093530542587,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3979,
      "step": 3008
    },
    {
      "epoch": 0.9444073914237514,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3387,
      "step": 3009
    },
    {
      "epoch": 0.9447212523049159,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.5077,
      "step": 3010
    },
    {
      "epoch": 0.9450351131860802,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.442,
      "step": 3011
    },
    {
      "epoch": 0.9453489740672447,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.5246,
      "step": 3012
    },
    {
      "epoch": 0.9456628349484091,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3485,
      "step": 3013
    },
    {
      "epoch": 0.9459766958295736,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4025,
      "step": 3014
    },
    {
      "epoch": 0.9462905567107379,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.4433,
      "step": 3015
    },
    {
      "epoch": 0.9466044175919024,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3678,
      "step": 3016
    },
    {
      "epoch": 0.9469182784730669,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.6643,
      "step": 3017
    },
    {
      "epoch": 0.9472321393542312,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4615,
      "step": 3018
    },
    {
      "epoch": 0.9475460002353957,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.5222,
      "step": 3019
    },
    {
      "epoch": 0.9478598611165601,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.4337,
      "step": 3020
    },
    {
      "epoch": 0.9481737219977245,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.5281,
      "step": 3021
    },
    {
      "epoch": 0.9484875828788889,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.4598,
      "step": 3022
    },
    {
      "epoch": 0.9488014437600534,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.6235,
      "step": 3023
    },
    {
      "epoch": 0.9491153046412177,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.5128,
      "step": 3024
    },
    {
      "epoch": 0.9494291655223822,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.8316,
      "step": 3025
    },
    {
      "epoch": 0.9497430264035466,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.6191,
      "step": 3026
    },
    {
      "epoch": 0.950056887284711,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.7582,
      "step": 3027
    },
    {
      "epoch": 0.9503707481658755,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.7131,
      "step": 3028
    },
    {
      "epoch": 0.9506846090470399,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.5641,
      "step": 3029
    },
    {
      "epoch": 0.9509984699282044,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.9696,
      "step": 3030
    },
    {
      "epoch": 0.9513123308093687,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 2.2189,
      "step": 3031
    },
    {
      "epoch": 0.9516261916905332,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.7686,
      "step": 3032
    },
    {
      "epoch": 0.9519400525716976,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.8801,
      "step": 3033
    },
    {
      "epoch": 0.952253913452862,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.7235,
      "step": 3034
    },
    {
      "epoch": 0.9525677743340264,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.9111,
      "step": 3035
    },
    {
      "epoch": 0.9528816352151909,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 2.1576,
      "step": 3036
    },
    {
      "epoch": 0.9531954960963553,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 2.4213,
      "step": 3037
    },
    {
      "epoch": 0.9535093569775197,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 2.5938,
      "step": 3038
    },
    {
      "epoch": 0.9538232178586842,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 2.451,
      "step": 3039
    },
    {
      "epoch": 0.9541370787398485,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.981,
      "step": 3040
    },
    {
      "epoch": 0.954450939621013,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.1268,
      "step": 3041
    },
    {
      "epoch": 0.9547648005021774,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.7128,
      "step": 3042
    },
    {
      "epoch": 0.9550786613833419,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.953,
      "step": 3043
    },
    {
      "epoch": 0.9553925222645062,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 2.6608,
      "step": 3044
    },
    {
      "epoch": 0.9557063831456707,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 2.1209,
      "step": 3045
    },
    {
      "epoch": 0.956020244026835,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 2.1233,
      "step": 3046
    },
    {
      "epoch": 0.9563341049079995,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.8969,
      "step": 3047
    },
    {
      "epoch": 0.956647965789164,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.8317,
      "step": 3048
    },
    {
      "epoch": 0.9569618266703284,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.983,
      "step": 3049
    },
    {
      "epoch": 0.9572756875514928,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.5904,
      "step": 3050
    },
    {
      "epoch": 0.9575895484326572,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.3929,
      "step": 3051
    },
    {
      "epoch": 0.9579034093138217,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3647,
      "step": 3052
    },
    {
      "epoch": 0.958217270194986,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.3026,
      "step": 3053
    },
    {
      "epoch": 0.9585311310761505,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.513,
      "step": 3054
    },
    {
      "epoch": 0.9588449919573149,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.5286,
      "step": 3055
    },
    {
      "epoch": 0.9591588528384793,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.427,
      "step": 3056
    },
    {
      "epoch": 0.9594727137196438,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4327,
      "step": 3057
    },
    {
      "epoch": 0.9597865746008082,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.5289,
      "step": 3058
    },
    {
      "epoch": 0.9601004354819727,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.5179,
      "step": 3059
    },
    {
      "epoch": 0.960414296363137,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3738,
      "step": 3060
    },
    {
      "epoch": 0.9607281572443015,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4025,
      "step": 3061
    },
    {
      "epoch": 0.9610420181254659,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.5099,
      "step": 3062
    },
    {
      "epoch": 0.9613558790066303,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4911,
      "step": 3063
    },
    {
      "epoch": 0.9616697398877947,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.5292,
      "step": 3064
    },
    {
      "epoch": 0.9619836007689592,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.347,
      "step": 3065
    },
    {
      "epoch": 0.9622974616501235,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.5005,
      "step": 3066
    },
    {
      "epoch": 0.962611322531288,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.3431,
      "step": 3067
    },
    {
      "epoch": 0.9629251834124525,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.4965,
      "step": 3068
    },
    {
      "epoch": 0.9632390442936168,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.5808,
      "step": 3069
    },
    {
      "epoch": 0.9635529051747813,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.7133,
      "step": 3070
    },
    {
      "epoch": 0.9638667660559457,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.2565,
      "step": 3071
    },
    {
      "epoch": 0.9641806269371102,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.4294,
      "step": 3072
    },
    {
      "epoch": 0.9644944878182745,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.5612,
      "step": 3073
    },
    {
      "epoch": 0.964808348699439,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.6573,
      "step": 3074
    },
    {
      "epoch": 0.9651222095806034,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.8696,
      "step": 3075
    },
    {
      "epoch": 0.9654360704617678,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.6256,
      "step": 3076
    },
    {
      "epoch": 0.9657499313429323,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.7176,
      "step": 3077
    },
    {
      "epoch": 0.9660637922240967,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.5864,
      "step": 3078
    },
    {
      "epoch": 0.9663776531052611,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4329,
      "step": 3079
    },
    {
      "epoch": 0.9666915139864255,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 2.2613,
      "step": 3080
    },
    {
      "epoch": 0.96700537486759,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 2.0,
      "step": 3081
    },
    {
      "epoch": 0.9673192357487543,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.8663,
      "step": 3082
    },
    {
      "epoch": 0.9676330966299188,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 2.303,
      "step": 3083
    },
    {
      "epoch": 0.9679469575110832,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.1145,
      "step": 3084
    },
    {
      "epoch": 0.9682608183922476,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.9761,
      "step": 3085
    },
    {
      "epoch": 0.968574679273412,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 2.0674,
      "step": 3086
    },
    {
      "epoch": 0.9688885401545765,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 2.1535,
      "step": 3087
    },
    {
      "epoch": 0.969202401035741,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 2.2142,
      "step": 3088
    },
    {
      "epoch": 0.9695162619169053,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 2.1012,
      "step": 3089
    },
    {
      "epoch": 0.9698301227980698,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.2116,
      "step": 3090
    },
    {
      "epoch": 0.9701439836792342,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.1366,
      "step": 3091
    },
    {
      "epoch": 0.9704578445603986,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 2.5627,
      "step": 3092
    },
    {
      "epoch": 0.970771705441563,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.1147,
      "step": 3093
    },
    {
      "epoch": 0.9710855663227275,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.7821,
      "step": 3094
    },
    {
      "epoch": 0.9713994272038918,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 2.6043,
      "step": 3095
    },
    {
      "epoch": 0.9717132880850563,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 2.2847,
      "step": 3096
    },
    {
      "epoch": 0.9720271489662208,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 2.2498,
      "step": 3097
    },
    {
      "epoch": 0.9723410098473851,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 2.1608,
      "step": 3098
    },
    {
      "epoch": 0.9726548707285496,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 2.4847,
      "step": 3099
    },
    {
      "epoch": 0.972968731609714,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 2.3834,
      "step": 3100
    },
    {
      "epoch": 0.9732825924908785,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0002,
      "loss": 1.3037,
      "step": 3101
    },
    {
      "epoch": 0.9735964533720428,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.4379,
      "step": 3102
    },
    {
      "epoch": 0.9739103142532073,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.3402,
      "step": 3103
    },
    {
      "epoch": 0.9742241751343717,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3735,
      "step": 3104
    },
    {
      "epoch": 0.9745380360155361,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.376,
      "step": 3105
    },
    {
      "epoch": 0.9748518968967005,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.5283,
      "step": 3106
    },
    {
      "epoch": 0.975165757777865,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.5694,
      "step": 3107
    },
    {
      "epoch": 0.9754796186590294,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.334,
      "step": 3108
    },
    {
      "epoch": 0.9757934795401938,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.528,
      "step": 3109
    },
    {
      "epoch": 0.9761073404213583,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.52,
      "step": 3110
    },
    {
      "epoch": 0.9764212013025226,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4252,
      "step": 3111
    },
    {
      "epoch": 0.9767350621836871,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.3186,
      "step": 3112
    },
    {
      "epoch": 0.9770489230648515,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.6066,
      "step": 3113
    },
    {
      "epoch": 0.977362783946016,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.544,
      "step": 3114
    },
    {
      "epoch": 0.9776766448271803,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.4584,
      "step": 3115
    },
    {
      "epoch": 0.9779905057083448,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3325,
      "step": 3116
    },
    {
      "epoch": 0.9783043665895093,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.3468,
      "step": 3117
    },
    {
      "epoch": 0.9786182274706736,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.4411,
      "step": 3118
    },
    {
      "epoch": 0.9789320883518381,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.2572,
      "step": 3119
    },
    {
      "epoch": 0.9792459492330025,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.4013,
      "step": 3120
    },
    {
      "epoch": 0.9795598101141669,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.4012,
      "step": 3121
    },
    {
      "epoch": 0.9798736709953313,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.4045,
      "step": 3122
    },
    {
      "epoch": 0.9801875318764958,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.1227,
      "step": 3123
    },
    {
      "epoch": 0.9805013927576601,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.6554,
      "step": 3124
    },
    {
      "epoch": 0.9808152536388246,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.5291,
      "step": 3125
    },
    {
      "epoch": 0.981129114519989,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.8137,
      "step": 3126
    },
    {
      "epoch": 0.9814429754011534,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.5328,
      "step": 3127
    },
    {
      "epoch": 0.9817568362823179,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.8186,
      "step": 3128
    },
    {
      "epoch": 0.9820706971634823,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 2.1706,
      "step": 3129
    },
    {
      "epoch": 0.9823845580446467,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.6254,
      "step": 3130
    },
    {
      "epoch": 0.9826984189258111,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 2.0098,
      "step": 3131
    },
    {
      "epoch": 0.9830122798069756,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.9764,
      "step": 3132
    },
    {
      "epoch": 0.98332614068814,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 2.1314,
      "step": 3133
    },
    {
      "epoch": 0.9836400015693044,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 2.2194,
      "step": 3134
    },
    {
      "epoch": 0.9839538624504688,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 2.5701,
      "step": 3135
    },
    {
      "epoch": 0.9842677233316333,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.304,
      "step": 3136
    },
    {
      "epoch": 0.9845815842127976,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 2.1723,
      "step": 3137
    },
    {
      "epoch": 0.9848954450939621,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 2.8468,
      "step": 3138
    },
    {
      "epoch": 0.9852093059751266,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 2.6077,
      "step": 3139
    },
    {
      "epoch": 0.9855231668562909,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.6458,
      "step": 3140
    },
    {
      "epoch": 0.9858370277374554,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.2956,
      "step": 3141
    },
    {
      "epoch": 0.9861508886186198,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 2.2138,
      "step": 3142
    },
    {
      "epoch": 0.9864647494997842,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 2.1963,
      "step": 3143
    },
    {
      "epoch": 0.9867786103809486,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.9578,
      "step": 3144
    },
    {
      "epoch": 0.9870924712621131,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 2.2266,
      "step": 3145
    },
    {
      "epoch": 0.9874063321432774,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.6495,
      "step": 3146
    },
    {
      "epoch": 0.9877201930244419,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 2.0621,
      "step": 3147
    },
    {
      "epoch": 0.9880340539056064,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.6402,
      "step": 3148
    },
    {
      "epoch": 0.9883479147867708,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002,
      "loss": 2.4478,
      "step": 3149
    },
    {
      "epoch": 0.9886617756679352,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 2.7114,
      "step": 3150
    },
    {
      "epoch": 0.9889756365490996,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0002,
      "loss": 1.3267,
      "step": 3151
    },
    {
      "epoch": 0.9892894974302641,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.333,
      "step": 3152
    },
    {
      "epoch": 0.9896033583114284,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.359,
      "step": 3153
    },
    {
      "epoch": 0.9899172191925929,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2814,
      "step": 3154
    },
    {
      "epoch": 0.9902310800737573,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3078,
      "step": 3155
    },
    {
      "epoch": 0.9905449409549217,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.5243,
      "step": 3156
    },
    {
      "epoch": 0.9908588018360861,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.4939,
      "step": 3157
    },
    {
      "epoch": 0.9911726627172506,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.5222,
      "step": 3158
    },
    {
      "epoch": 0.991486523598415,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.5443,
      "step": 3159
    },
    {
      "epoch": 0.9918003844795794,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.4496,
      "step": 3160
    },
    {
      "epoch": 0.9921142453607439,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4076,
      "step": 3161
    },
    {
      "epoch": 0.9924281062419082,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.387,
      "step": 3162
    },
    {
      "epoch": 0.9927419671230727,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.4078,
      "step": 3163
    },
    {
      "epoch": 0.9930558280042371,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.4414,
      "step": 3164
    },
    {
      "epoch": 0.9933696888854016,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.526,
      "step": 3165
    },
    {
      "epoch": 0.9936835497665659,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.6098,
      "step": 3166
    },
    {
      "epoch": 0.9939974106477304,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.6074,
      "step": 3167
    },
    {
      "epoch": 0.9943112715288949,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.4543,
      "step": 3168
    },
    {
      "epoch": 0.9946251324100592,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.5867,
      "step": 3169
    },
    {
      "epoch": 0.9949389932912237,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.5843,
      "step": 3170
    },
    {
      "epoch": 0.9952528541723881,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 2.0918,
      "step": 3171
    },
    {
      "epoch": 0.9955667150535525,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 2.0152,
      "step": 3172
    },
    {
      "epoch": 0.9958805759347169,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.5978,
      "step": 3173
    },
    {
      "epoch": 0.9961944368158814,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 2.2455,
      "step": 3174
    },
    {
      "epoch": 0.9965082976970457,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 2.0231,
      "step": 3175
    },
    {
      "epoch": 0.9968221585782102,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.9791,
      "step": 3176
    },
    {
      "epoch": 0.9971360194593746,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.3723,
      "step": 3177
    },
    {
      "epoch": 0.997449880340539,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002,
      "loss": 2.5541,
      "step": 3178
    },
    {
      "epoch": 0.9977637412217035,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 2.4854,
      "step": 3179
    },
    {
      "epoch": 0.9977637412217035,
      "eval_loss": 1.7612242698669434,
      "eval_runtime": 122.9212,
      "eval_samples_per_second": 8.135,
      "eval_steps_per_second": 8.135,
      "step": 3179
    },
    {
      "epoch": 0.9977637412217035,
      "mmlu_eval_accuracy": 0.4094543774249976,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.35714285714285715,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3448275862068966,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "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": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.25,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.19230769230769232,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.55,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.4230769230769231,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.6162790697674418,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.331150726058139,
      "step": 3179
    },
    {
      "epoch": 0.9980776021028679,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.5063,
      "step": 3180
    },
    {
      "epoch": 0.9983914629840324,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 2.0594,
      "step": 3181
    },
    {
      "epoch": 0.9987053238651967,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.7996,
      "step": 3182
    },
    {
      "epoch": 0.9990191847463612,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 2.5581,
      "step": 3183
    },
    {
      "epoch": 0.9993330456275256,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 2.3198,
      "step": 3184
    },
    {
      "epoch": 0.99964690650869,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 2.0987,
      "step": 3185
    },
    {
      "epoch": 0.9999607673898544,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.6696,
      "step": 3186
    },
    {
      "epoch": 1.0002746282710189,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 2.0604,
      "step": 3187
    },
    {
      "epoch": 1.0005884891521832,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0002,
      "loss": 1.3877,
      "step": 3188
    },
    {
      "epoch": 1.0009023500333478,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.3389,
      "step": 3189
    },
    {
      "epoch": 1.0012162109145122,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3807,
      "step": 3190
    },
    {
      "epoch": 1.0015300717956765,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3194,
      "step": 3191
    },
    {
      "epoch": 1.001843932676841,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.3926,
      "step": 3192
    },
    {
      "epoch": 1.0021577935580055,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3041,
      "step": 3193
    },
    {
      "epoch": 1.0024716544391699,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3575,
      "step": 3194
    },
    {
      "epoch": 1.0027855153203342,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2774,
      "step": 3195
    },
    {
      "epoch": 1.0030993762014986,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3192,
      "step": 3196
    },
    {
      "epoch": 1.0034132370826632,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4229,
      "step": 3197
    },
    {
      "epoch": 1.0037270979638275,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3111,
      "step": 3198
    },
    {
      "epoch": 1.004040958844992,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2728,
      "step": 3199
    },
    {
      "epoch": 1.0043548197261565,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3139,
      "step": 3200
    },
    {
      "epoch": 1.0046686806073208,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2393,
      "step": 3201
    },
    {
      "epoch": 1.0049825414884852,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.3164,
      "step": 3202
    },
    {
      "epoch": 1.0052964023696496,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3657,
      "step": 3203
    },
    {
      "epoch": 1.0056102632508142,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.4096,
      "step": 3204
    },
    {
      "epoch": 1.0059241241319785,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.411,
      "step": 3205
    },
    {
      "epoch": 1.0062379850131429,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.2077,
      "step": 3206
    },
    {
      "epoch": 1.0065518458943072,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.2669,
      "step": 3207
    },
    {
      "epoch": 1.0068657067754718,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.4376,
      "step": 3208
    },
    {
      "epoch": 1.0071795676566362,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.1823,
      "step": 3209
    },
    {
      "epoch": 1.0074934285378006,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.4707,
      "step": 3210
    },
    {
      "epoch": 1.0078072894189651,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.4341,
      "step": 3211
    },
    {
      "epoch": 1.0081211503001295,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.5259,
      "step": 3212
    },
    {
      "epoch": 1.0084350111812939,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.3221,
      "step": 3213
    },
    {
      "epoch": 1.0087488720624582,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.7058,
      "step": 3214
    },
    {
      "epoch": 1.0090627329436228,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.5108,
      "step": 3215
    },
    {
      "epoch": 1.0093765938247872,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.6032,
      "step": 3216
    },
    {
      "epoch": 1.0096904547059515,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.7211,
      "step": 3217
    },
    {
      "epoch": 1.0100043155871161,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.6075,
      "step": 3218
    },
    {
      "epoch": 1.0103181764682805,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 2.0775,
      "step": 3219
    },
    {
      "epoch": 1.0106320373494448,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.9127,
      "step": 3220
    },
    {
      "epoch": 1.0109458982306092,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.9032,
      "step": 3221
    },
    {
      "epoch": 1.0112597591117738,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 2.2753,
      "step": 3222
    },
    {
      "epoch": 1.0115736199929382,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.9234,
      "step": 3223
    },
    {
      "epoch": 1.0118874808741025,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.9765,
      "step": 3224
    },
    {
      "epoch": 1.0122013417552669,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.9529,
      "step": 3225
    },
    {
      "epoch": 1.0125152026364315,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.9424,
      "step": 3226
    },
    {
      "epoch": 1.0128290635175958,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 2.0726,
      "step": 3227
    },
    {
      "epoch": 1.0131429243987602,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.0097,
      "step": 3228
    },
    {
      "epoch": 1.0134567852799248,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.1058,
      "step": 3229
    },
    {
      "epoch": 1.0137706461610891,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 2.1626,
      "step": 3230
    },
    {
      "epoch": 1.0140845070422535,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.7617,
      "step": 3231
    },
    {
      "epoch": 1.0143983679234179,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.7791,
      "step": 3232
    },
    {
      "epoch": 1.0147122288045825,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.5761,
      "step": 3233
    },
    {
      "epoch": 1.0150260896857468,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.8905,
      "step": 3234
    },
    {
      "epoch": 1.0153399505669112,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.9349,
      "step": 3235
    },
    {
      "epoch": 1.0156538114480755,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.2521,
      "step": 3236
    },
    {
      "epoch": 1.0159676723292401,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.7394,
      "step": 3237
    },
    {
      "epoch": 1.0162815332104045,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0921,
      "step": 3238
    },
    {
      "epoch": 1.0165953940915688,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.435,
      "step": 3239
    },
    {
      "epoch": 1.0169092549727334,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.322,
      "step": 3240
    },
    {
      "epoch": 1.0172231158538978,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.5587,
      "step": 3241
    },
    {
      "epoch": 1.0175369767350622,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2465,
      "step": 3242
    },
    {
      "epoch": 1.0178508376162265,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3037,
      "step": 3243
    },
    {
      "epoch": 1.018164698497391,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3711,
      "step": 3244
    },
    {
      "epoch": 1.0184785593785555,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.5833,
      "step": 3245
    },
    {
      "epoch": 1.0187924202597198,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2177,
      "step": 3246
    },
    {
      "epoch": 1.0191062811408842,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.3321,
      "step": 3247
    },
    {
      "epoch": 1.0194201420220488,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2415,
      "step": 3248
    },
    {
      "epoch": 1.0197340029032131,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.2932,
      "step": 3249
    },
    {
      "epoch": 1.0200478637843775,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.3618,
      "step": 3250
    },
    {
      "epoch": 1.020361724665542,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.1451,
      "step": 3251
    },
    {
      "epoch": 1.0206755855467065,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.3904,
      "step": 3252
    },
    {
      "epoch": 1.0209894464278708,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.3366,
      "step": 3253
    },
    {
      "epoch": 1.0213033073090352,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.47,
      "step": 3254
    },
    {
      "epoch": 1.0216171681901998,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.2812,
      "step": 3255
    },
    {
      "epoch": 1.0219310290713641,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.3385,
      "step": 3256
    },
    {
      "epoch": 1.0222448899525285,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.6318,
      "step": 3257
    },
    {
      "epoch": 1.022558750833693,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3832,
      "step": 3258
    },
    {
      "epoch": 1.0228726117148574,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.2538,
      "step": 3259
    },
    {
      "epoch": 1.0231864725960218,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.6401,
      "step": 3260
    },
    {
      "epoch": 1.0235003334771862,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.3818,
      "step": 3261
    },
    {
      "epoch": 1.0238141943583507,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.5104,
      "step": 3262
    },
    {
      "epoch": 1.0241280552395151,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.704,
      "step": 3263
    },
    {
      "epoch": 1.0244419161206795,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 2.0966,
      "step": 3264
    },
    {
      "epoch": 1.0247557770018438,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.7171,
      "step": 3265
    },
    {
      "epoch": 1.0250696378830084,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.6738,
      "step": 3266
    },
    {
      "epoch": 1.0253834987641728,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.6724,
      "step": 3267
    },
    {
      "epoch": 1.0256973596453371,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.9199,
      "step": 3268
    },
    {
      "epoch": 1.0260112205265017,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.8901,
      "step": 3269
    },
    {
      "epoch": 1.026325081407666,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.9366,
      "step": 3270
    },
    {
      "epoch": 1.0266389422888305,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.8268,
      "step": 3271
    },
    {
      "epoch": 1.0269528031699948,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.751,
      "step": 3272
    },
    {
      "epoch": 1.0272666640511594,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 2.2259,
      "step": 3273
    },
    {
      "epoch": 1.0275805249323238,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 2.0009,
      "step": 3274
    },
    {
      "epoch": 1.0278943858134881,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 2.0155,
      "step": 3275
    },
    {
      "epoch": 1.0282082466946525,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.7522,
      "step": 3276
    },
    {
      "epoch": 1.028522107575817,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 2.2466,
      "step": 3277
    },
    {
      "epoch": 1.0288359684569814,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.8462,
      "step": 3278
    },
    {
      "epoch": 1.0291498293381458,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.9953,
      "step": 3279
    },
    {
      "epoch": 1.0294636902193104,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 2.185,
      "step": 3280
    },
    {
      "epoch": 1.0297775511004748,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.9967,
      "step": 3281
    },
    {
      "epoch": 1.0300914119816391,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.7523,
      "step": 3282
    },
    {
      "epoch": 1.0304052728628035,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.7159,
      "step": 3283
    },
    {
      "epoch": 1.030719133743968,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.3885,
      "step": 3284
    },
    {
      "epoch": 1.0310329946251324,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.5344,
      "step": 3285
    },
    {
      "epoch": 1.0313468555062968,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002,
      "loss": 2.2785,
      "step": 3286
    },
    {
      "epoch": 1.0316607163874612,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.4047,
      "step": 3287
    },
    {
      "epoch": 1.0319745772686257,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4677,
      "step": 3288
    },
    {
      "epoch": 1.03228843814979,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.6365,
      "step": 3289
    },
    {
      "epoch": 1.0326022990309545,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3259,
      "step": 3290
    },
    {
      "epoch": 1.032916159912119,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.2246,
      "step": 3291
    },
    {
      "epoch": 1.0332300207932834,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.3766,
      "step": 3292
    },
    {
      "epoch": 1.0335438816744478,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4346,
      "step": 3293
    },
    {
      "epoch": 1.0338577425556121,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.4882,
      "step": 3294
    },
    {
      "epoch": 1.0341716034367767,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2832,
      "step": 3295
    },
    {
      "epoch": 1.034485464317941,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.3905,
      "step": 3296
    },
    {
      "epoch": 1.0347993251991054,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.175,
      "step": 3297
    },
    {
      "epoch": 1.0351131860802698,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.4527,
      "step": 3298
    },
    {
      "epoch": 1.0354270469614344,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.2958,
      "step": 3299
    },
    {
      "epoch": 1.0357409078425988,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.272,
      "step": 3300
    },
    {
      "epoch": 1.0360547687237631,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.3249,
      "step": 3301
    },
    {
      "epoch": 1.0363686296049277,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2704,
      "step": 3302
    },
    {
      "epoch": 1.036682490486092,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.4118,
      "step": 3303
    },
    {
      "epoch": 1.0369963513672564,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.3258,
      "step": 3304
    },
    {
      "epoch": 1.0373102122484208,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.3525,
      "step": 3305
    },
    {
      "epoch": 1.0376240731295854,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.3024,
      "step": 3306
    },
    {
      "epoch": 1.0379379340107497,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.4737,
      "step": 3307
    },
    {
      "epoch": 1.038251794891914,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.5573,
      "step": 3308
    },
    {
      "epoch": 1.0385656557730787,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.495,
      "step": 3309
    },
    {
      "epoch": 1.038879516654243,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.4378,
      "step": 3310
    },
    {
      "epoch": 1.0391933775354074,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.5224,
      "step": 3311
    },
    {
      "epoch": 1.0395072384165718,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.5927,
      "step": 3312
    },
    {
      "epoch": 1.0398210992977364,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.3604,
      "step": 3313
    },
    {
      "epoch": 1.0401349601789007,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.4609,
      "step": 3314
    },
    {
      "epoch": 1.040448821060065,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.7497,
      "step": 3315
    },
    {
      "epoch": 1.0407626819412295,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.7503,
      "step": 3316
    },
    {
      "epoch": 1.041076542822394,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.9545,
      "step": 3317
    },
    {
      "epoch": 1.0413904037035584,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.0486,
      "step": 3318
    },
    {
      "epoch": 1.0417042645847228,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.8276,
      "step": 3319
    },
    {
      "epoch": 1.0420181254658873,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.8632,
      "step": 3320
    },
    {
      "epoch": 1.0423319863470517,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.9163,
      "step": 3321
    },
    {
      "epoch": 1.042645847228216,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.9768,
      "step": 3322
    },
    {
      "epoch": 1.0429597081093804,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.9516,
      "step": 3323
    },
    {
      "epoch": 1.043273568990545,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.8407,
      "step": 3324
    },
    {
      "epoch": 1.0435874298717094,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.9521,
      "step": 3325
    },
    {
      "epoch": 1.0439012907528737,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.9723,
      "step": 3326
    },
    {
      "epoch": 1.044215151634038,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.6469,
      "step": 3327
    },
    {
      "epoch": 1.0445290125152027,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 2.2362,
      "step": 3328
    },
    {
      "epoch": 1.044842873396367,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.6164,
      "step": 3329
    },
    {
      "epoch": 1.0451567342775314,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.5419,
      "step": 3330
    },
    {
      "epoch": 1.045470595158696,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0002,
      "loss": 2.2275,
      "step": 3331
    },
    {
      "epoch": 1.0457844560398604,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.0776,
      "step": 3332
    },
    {
      "epoch": 1.0460983169210247,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.9309,
      "step": 3333
    },
    {
      "epoch": 1.046412177802189,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 2.0388,
      "step": 3334
    },
    {
      "epoch": 1.0467260386833537,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 2.0802,
      "step": 3335
    },
    {
      "epoch": 1.047039899564518,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.5374,
      "step": 3336
    },
    {
      "epoch": 1.0473537604456824,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.5432,
      "step": 3337
    },
    {
      "epoch": 1.047667621326847,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.4247,
      "step": 3338
    },
    {
      "epoch": 1.0479814822080114,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2922,
      "step": 3339
    },
    {
      "epoch": 1.0482953430891757,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2041,
      "step": 3340
    },
    {
      "epoch": 1.04860920397034,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4376,
      "step": 3341
    },
    {
      "epoch": 1.0489230648515047,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3131,
      "step": 3342
    },
    {
      "epoch": 1.049236925732669,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.4706,
      "step": 3343
    },
    {
      "epoch": 1.0495507866138334,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.228,
      "step": 3344
    },
    {
      "epoch": 1.0498646474949977,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.3428,
      "step": 3345
    },
    {
      "epoch": 1.0501785083761623,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2709,
      "step": 3346
    },
    {
      "epoch": 1.0504923692573267,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2898,
      "step": 3347
    },
    {
      "epoch": 1.050806230138491,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.292,
      "step": 3348
    },
    {
      "epoch": 1.0511200910196556,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.3754,
      "step": 3349
    },
    {
      "epoch": 1.05143395190082,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.2088,
      "step": 3350
    },
    {
      "epoch": 1.0517478127819844,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3938,
      "step": 3351
    },
    {
      "epoch": 1.0520616736631487,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.359,
      "step": 3352
    },
    {
      "epoch": 1.0523755345443133,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.4313,
      "step": 3353
    },
    {
      "epoch": 1.0526893954254777,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.4475,
      "step": 3354
    },
    {
      "epoch": 1.053003256306642,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.2955,
      "step": 3355
    },
    {
      "epoch": 1.0533171171878064,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.4495,
      "step": 3356
    },
    {
      "epoch": 1.053630978068971,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.512,
      "step": 3357
    },
    {
      "epoch": 1.0539448389501354,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.3396,
      "step": 3358
    },
    {
      "epoch": 1.0542586998312997,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.5656,
      "step": 3359
    },
    {
      "epoch": 1.0545725607124643,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.4603,
      "step": 3360
    },
    {
      "epoch": 1.0548864215936287,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.5615,
      "step": 3361
    },
    {
      "epoch": 1.055200282474793,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.4789,
      "step": 3362
    },
    {
      "epoch": 1.0555141433559574,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.4517,
      "step": 3363
    },
    {
      "epoch": 1.055828004237122,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.6078,
      "step": 3364
    },
    {
      "epoch": 1.0561418651182863,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.7573,
      "step": 3365
    },
    {
      "epoch": 1.0564557259994507,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.919,
      "step": 3366
    },
    {
      "epoch": 1.0564557259994507,
      "eval_loss": 1.782015085220337,
      "eval_runtime": 122.9733,
      "eval_samples_per_second": 8.132,
      "eval_steps_per_second": 8.132,
      "step": 3366
    },
    {
      "epoch": 1.0564557259994507,
      "mmlu_eval_accuracy": 0.4088775674784756,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "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": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.21875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.19230769230769232,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.6,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.38461538461538464,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.6162790697674418,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.48484848484848486,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.37142857142857144,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.36363636363636365,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.275355286920561,
      "step": 3366
    },
    {
      "epoch": 1.056769586880615,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.7448,
      "step": 3367
    },
    {
      "epoch": 1.0570834477617796,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.7405,
      "step": 3368
    },
    {
      "epoch": 1.057397308642944,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.8165,
      "step": 3369
    },
    {
      "epoch": 1.0577111695241084,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 2.2684,
      "step": 3370
    },
    {
      "epoch": 1.058025030405273,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.6534,
      "step": 3371
    },
    {
      "epoch": 1.0583388912864373,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 2.0191,
      "step": 3372
    },
    {
      "epoch": 1.0586527521676017,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.571,
      "step": 3373
    },
    {
      "epoch": 1.058966613048766,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.961,
      "step": 3374
    },
    {
      "epoch": 1.0592804739299306,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.9378,
      "step": 3375
    },
    {
      "epoch": 1.059594334811095,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.8649,
      "step": 3376
    },
    {
      "epoch": 1.0599081956922594,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.7798,
      "step": 3377
    },
    {
      "epoch": 1.0602220565734237,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.8226,
      "step": 3378
    },
    {
      "epoch": 1.0605359174545883,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.0467,
      "step": 3379
    },
    {
      "epoch": 1.0608497783357527,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 2.1341,
      "step": 3380
    },
    {
      "epoch": 1.061163639216917,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.6338,
      "step": 3381
    },
    {
      "epoch": 1.0614775000980816,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 2.1787,
      "step": 3382
    },
    {
      "epoch": 1.061791360979246,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.8559,
      "step": 3383
    },
    {
      "epoch": 1.0621052218604103,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.1064,
      "step": 3384
    },
    {
      "epoch": 1.0624190827415747,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.6618,
      "step": 3385
    },
    {
      "epoch": 1.0627329436227393,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002,
      "loss": 2.4034,
      "step": 3386
    },
    {
      "epoch": 1.0630468045039037,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.4687,
      "step": 3387
    },
    {
      "epoch": 1.063360665385068,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.3391,
      "step": 3388
    },
    {
      "epoch": 1.0636745262662326,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.3221,
      "step": 3389
    },
    {
      "epoch": 1.063988387147397,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3004,
      "step": 3390
    },
    {
      "epoch": 1.0643022480285613,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3623,
      "step": 3391
    },
    {
      "epoch": 1.0646161089097257,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.207,
      "step": 3392
    },
    {
      "epoch": 1.0649299697908903,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3389,
      "step": 3393
    },
    {
      "epoch": 1.0652438306720546,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4117,
      "step": 3394
    },
    {
      "epoch": 1.065557691553219,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.4149,
      "step": 3395
    },
    {
      "epoch": 1.0658715524343834,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.5053,
      "step": 3396
    },
    {
      "epoch": 1.066185413315548,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.6008,
      "step": 3397
    },
    {
      "epoch": 1.0664992741967123,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.401,
      "step": 3398
    },
    {
      "epoch": 1.0668131350778767,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.4048,
      "step": 3399
    },
    {
      "epoch": 1.0671269959590413,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.3891,
      "step": 3400
    },
    {
      "epoch": 1.0674408568402056,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3934,
      "step": 3401
    },
    {
      "epoch": 1.06775471772137,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.4672,
      "step": 3402
    },
    {
      "epoch": 1.0680685786025343,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.3179,
      "step": 3403
    },
    {
      "epoch": 1.068382439483699,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3361,
      "step": 3404
    },
    {
      "epoch": 1.0686963003648633,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.2874,
      "step": 3405
    },
    {
      "epoch": 1.0690101612460277,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.4479,
      "step": 3406
    },
    {
      "epoch": 1.069324022127192,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.2963,
      "step": 3407
    },
    {
      "epoch": 1.0696378830083566,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.352,
      "step": 3408
    },
    {
      "epoch": 1.069951743889521,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.2518,
      "step": 3409
    },
    {
      "epoch": 1.0702656047706853,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.3461,
      "step": 3410
    },
    {
      "epoch": 1.07057946565185,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.3596,
      "step": 3411
    },
    {
      "epoch": 1.0708933265330143,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.4757,
      "step": 3412
    },
    {
      "epoch": 1.0712071874141786,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.4412,
      "step": 3413
    },
    {
      "epoch": 1.071521048295343,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3274,
      "step": 3414
    },
    {
      "epoch": 1.0718349091765076,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.6775,
      "step": 3415
    },
    {
      "epoch": 1.072148770057672,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.5285,
      "step": 3416
    },
    {
      "epoch": 1.0724626309388363,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.9699,
      "step": 3417
    },
    {
      "epoch": 1.072776491820001,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.8587,
      "step": 3418
    },
    {
      "epoch": 1.0730903527011653,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 2.1302,
      "step": 3419
    },
    {
      "epoch": 1.0734042135823296,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 2.017,
      "step": 3420
    },
    {
      "epoch": 1.073718074463494,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.7599,
      "step": 3421
    },
    {
      "epoch": 1.0740319353446586,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.8603,
      "step": 3422
    },
    {
      "epoch": 1.074345796225823,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.7622,
      "step": 3423
    },
    {
      "epoch": 1.0746596571069873,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 1.8043,
      "step": 3424
    },
    {
      "epoch": 1.0749735179881517,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 2.0542,
      "step": 3425
    },
    {
      "epoch": 1.0752873788693162,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 2.1464,
      "step": 3426
    },
    {
      "epoch": 1.0756012397504806,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.0404,
      "step": 3427
    },
    {
      "epoch": 1.075915100631645,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.6908,
      "step": 3428
    },
    {
      "epoch": 1.0762289615128093,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.0362,
      "step": 3429
    },
    {
      "epoch": 1.076542822393974,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 2.2215,
      "step": 3430
    },
    {
      "epoch": 1.0768566832751383,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.4921,
      "step": 3431
    },
    {
      "epoch": 1.0771705441563026,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.6788,
      "step": 3432
    },
    {
      "epoch": 1.0774844050374672,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.6164,
      "step": 3433
    },
    {
      "epoch": 1.0777982659186316,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 2.1122,
      "step": 3434
    },
    {
      "epoch": 1.078112126799796,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.5735,
      "step": 3435
    },
    {
      "epoch": 1.0784259876809603,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 2.0643,
      "step": 3436
    },
    {
      "epoch": 1.078739848562125,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.5819,
      "step": 3437
    },
    {
      "epoch": 1.0790537094432893,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.2702,
      "step": 3438
    },
    {
      "epoch": 1.0793675703244536,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.3787,
      "step": 3439
    },
    {
      "epoch": 1.0796814312056182,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3857,
      "step": 3440
    },
    {
      "epoch": 1.0799952920867826,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3512,
      "step": 3441
    },
    {
      "epoch": 1.080309152967947,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2302,
      "step": 3442
    },
    {
      "epoch": 1.0806230138491113,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4563,
      "step": 3443
    },
    {
      "epoch": 1.0809368747302759,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.3174,
      "step": 3444
    },
    {
      "epoch": 1.0812507356114403,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.5022,
      "step": 3445
    },
    {
      "epoch": 1.0815645964926046,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2804,
      "step": 3446
    },
    {
      "epoch": 1.081878457373769,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.402,
      "step": 3447
    },
    {
      "epoch": 1.0821923182549336,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.3564,
      "step": 3448
    },
    {
      "epoch": 1.082506179136098,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.4267,
      "step": 3449
    },
    {
      "epoch": 1.0828200400172623,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.2007,
      "step": 3450
    },
    {
      "epoch": 1.0831339008984269,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.3224,
      "step": 3451
    },
    {
      "epoch": 1.0834477617795912,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.3205,
      "step": 3452
    },
    {
      "epoch": 1.0837616226607556,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.3749,
      "step": 3453
    },
    {
      "epoch": 1.08407548354192,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2017,
      "step": 3454
    },
    {
      "epoch": 1.0843893444230845,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.2634,
      "step": 3455
    },
    {
      "epoch": 1.084703205304249,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.3602,
      "step": 3456
    },
    {
      "epoch": 1.0850170661854133,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.3545,
      "step": 3457
    },
    {
      "epoch": 1.0853309270665776,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.3156,
      "step": 3458
    },
    {
      "epoch": 1.0856447879477422,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.6673,
      "step": 3459
    },
    {
      "epoch": 1.0859586488289066,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.5901,
      "step": 3460
    },
    {
      "epoch": 1.086272509710071,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.2643,
      "step": 3461
    },
    {
      "epoch": 1.0865863705912355,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.6226,
      "step": 3462
    },
    {
      "epoch": 1.0869002314724,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.3418,
      "step": 3463
    },
    {
      "epoch": 1.0872140923535643,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.7287,
      "step": 3464
    },
    {
      "epoch": 1.0875279532347286,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.5898,
      "step": 3465
    },
    {
      "epoch": 1.0878418141158932,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.8764,
      "step": 3466
    },
    {
      "epoch": 1.0881556749970576,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.5382,
      "step": 3467
    },
    {
      "epoch": 1.088469535878222,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.8086,
      "step": 3468
    },
    {
      "epoch": 1.0887833967593865,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.5665,
      "step": 3469
    },
    {
      "epoch": 1.0890972576405509,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 2.0351,
      "step": 3470
    },
    {
      "epoch": 1.0894111185217152,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.0884,
      "step": 3471
    },
    {
      "epoch": 1.0897249794028796,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 1.8107,
      "step": 3472
    },
    {
      "epoch": 1.0900388402840442,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.7368,
      "step": 3473
    },
    {
      "epoch": 1.0903527011652085,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.4573,
      "step": 3474
    },
    {
      "epoch": 1.090666562046373,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.1211,
      "step": 3475
    },
    {
      "epoch": 1.0909804229275373,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 2.2903,
      "step": 3476
    },
    {
      "epoch": 1.0912942838087019,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.309,
      "step": 3477
    },
    {
      "epoch": 1.0916081446898662,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002,
      "loss": 2.4323,
      "step": 3478
    },
    {
      "epoch": 1.0919220055710306,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.7412,
      "step": 3479
    },
    {
      "epoch": 1.092235866452195,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.823,
      "step": 3480
    },
    {
      "epoch": 1.0925497273333595,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 2.1499,
      "step": 3481
    },
    {
      "epoch": 1.092863588214524,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.7287,
      "step": 3482
    },
    {
      "epoch": 1.0931774490956883,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.7538,
      "step": 3483
    },
    {
      "epoch": 1.0934913099768528,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.8354,
      "step": 3484
    },
    {
      "epoch": 1.0938051708580172,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.2461,
      "step": 3485
    },
    {
      "epoch": 1.0941190317391816,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 2.5617,
      "step": 3486
    },
    {
      "epoch": 1.094432892620346,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.5741,
      "step": 3487
    },
    {
      "epoch": 1.0947467535015105,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.3492,
      "step": 3488
    },
    {
      "epoch": 1.0950606143826749,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.3904,
      "step": 3489
    },
    {
      "epoch": 1.0953744752638392,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.3113,
      "step": 3490
    },
    {
      "epoch": 1.0956883361450038,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.3793,
      "step": 3491
    },
    {
      "epoch": 1.0960021970261682,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.3095,
      "step": 3492
    },
    {
      "epoch": 1.0963160579073326,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.3662,
      "step": 3493
    },
    {
      "epoch": 1.096629918788497,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2684,
      "step": 3494
    },
    {
      "epoch": 1.0969437796696615,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3659,
      "step": 3495
    },
    {
      "epoch": 1.0972576405508259,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.3242,
      "step": 3496
    },
    {
      "epoch": 1.0975715014319902,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3947,
      "step": 3497
    },
    {
      "epoch": 1.0978853623131548,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.2836,
      "step": 3498
    },
    {
      "epoch": 1.0981992231943192,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.1691,
      "step": 3499
    },
    {
      "epoch": 1.0985130840754835,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.3641,
      "step": 3500
    },
    {
      "epoch": 1.098826944956648,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.4494,
      "step": 3501
    },
    {
      "epoch": 1.0991408058378125,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.2806,
      "step": 3502
    },
    {
      "epoch": 1.0994546667189768,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.3992,
      "step": 3503
    },
    {
      "epoch": 1.0997685276001412,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.4798,
      "step": 3504
    },
    {
      "epoch": 1.1000823884813056,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.332,
      "step": 3505
    },
    {
      "epoch": 1.1003962493624702,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.3876,
      "step": 3506
    },
    {
      "epoch": 1.1007101102436345,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.5442,
      "step": 3507
    },
    {
      "epoch": 1.1010239711247989,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.2679,
      "step": 3508
    },
    {
      "epoch": 1.1013378320059632,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.4113,
      "step": 3509
    },
    {
      "epoch": 1.1016516928871278,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.5659,
      "step": 3510
    },
    {
      "epoch": 1.1019655537682922,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.6196,
      "step": 3511
    },
    {
      "epoch": 1.1022794146494566,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.4541,
      "step": 3512
    },
    {
      "epoch": 1.1025932755306211,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.64,
      "step": 3513
    },
    {
      "epoch": 1.1029071364117855,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.4209,
      "step": 3514
    },
    {
      "epoch": 1.1032209972929499,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.5357,
      "step": 3515
    },
    {
      "epoch": 1.1035348581741142,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.8676,
      "step": 3516
    },
    {
      "epoch": 1.1038487190552788,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.4921,
      "step": 3517
    },
    {
      "epoch": 1.1041625799364432,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.8682,
      "step": 3518
    },
    {
      "epoch": 1.1044764408176075,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.6149,
      "step": 3519
    },
    {
      "epoch": 1.1047903016987721,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.7332,
      "step": 3520
    },
    {
      "epoch": 1.1051041625799365,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 2.0951,
      "step": 3521
    },
    {
      "epoch": 1.1054180234611009,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.7416,
      "step": 3522
    },
    {
      "epoch": 1.1057318843422652,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.9467,
      "step": 3523
    },
    {
      "epoch": 1.1060457452234298,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.9566,
      "step": 3524
    },
    {
      "epoch": 1.1063596061045942,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.1813,
      "step": 3525
    },
    {
      "epoch": 1.1066734669857585,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.0633,
      "step": 3526
    },
    {
      "epoch": 1.1069873278669229,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.1788,
      "step": 3527
    },
    {
      "epoch": 1.1073011887480875,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.8167,
      "step": 3528
    },
    {
      "epoch": 1.1076150496292518,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.9935,
      "step": 3529
    },
    {
      "epoch": 1.1079289105104162,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.9675,
      "step": 3530
    },
    {
      "epoch": 1.1082427713915808,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 2.2365,
      "step": 3531
    },
    {
      "epoch": 1.1085566322727451,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 1.8775,
      "step": 3532
    },
    {
      "epoch": 1.1088704931539095,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.58,
      "step": 3533
    },
    {
      "epoch": 1.1091843540350739,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.2402,
      "step": 3534
    },
    {
      "epoch": 1.1094982149162385,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 1.6955,
      "step": 3535
    },
    {
      "epoch": 1.1098120757974028,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.1472,
      "step": 3536
    },
    {
      "epoch": 1.1101259366785672,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4235,
      "step": 3537
    },
    {
      "epoch": 1.1104397975597315,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.299,
      "step": 3538
    },
    {
      "epoch": 1.1107536584408961,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2197,
      "step": 3539
    },
    {
      "epoch": 1.1110675193220605,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.5051,
      "step": 3540
    },
    {
      "epoch": 1.1113813802032249,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.3248,
      "step": 3541
    },
    {
      "epoch": 1.1116952410843894,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.5145,
      "step": 3542
    },
    {
      "epoch": 1.1120091019655538,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3977,
      "step": 3543
    },
    {
      "epoch": 1.1123229628467182,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.4126,
      "step": 3544
    },
    {
      "epoch": 1.1126368237278825,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4288,
      "step": 3545
    },
    {
      "epoch": 1.1129506846090471,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2885,
      "step": 3546
    },
    {
      "epoch": 1.1132645454902115,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2296,
      "step": 3547
    },
    {
      "epoch": 1.1135784063713758,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.5819,
      "step": 3548
    },
    {
      "epoch": 1.1138922672525404,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.3405,
      "step": 3549
    },
    {
      "epoch": 1.1142061281337048,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.38,
      "step": 3550
    },
    {
      "epoch": 1.1145199890148692,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.3125,
      "step": 3551
    },
    {
      "epoch": 1.1148338498960335,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.2946,
      "step": 3552
    },
    {
      "epoch": 1.115147710777198,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.4761,
      "step": 3553
    },
    {
      "epoch": 1.115147710777198,
      "eval_loss": 1.785923719406128,
      "eval_runtime": 122.646,
      "eval_samples_per_second": 8.154,
      "eval_steps_per_second": 8.154,
      "step": 3553
    },
    {
      "epoch": 1.115147710777198,
      "mmlu_eval_accuracy": 0.41082880955145434,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "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": 0.2,
      "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": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.6,
      "mmlu_eval_accuracy_high_school_statistics": 0.17391304347826086,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.38461538461538464,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5697674418604651,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.48484848484848486,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.36363636363636365,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.1708553617176551,
      "step": 3553
    },
    {
      "epoch": 1.1154615716583625,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.4877,
      "step": 3554
    },
    {
      "epoch": 1.1157754325395268,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.3422,
      "step": 3555
    },
    {
      "epoch": 1.1160892934206912,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.5457,
      "step": 3556
    },
    {
      "epoch": 1.1164031543018558,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.5776,
      "step": 3557
    },
    {
      "epoch": 1.1167170151830201,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.3775,
      "step": 3558
    },
    {
      "epoch": 1.1170308760641845,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.4051,
      "step": 3559
    },
    {
      "epoch": 1.1173447369453489,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.2419,
      "step": 3560
    },
    {
      "epoch": 1.1176585978265134,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.3213,
      "step": 3561
    },
    {
      "epoch": 1.1179724587076778,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.5882,
      "step": 3562
    },
    {
      "epoch": 1.1182863195888422,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.4452,
      "step": 3563
    },
    {
      "epoch": 1.1186001804700068,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.5798,
      "step": 3564
    },
    {
      "epoch": 1.1189140413511711,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.8714,
      "step": 3565
    },
    {
      "epoch": 1.1192279022323355,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.5531,
      "step": 3566
    },
    {
      "epoch": 1.1195417631134998,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.5999,
      "step": 3567
    },
    {
      "epoch": 1.1198556239946644,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.6179,
      "step": 3568
    },
    {
      "epoch": 1.1201694848758288,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.8053,
      "step": 3569
    },
    {
      "epoch": 1.1204833457569932,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.7818,
      "step": 3570
    },
    {
      "epoch": 1.1207972066381577,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.8914,
      "step": 3571
    },
    {
      "epoch": 1.121111067519322,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.9739,
      "step": 3572
    },
    {
      "epoch": 1.1214249284004865,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 2.1106,
      "step": 3573
    },
    {
      "epoch": 1.1217387892816508,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.9206,
      "step": 3574
    },
    {
      "epoch": 1.1220526501628154,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.8017,
      "step": 3575
    },
    {
      "epoch": 1.1223665110439798,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.6244,
      "step": 3576
    },
    {
      "epoch": 1.1226803719251441,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.9483,
      "step": 3577
    },
    {
      "epoch": 1.1229942328063085,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.5826,
      "step": 3578
    },
    {
      "epoch": 1.123308093687473,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.9219,
      "step": 3579
    },
    {
      "epoch": 1.1236219545686374,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 2.1785,
      "step": 3580
    },
    {
      "epoch": 1.1239358154498018,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.1084,
      "step": 3581
    },
    {
      "epoch": 1.1242496763309664,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 2.7719,
      "step": 3582
    },
    {
      "epoch": 1.1245635372121308,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.6691,
      "step": 3583
    },
    {
      "epoch": 1.1248773980932951,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.8765,
      "step": 3584
    },
    {
      "epoch": 1.1251912589744595,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.8913,
      "step": 3585
    },
    {
      "epoch": 1.125505119855624,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002,
      "loss": 2.8588,
      "step": 3586
    },
    {
      "epoch": 1.1258189807367884,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.473,
      "step": 3587
    },
    {
      "epoch": 1.1261328416179528,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2529,
      "step": 3588
    },
    {
      "epoch": 1.1264467024991172,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3749,
      "step": 3589
    },
    {
      "epoch": 1.1267605633802817,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2713,
      "step": 3590
    },
    {
      "epoch": 1.127074424261446,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4169,
      "step": 3591
    },
    {
      "epoch": 1.1273882851426105,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.245,
      "step": 3592
    },
    {
      "epoch": 1.127702146023775,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3938,
      "step": 3593
    },
    {
      "epoch": 1.1280160069049394,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3715,
      "step": 3594
    },
    {
      "epoch": 1.1283298677861038,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.3243,
      "step": 3595
    },
    {
      "epoch": 1.1286437286672681,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2939,
      "step": 3596
    },
    {
      "epoch": 1.1289575895484327,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2146,
      "step": 3597
    },
    {
      "epoch": 1.129271450429597,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3606,
      "step": 3598
    },
    {
      "epoch": 1.1295853113107615,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.4294,
      "step": 3599
    },
    {
      "epoch": 1.129899172191926,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.4255,
      "step": 3600
    },
    {
      "epoch": 1.1302130330730904,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.4143,
      "step": 3601
    },
    {
      "epoch": 1.1305268939542548,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.3901,
      "step": 3602
    },
    {
      "epoch": 1.1308407548354191,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.4703,
      "step": 3603
    },
    {
      "epoch": 1.1311546157165837,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.2592,
      "step": 3604
    },
    {
      "epoch": 1.131468476597748,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.3548,
      "step": 3605
    },
    {
      "epoch": 1.1317823374789124,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.5083,
      "step": 3606
    },
    {
      "epoch": 1.1320961983600768,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.5108,
      "step": 3607
    },
    {
      "epoch": 1.1324100592412414,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.2707,
      "step": 3608
    },
    {
      "epoch": 1.1327239201224057,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.5571,
      "step": 3609
    },
    {
      "epoch": 1.13303778100357,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.5364,
      "step": 3610
    },
    {
      "epoch": 1.1333516418847345,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.2,
      "step": 3611
    },
    {
      "epoch": 1.133665502765899,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.3271,
      "step": 3612
    },
    {
      "epoch": 1.1339793636470634,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.3633,
      "step": 3613
    },
    {
      "epoch": 1.1342932245282278,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.514,
      "step": 3614
    },
    {
      "epoch": 1.1346070854093924,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.6272,
      "step": 3615
    },
    {
      "epoch": 1.1349209462905567,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.6598,
      "step": 3616
    },
    {
      "epoch": 1.135234807171721,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.7269,
      "step": 3617
    },
    {
      "epoch": 1.1355486680528855,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.8545,
      "step": 3618
    },
    {
      "epoch": 1.13586252893405,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.9285,
      "step": 3619
    },
    {
      "epoch": 1.1361763898152144,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.7557,
      "step": 3620
    },
    {
      "epoch": 1.1364902506963788,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.9605,
      "step": 3621
    },
    {
      "epoch": 1.1368041115775434,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 2.262,
      "step": 3622
    },
    {
      "epoch": 1.1371179724587077,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.7123,
      "step": 3623
    },
    {
      "epoch": 1.137431833339872,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 2.228,
      "step": 3624
    },
    {
      "epoch": 1.1377456942210364,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.6424,
      "step": 3625
    },
    {
      "epoch": 1.138059555102201,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 2.1712,
      "step": 3626
    },
    {
      "epoch": 1.1383734159833654,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.734,
      "step": 3627
    },
    {
      "epoch": 1.1386872768645298,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.7847,
      "step": 3628
    },
    {
      "epoch": 1.1390011377456943,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.7135,
      "step": 3629
    },
    {
      "epoch": 1.1393149986268587,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 2.2577,
      "step": 3630
    },
    {
      "epoch": 1.139628859508023,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.34,
      "step": 3631
    },
    {
      "epoch": 1.1399427203891874,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.2824,
      "step": 3632
    },
    {
      "epoch": 1.140256581270352,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.7479,
      "step": 3633
    },
    {
      "epoch": 1.1405704421515164,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.5148,
      "step": 3634
    },
    {
      "epoch": 1.1408843030326807,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 1.7588,
      "step": 3635
    },
    {
      "epoch": 1.141198163913845,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.4824,
      "step": 3636
    },
    {
      "epoch": 1.1415120247950097,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.9549,
      "step": 3637
    },
    {
      "epoch": 1.141825885676174,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.3516,
      "step": 3638
    },
    {
      "epoch": 1.1421397465573384,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3547,
      "step": 3639
    },
    {
      "epoch": 1.1424536074385028,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.2967,
      "step": 3640
    },
    {
      "epoch": 1.1427674683196674,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.4629,
      "step": 3641
    },
    {
      "epoch": 1.1430813292008317,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.6071,
      "step": 3642
    },
    {
      "epoch": 1.143395190081996,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.5825,
      "step": 3643
    },
    {
      "epoch": 1.1437090509631607,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3035,
      "step": 3644
    },
    {
      "epoch": 1.144022911844325,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4255,
      "step": 3645
    },
    {
      "epoch": 1.1443367727254894,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2383,
      "step": 3646
    },
    {
      "epoch": 1.1446506336066538,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.5538,
      "step": 3647
    },
    {
      "epoch": 1.1449644944878183,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.3389,
      "step": 3648
    },
    {
      "epoch": 1.1452783553689827,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.4563,
      "step": 3649
    },
    {
      "epoch": 1.145592216250147,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.2892,
      "step": 3650
    },
    {
      "epoch": 1.1459060771313117,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.3073,
      "step": 3651
    },
    {
      "epoch": 1.146219938012476,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.3571,
      "step": 3652
    },
    {
      "epoch": 1.1465337988936404,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2609,
      "step": 3653
    },
    {
      "epoch": 1.1468476597748047,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.4362,
      "step": 3654
    },
    {
      "epoch": 1.1471615206559693,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.4871,
      "step": 3655
    },
    {
      "epoch": 1.1474753815371337,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.2907,
      "step": 3656
    },
    {
      "epoch": 1.147789242418298,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.258,
      "step": 3657
    },
    {
      "epoch": 1.1481031032994626,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.3671,
      "step": 3658
    },
    {
      "epoch": 1.148416964180627,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.233,
      "step": 3659
    },
    {
      "epoch": 1.1487308250617914,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.3437,
      "step": 3660
    },
    {
      "epoch": 1.1490446859429557,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.419,
      "step": 3661
    },
    {
      "epoch": 1.14935854682412,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.6437,
      "step": 3662
    },
    {
      "epoch": 1.1496724077052847,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.9029,
      "step": 3663
    },
    {
      "epoch": 1.149986268586449,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.5138,
      "step": 3664
    },
    {
      "epoch": 1.1503001294676134,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.5226,
      "step": 3665
    },
    {
      "epoch": 1.150613990348778,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.9024,
      "step": 3666
    },
    {
      "epoch": 1.1509278512299423,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.8113,
      "step": 3667
    },
    {
      "epoch": 1.1512417121111067,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 2.169,
      "step": 3668
    },
    {
      "epoch": 1.151555572992271,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.7122,
      "step": 3669
    },
    {
      "epoch": 1.1518694338734357,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 2.3097,
      "step": 3670
    },
    {
      "epoch": 1.1521832947546,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 2.2051,
      "step": 3671
    },
    {
      "epoch": 1.1524971556357644,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.7081,
      "step": 3672
    },
    {
      "epoch": 1.152811016516929,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.8949,
      "step": 3673
    },
    {
      "epoch": 1.1531248773980933,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 2.0396,
      "step": 3674
    },
    {
      "epoch": 1.1534387382792577,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.1069,
      "step": 3675
    },
    {
      "epoch": 1.153752599160422,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 2.4829,
      "step": 3676
    },
    {
      "epoch": 1.1540664600415866,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.8779,
      "step": 3677
    },
    {
      "epoch": 1.154380320922751,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.9209,
      "step": 3678
    },
    {
      "epoch": 1.1546941818039154,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.9027,
      "step": 3679
    },
    {
      "epoch": 1.15500804268508,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.2018,
      "step": 3680
    },
    {
      "epoch": 1.1553219035662443,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.5032,
      "step": 3681
    },
    {
      "epoch": 1.1556357644474087,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.9819,
      "step": 3682
    },
    {
      "epoch": 1.155949625328573,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.6673,
      "step": 3683
    },
    {
      "epoch": 1.1562634862097376,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 2.1791,
      "step": 3684
    },
    {
      "epoch": 1.156577347090902,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.086,
      "step": 3685
    },
    {
      "epoch": 1.1568912079720663,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 2.2853,
      "step": 3686
    },
    {
      "epoch": 1.1572050688532307,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.5367,
      "step": 3687
    },
    {
      "epoch": 1.1575189297343953,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3871,
      "step": 3688
    },
    {
      "epoch": 1.1578327906155597,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4207,
      "step": 3689
    },
    {
      "epoch": 1.158146651496724,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.2606,
      "step": 3690
    },
    {
      "epoch": 1.1584605123778884,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.5775,
      "step": 3691
    },
    {
      "epoch": 1.158774373259053,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.3002,
      "step": 3692
    },
    {
      "epoch": 1.1590882341402173,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.2388,
      "step": 3693
    },
    {
      "epoch": 1.1594020950213817,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.3587,
      "step": 3694
    },
    {
      "epoch": 1.1597159559025463,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.3413,
      "step": 3695
    },
    {
      "epoch": 1.1600298167837106,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.352,
      "step": 3696
    },
    {
      "epoch": 1.160343677664875,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.1805,
      "step": 3697
    },
    {
      "epoch": 1.1606575385460394,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.4525,
      "step": 3698
    },
    {
      "epoch": 1.160971399427204,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.3703,
      "step": 3699
    },
    {
      "epoch": 1.1612852603083683,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.3686,
      "step": 3700
    },
    {
      "epoch": 1.1615991211895327,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.2846,
      "step": 3701
    },
    {
      "epoch": 1.1619129820706973,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.3523,
      "step": 3702
    },
    {
      "epoch": 1.1622268429518616,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.4451,
      "step": 3703
    },
    {
      "epoch": 1.162540703833026,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.4352,
      "step": 3704
    },
    {
      "epoch": 1.1628545647141904,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.4363,
      "step": 3705
    },
    {
      "epoch": 1.163168425595355,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.3166,
      "step": 3706
    },
    {
      "epoch": 1.1634822864765193,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.3889,
      "step": 3707
    },
    {
      "epoch": 1.1637961473576837,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.4201,
      "step": 3708
    },
    {
      "epoch": 1.1641100082388482,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.3659,
      "step": 3709
    },
    {
      "epoch": 1.1644238691200126,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.6474,
      "step": 3710
    },
    {
      "epoch": 1.164737730001177,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.538,
      "step": 3711
    },
    {
      "epoch": 1.1650515908823413,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.8065,
      "step": 3712
    },
    {
      "epoch": 1.165365451763506,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.5863,
      "step": 3713
    },
    {
      "epoch": 1.1656793126446703,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.583,
      "step": 3714
    },
    {
      "epoch": 1.1659931735258346,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.7434,
      "step": 3715
    },
    {
      "epoch": 1.166307034406999,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.264,
      "step": 3716
    },
    {
      "epoch": 1.1666208952881636,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.3905,
      "step": 3717
    },
    {
      "epoch": 1.166934756169328,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.6922,
      "step": 3718
    },
    {
      "epoch": 1.1672486170504923,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002,
      "loss": 2.1253,
      "step": 3719
    },
    {
      "epoch": 1.1675624779316567,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.9142,
      "step": 3720
    },
    {
      "epoch": 1.1678763388128213,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 2.3627,
      "step": 3721
    },
    {
      "epoch": 1.1681901996939856,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.8973,
      "step": 3722
    },
    {
      "epoch": 1.16850406057515,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.0719,
      "step": 3723
    },
    {
      "epoch": 1.1688179214563146,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 2.2261,
      "step": 3724
    },
    {
      "epoch": 1.169131782337479,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.9728,
      "step": 3725
    },
    {
      "epoch": 1.1694456432186433,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.9261,
      "step": 3726
    },
    {
      "epoch": 1.1697595040998077,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.9135,
      "step": 3727
    },
    {
      "epoch": 1.1700733649809723,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 1.9522,
      "step": 3728
    },
    {
      "epoch": 1.1703872258621366,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.5351,
      "step": 3729
    },
    {
      "epoch": 1.170701086743301,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.7964,
      "step": 3730
    },
    {
      "epoch": 1.1710149476244656,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.8993,
      "step": 3731
    },
    {
      "epoch": 1.17132880850563,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.8904,
      "step": 3732
    },
    {
      "epoch": 1.1716426693867943,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.7531,
      "step": 3733
    },
    {
      "epoch": 1.1719565302679587,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.8729,
      "step": 3734
    },
    {
      "epoch": 1.1722703911491232,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.9941,
      "step": 3735
    },
    {
      "epoch": 1.1725842520302876,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002,
      "loss": 3.032,
      "step": 3736
    },
    {
      "epoch": 1.172898112911452,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.227,
      "step": 3737
    },
    {
      "epoch": 1.1732119737926165,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2885,
      "step": 3738
    },
    {
      "epoch": 1.173525834673781,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2461,
      "step": 3739
    },
    {
      "epoch": 1.1738396955549453,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.4896,
      "step": 3740
    },
    {
      "epoch": 1.1738396955549453,
      "eval_loss": 1.7994587421417236,
      "eval_runtime": 123.3438,
      "eval_samples_per_second": 8.107,
      "eval_steps_per_second": 8.107,
      "step": 3740
    },
    {
      "epoch": 1.1738396955549453,
      "mmlu_eval_accuracy": 0.40424781951426475,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.18181818181818182,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.3333333333333333,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.42857142857142855,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.5666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.38461538461538464,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5465116279069767,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.2571428571428571,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.36363636363636365,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.0847891762434152,
      "step": 3740
    },
    {
      "epoch": 1.1741535564361096,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.5034,
      "step": 3741
    },
    {
      "epoch": 1.174467417317274,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.4662,
      "step": 3742
    },
    {
      "epoch": 1.1747812781984386,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.3567,
      "step": 3743
    },
    {
      "epoch": 1.175095139079603,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.4361,
      "step": 3744
    },
    {
      "epoch": 1.1754089999607673,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.4061,
      "step": 3745
    },
    {
      "epoch": 1.175722860841932,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.4657,
      "step": 3746
    },
    {
      "epoch": 1.1760367217230963,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.2493,
      "step": 3747
    },
    {
      "epoch": 1.1763505826042606,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.369,
      "step": 3748
    },
    {
      "epoch": 1.176664443485425,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.6183,
      "step": 3749
    },
    {
      "epoch": 1.1769783043665896,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.326,
      "step": 3750
    },
    {
      "epoch": 1.177292165247754,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.4562,
      "step": 3751
    },
    {
      "epoch": 1.1776060261289183,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.3425,
      "step": 3752
    },
    {
      "epoch": 1.1779198870100829,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.2195,
      "step": 3753
    },
    {
      "epoch": 1.1782337478912472,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.2509,
      "step": 3754
    },
    {
      "epoch": 1.1785476087724116,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.2863,
      "step": 3755
    },
    {
      "epoch": 1.178861469653576,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.3202,
      "step": 3756
    },
    {
      "epoch": 1.1791753305347406,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.4818,
      "step": 3757
    },
    {
      "epoch": 1.179489191415905,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.2737,
      "step": 3758
    },
    {
      "epoch": 1.1798030522970693,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.3306,
      "step": 3759
    },
    {
      "epoch": 1.1801169131782339,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.3687,
      "step": 3760
    },
    {
      "epoch": 1.1804307740593982,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.5493,
      "step": 3761
    },
    {
      "epoch": 1.1807446349405626,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.6763,
      "step": 3762
    },
    {
      "epoch": 1.181058495821727,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.6181,
      "step": 3763
    },
    {
      "epoch": 1.1813723567028915,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.8927,
      "step": 3764
    },
    {
      "epoch": 1.181686217584056,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.9099,
      "step": 3765
    },
    {
      "epoch": 1.1820000784652203,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 2.0363,
      "step": 3766
    },
    {
      "epoch": 1.1823139393463846,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.8547,
      "step": 3767
    },
    {
      "epoch": 1.1826278002275492,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.9647,
      "step": 3768
    },
    {
      "epoch": 1.1829416611087136,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.7915,
      "step": 3769
    },
    {
      "epoch": 1.183255521989878,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 2.1384,
      "step": 3770
    },
    {
      "epoch": 1.1835693828710423,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.9578,
      "step": 3771
    },
    {
      "epoch": 1.1838832437522069,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.1569,
      "step": 3772
    },
    {
      "epoch": 1.1841971046333712,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.8285,
      "step": 3773
    },
    {
      "epoch": 1.1845109655145356,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.6542,
      "step": 3774
    },
    {
      "epoch": 1.1848248263957002,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.9014,
      "step": 3775
    },
    {
      "epoch": 1.1851386872768646,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.6977,
      "step": 3776
    },
    {
      "epoch": 1.185452548158029,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.8656,
      "step": 3777
    },
    {
      "epoch": 1.1857664090391933,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.8882,
      "step": 3778
    },
    {
      "epoch": 1.1860802699203579,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 2.0131,
      "step": 3779
    },
    {
      "epoch": 1.1863941308015222,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.326,
      "step": 3780
    },
    {
      "epoch": 1.1867079916826866,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 2.2844,
      "step": 3781
    },
    {
      "epoch": 1.1870218525638512,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.4935,
      "step": 3782
    },
    {
      "epoch": 1.1873357134450155,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.8096,
      "step": 3783
    },
    {
      "epoch": 1.18764957432618,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.9125,
      "step": 3784
    },
    {
      "epoch": 1.1879634352073443,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 2.1215,
      "step": 3785
    },
    {
      "epoch": 1.1882772960885088,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 2.6974,
      "step": 3786
    },
    {
      "epoch": 1.1885911569696732,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.4455,
      "step": 3787
    },
    {
      "epoch": 1.1889050178508376,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3066,
      "step": 3788
    },
    {
      "epoch": 1.1892188787320022,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.2273,
      "step": 3789
    },
    {
      "epoch": 1.1895327396131665,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.3491,
      "step": 3790
    },
    {
      "epoch": 1.1898466004943309,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3615,
      "step": 3791
    },
    {
      "epoch": 1.1901604613754952,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2648,
      "step": 3792
    },
    {
      "epoch": 1.1904743222566596,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.5231,
      "step": 3793
    },
    {
      "epoch": 1.1907881831378242,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.4563,
      "step": 3794
    },
    {
      "epoch": 1.1911020440189886,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.3943,
      "step": 3795
    },
    {
      "epoch": 1.191415904900153,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3072,
      "step": 3796
    },
    {
      "epoch": 1.1917297657813175,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.1707,
      "step": 3797
    },
    {
      "epoch": 1.1920436266624819,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.3971,
      "step": 3798
    },
    {
      "epoch": 1.1923574875436462,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.4205,
      "step": 3799
    },
    {
      "epoch": 1.1926713484248106,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.2905,
      "step": 3800
    },
    {
      "epoch": 1.1929852093059752,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.2874,
      "step": 3801
    },
    {
      "epoch": 1.1932990701871395,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.4268,
      "step": 3802
    },
    {
      "epoch": 1.193612931068304,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.2796,
      "step": 3803
    },
    {
      "epoch": 1.1939267919494685,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.3508,
      "step": 3804
    },
    {
      "epoch": 1.1942406528306329,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.2908,
      "step": 3805
    },
    {
      "epoch": 1.1945545137117972,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.3608,
      "step": 3806
    },
    {
      "epoch": 1.1948683745929616,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.322,
      "step": 3807
    },
    {
      "epoch": 1.1951822354741262,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.2416,
      "step": 3808
    },
    {
      "epoch": 1.1954960963552905,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.426,
      "step": 3809
    },
    {
      "epoch": 1.195809957236455,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.4441,
      "step": 3810
    },
    {
      "epoch": 1.1961238181176195,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.4874,
      "step": 3811
    },
    {
      "epoch": 1.1964376789987838,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.6444,
      "step": 3812
    },
    {
      "epoch": 1.1967515398799482,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.3802,
      "step": 3813
    },
    {
      "epoch": 1.1970654007611126,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.4772,
      "step": 3814
    },
    {
      "epoch": 1.1973792616422771,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.7324,
      "step": 3815
    },
    {
      "epoch": 1.1976931225234415,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.5248,
      "step": 3816
    },
    {
      "epoch": 1.1980069834046059,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.651,
      "step": 3817
    },
    {
      "epoch": 1.1983208442857702,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.5842,
      "step": 3818
    },
    {
      "epoch": 1.1986347051669348,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.7978,
      "step": 3819
    },
    {
      "epoch": 1.1989485660480992,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.7602,
      "step": 3820
    },
    {
      "epoch": 1.1992624269292635,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.8248,
      "step": 3821
    },
    {
      "epoch": 1.199576287810428,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.8459,
      "step": 3822
    },
    {
      "epoch": 1.1998901486915925,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.8383,
      "step": 3823
    },
    {
      "epoch": 1.2002040095727569,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.9263,
      "step": 3824
    },
    {
      "epoch": 1.2005178704539212,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.1008,
      "step": 3825
    },
    {
      "epoch": 1.2008317313350858,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 1.9617,
      "step": 3826
    },
    {
      "epoch": 1.2011455922162502,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.8431,
      "step": 3827
    },
    {
      "epoch": 1.2014594530974145,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.1326,
      "step": 3828
    },
    {
      "epoch": 1.201773313978579,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 2.2094,
      "step": 3829
    },
    {
      "epoch": 1.2020871748597435,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.9739,
      "step": 3830
    },
    {
      "epoch": 1.2024010357409078,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.2287,
      "step": 3831
    },
    {
      "epoch": 1.2027148966220722,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 2.101,
      "step": 3832
    },
    {
      "epoch": 1.2030287575032368,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.8535,
      "step": 3833
    },
    {
      "epoch": 1.2033426183844012,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.7292,
      "step": 3834
    },
    {
      "epoch": 1.2036564792655655,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 2.4744,
      "step": 3835
    },
    {
      "epoch": 1.2039703401467299,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 2.7216,
      "step": 3836
    },
    {
      "epoch": 1.2042842010278945,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.3469,
      "step": 3837
    },
    {
      "epoch": 1.2045980619090588,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3098,
      "step": 3838
    },
    {
      "epoch": 1.2049119227902232,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2628,
      "step": 3839
    },
    {
      "epoch": 1.2052257836713878,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.324,
      "step": 3840
    },
    {
      "epoch": 1.2055396445525521,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.3925,
      "step": 3841
    },
    {
      "epoch": 1.2058535054337165,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.3853,
      "step": 3842
    },
    {
      "epoch": 1.2061673663148809,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.4041,
      "step": 3843
    },
    {
      "epoch": 1.2064812271960454,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.4943,
      "step": 3844
    },
    {
      "epoch": 1.2067950880772098,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.4522,
      "step": 3845
    },
    {
      "epoch": 1.2071089489583742,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.395,
      "step": 3846
    },
    {
      "epoch": 1.2074228098395385,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.4094,
      "step": 3847
    },
    {
      "epoch": 1.2077366707207031,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2752,
      "step": 3848
    },
    {
      "epoch": 1.2080505316018675,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.3864,
      "step": 3849
    },
    {
      "epoch": 1.2083643924830318,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.3639,
      "step": 3850
    },
    {
      "epoch": 1.2086782533641962,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.3288,
      "step": 3851
    },
    {
      "epoch": 1.2089921142453608,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.4141,
      "step": 3852
    },
    {
      "epoch": 1.2093059751265252,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.3423,
      "step": 3853
    },
    {
      "epoch": 1.2096198360076895,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.3284,
      "step": 3854
    },
    {
      "epoch": 1.209933696888854,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.4631,
      "step": 3855
    },
    {
      "epoch": 1.2102475577700185,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.2618,
      "step": 3856
    },
    {
      "epoch": 1.2105614186511828,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.4483,
      "step": 3857
    },
    {
      "epoch": 1.2108752795323472,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.4485,
      "step": 3858
    },
    {
      "epoch": 1.2111891404135118,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.4875,
      "step": 3859
    },
    {
      "epoch": 1.2115030012946761,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.4438,
      "step": 3860
    },
    {
      "epoch": 1.2118168621758405,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.4452,
      "step": 3861
    },
    {
      "epoch": 1.212130723057005,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.5181,
      "step": 3862
    },
    {
      "epoch": 1.2124445839381695,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.4851,
      "step": 3863
    },
    {
      "epoch": 1.2127584448193338,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.5521,
      "step": 3864
    },
    {
      "epoch": 1.2130723057004982,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.69,
      "step": 3865
    },
    {
      "epoch": 1.2133861665816628,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.7878,
      "step": 3866
    },
    {
      "epoch": 1.2137000274628271,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.5545,
      "step": 3867
    },
    {
      "epoch": 1.2140138883439915,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.5646,
      "step": 3868
    },
    {
      "epoch": 1.214327749225156,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.9537,
      "step": 3869
    },
    {
      "epoch": 1.2146416101063204,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.7254,
      "step": 3870
    },
    {
      "epoch": 1.2149554709874848,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.6222,
      "step": 3871
    },
    {
      "epoch": 1.2152693318686492,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 2.2117,
      "step": 3872
    },
    {
      "epoch": 1.2155831927498135,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.3514,
      "step": 3873
    },
    {
      "epoch": 1.215897053630978,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.8412,
      "step": 3874
    },
    {
      "epoch": 1.2162109145121425,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002,
      "loss": 2.1561,
      "step": 3875
    },
    {
      "epoch": 1.2165247753933068,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 2.0765,
      "step": 3876
    },
    {
      "epoch": 1.2168386362744714,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.8924,
      "step": 3877
    },
    {
      "epoch": 1.2171524971556358,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.2694,
      "step": 3878
    },
    {
      "epoch": 1.2174663580368001,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 2.0347,
      "step": 3879
    },
    {
      "epoch": 1.2177802189179645,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.9706,
      "step": 3880
    },
    {
      "epoch": 1.218094079799129,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 2.0052,
      "step": 3881
    },
    {
      "epoch": 1.2184079406802935,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.8245,
      "step": 3882
    },
    {
      "epoch": 1.2187218015614578,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.787,
      "step": 3883
    },
    {
      "epoch": 1.2190356624426224,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.7334,
      "step": 3884
    },
    {
      "epoch": 1.2193495233237868,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 1.853,
      "step": 3885
    },
    {
      "epoch": 1.2196633842049511,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002,
      "loss": 2.5601,
      "step": 3886
    },
    {
      "epoch": 1.2199772450861155,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.5006,
      "step": 3887
    },
    {
      "epoch": 1.22029110596728,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3305,
      "step": 3888
    },
    {
      "epoch": 1.2206049668484444,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3465,
      "step": 3889
    },
    {
      "epoch": 1.2209188277296088,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4851,
      "step": 3890
    },
    {
      "epoch": 1.2212326886107734,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3539,
      "step": 3891
    },
    {
      "epoch": 1.2215465494919377,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.486,
      "step": 3892
    },
    {
      "epoch": 1.2218604103731021,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2972,
      "step": 3893
    },
    {
      "epoch": 1.2221742712542665,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3355,
      "step": 3894
    },
    {
      "epoch": 1.222488132135431,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3257,
      "step": 3895
    },
    {
      "epoch": 1.2228019930165954,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.4074,
      "step": 3896
    },
    {
      "epoch": 1.2231158538977598,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.3131,
      "step": 3897
    },
    {
      "epoch": 1.2234297147789241,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.3275,
      "step": 3898
    },
    {
      "epoch": 1.2237435756600887,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.3617,
      "step": 3899
    },
    {
      "epoch": 1.224057436541253,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.3141,
      "step": 3900
    },
    {
      "epoch": 1.2243712974224175,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.2291,
      "step": 3901
    },
    {
      "epoch": 1.2246851583035818,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.1175,
      "step": 3902
    },
    {
      "epoch": 1.2249990191847464,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.5335,
      "step": 3903
    },
    {
      "epoch": 1.2253128800659108,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.4631,
      "step": 3904
    },
    {
      "epoch": 1.2256267409470751,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.1809,
      "step": 3905
    },
    {
      "epoch": 1.2259406018282397,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.3521,
      "step": 3906
    },
    {
      "epoch": 1.226254462709404,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.1823,
      "step": 3907
    },
    {
      "epoch": 1.2265683235905684,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.3867,
      "step": 3908
    },
    {
      "epoch": 1.2268821844717328,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.3609,
      "step": 3909
    },
    {
      "epoch": 1.2271960453528974,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.4966,
      "step": 3910
    },
    {
      "epoch": 1.2275099062340618,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.2672,
      "step": 3911
    },
    {
      "epoch": 1.2278237671152261,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.2476,
      "step": 3912
    },
    {
      "epoch": 1.2281376279963907,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.6099,
      "step": 3913
    },
    {
      "epoch": 1.228451488877555,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.5465,
      "step": 3914
    },
    {
      "epoch": 1.2287653497587194,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.7972,
      "step": 3915
    },
    {
      "epoch": 1.2290792106398838,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.8126,
      "step": 3916
    },
    {
      "epoch": 1.2293930715210484,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.6936,
      "step": 3917
    },
    {
      "epoch": 1.2297069324022127,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.853,
      "step": 3918
    },
    {
      "epoch": 1.230020793283377,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.8028,
      "step": 3919
    },
    {
      "epoch": 1.2303346541645417,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.4233,
      "step": 3920
    },
    {
      "epoch": 1.230648515045706,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.8537,
      "step": 3921
    },
    {
      "epoch": 1.2309623759268704,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.1938,
      "step": 3922
    },
    {
      "epoch": 1.2312762368080348,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.5935,
      "step": 3923
    },
    {
      "epoch": 1.2315900976891994,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.5907,
      "step": 3924
    },
    {
      "epoch": 1.2319039585703637,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.5777,
      "step": 3925
    },
    {
      "epoch": 1.232217819451528,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 2.1,
      "step": 3926
    },
    {
      "epoch": 1.2325316803326924,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.9012,
      "step": 3927
    },
    {
      "epoch": 1.2325316803326924,
      "eval_loss": 1.7945857048034668,
      "eval_runtime": 123.0461,
      "eval_samples_per_second": 8.127,
      "eval_steps_per_second": 8.127,
      "step": 3927
    },
    {
      "epoch": 1.2325316803326924,
      "mmlu_eval_accuracy": 0.3989423337967472,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.25,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.3333333333333333,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
      "mmlu_eval_accuracy_high_school_psychology": 0.55,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.4230769230769231,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5116279069767442,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.48484848484848486,
      "mmlu_eval_accuracy_philosophy": 0.4117647058823529,
      "mmlu_eval_accuracy_prehistory": 0.2857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5,
      "mmlu_eval_accuracy_us_foreign_policy": 0.36363636363636365,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.1527687325187947,
      "step": 3927
    },
    {
      "epoch": 1.232845541213857,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.9394,
      "step": 3928
    },
    {
      "epoch": 1.2331594020950214,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.5628,
      "step": 3929
    },
    {
      "epoch": 1.2334732629761858,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 2.0467,
      "step": 3930
    },
    {
      "epoch": 1.2337871238573501,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 2.5013,
      "step": 3931
    },
    {
      "epoch": 1.2341009847385147,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.7232,
      "step": 3932
    },
    {
      "epoch": 1.234414845619679,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 2.0227,
      "step": 3933
    },
    {
      "epoch": 1.2347287065008434,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.9215,
      "step": 3934
    },
    {
      "epoch": 1.235042567382008,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 2.0873,
      "step": 3935
    },
    {
      "epoch": 1.2353564282631724,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.3414,
      "step": 3936
    },
    {
      "epoch": 1.2356702891443367,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.5739,
      "step": 3937
    },
    {
      "epoch": 1.235984150025501,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3787,
      "step": 3938
    },
    {
      "epoch": 1.2362980109066657,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.4379,
      "step": 3939
    },
    {
      "epoch": 1.23661187178783,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.5092,
      "step": 3940
    },
    {
      "epoch": 1.2369257326689944,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.2708,
      "step": 3941
    },
    {
      "epoch": 1.237239593550159,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3394,
      "step": 3942
    },
    {
      "epoch": 1.2375534544313234,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.2485,
      "step": 3943
    },
    {
      "epoch": 1.2378673153124877,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.2642,
      "step": 3944
    },
    {
      "epoch": 1.238181176193652,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.2613,
      "step": 3945
    },
    {
      "epoch": 1.2384950370748167,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.3982,
      "step": 3946
    },
    {
      "epoch": 1.238808897955981,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.5133,
      "step": 3947
    },
    {
      "epoch": 1.2391227588371454,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.3276,
      "step": 3948
    },
    {
      "epoch": 1.2394366197183098,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2339,
      "step": 3949
    },
    {
      "epoch": 1.2397504805994743,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.5666,
      "step": 3950
    },
    {
      "epoch": 1.2400643414806387,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.4242,
      "step": 3951
    },
    {
      "epoch": 1.240378202361803,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.524,
      "step": 3952
    },
    {
      "epoch": 1.2406920632429674,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.4408,
      "step": 3953
    },
    {
      "epoch": 1.241005924124132,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.3607,
      "step": 3954
    },
    {
      "epoch": 1.2413197850052964,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.3292,
      "step": 3955
    },
    {
      "epoch": 1.2416336458864607,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.3111,
      "step": 3956
    },
    {
      "epoch": 1.2419475067676253,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.5065,
      "step": 3957
    },
    {
      "epoch": 1.2422613676487897,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.4442,
      "step": 3958
    },
    {
      "epoch": 1.242575228529954,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.6287,
      "step": 3959
    },
    {
      "epoch": 1.2428890894111184,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.4796,
      "step": 3960
    },
    {
      "epoch": 1.243202950292283,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.3231,
      "step": 3961
    },
    {
      "epoch": 1.2435168111734474,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.4953,
      "step": 3962
    },
    {
      "epoch": 1.2438306720546117,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.4468,
      "step": 3963
    },
    {
      "epoch": 1.2441445329357763,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.7283,
      "step": 3964
    },
    {
      "epoch": 1.2444583938169407,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.7535,
      "step": 3965
    },
    {
      "epoch": 1.244772254698105,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.2766,
      "step": 3966
    },
    {
      "epoch": 1.2450861155792694,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.5373,
      "step": 3967
    },
    {
      "epoch": 1.245399976460434,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.6022,
      "step": 3968
    },
    {
      "epoch": 1.2457138373415984,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.5153,
      "step": 3969
    },
    {
      "epoch": 1.2460276982227627,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.7013,
      "step": 3970
    },
    {
      "epoch": 1.2463415591039273,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 2.0935,
      "step": 3971
    },
    {
      "epoch": 1.2466554199850917,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.8361,
      "step": 3972
    },
    {
      "epoch": 1.246969280866256,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.0163,
      "step": 3973
    },
    {
      "epoch": 1.2472831417474204,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.9271,
      "step": 3974
    },
    {
      "epoch": 1.247597002628585,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.6085,
      "step": 3975
    },
    {
      "epoch": 1.2479108635097493,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 2.0353,
      "step": 3976
    },
    {
      "epoch": 1.2482247243909137,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 2.0623,
      "step": 3977
    },
    {
      "epoch": 1.248538585272078,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.6417,
      "step": 3978
    },
    {
      "epoch": 1.2488524461532426,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.8706,
      "step": 3979
    },
    {
      "epoch": 1.249166307034407,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 2.0887,
      "step": 3980
    },
    {
      "epoch": 1.2494801679155714,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 2.0637,
      "step": 3981
    },
    {
      "epoch": 1.2497940287967357,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.684,
      "step": 3982
    },
    {
      "epoch": 1.2501078896779003,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.6825,
      "step": 3983
    },
    {
      "epoch": 1.2504217505590647,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.5943,
      "step": 3984
    },
    {
      "epoch": 1.250735611440229,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.3663,
      "step": 3985
    },
    {
      "epoch": 1.2510494723213936,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 2.5565,
      "step": 3986
    },
    {
      "epoch": 1.251363333202558,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.5542,
      "step": 3987
    },
    {
      "epoch": 1.2516771940837224,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.0363,
      "step": 3988
    },
    {
      "epoch": 1.2519910549648867,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.3929,
      "step": 3989
    },
    {
      "epoch": 1.2523049158460513,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.324,
      "step": 3990
    },
    {
      "epoch": 1.2526187767272157,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.413,
      "step": 3991
    },
    {
      "epoch": 1.25293263760838,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.4615,
      "step": 3992
    },
    {
      "epoch": 1.2532464984895446,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.3599,
      "step": 3993
    },
    {
      "epoch": 1.253560359370709,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.3142,
      "step": 3994
    },
    {
      "epoch": 1.2538742202518733,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.4337,
      "step": 3995
    },
    {
      "epoch": 1.2541880811330377,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.3202,
      "step": 3996
    },
    {
      "epoch": 1.2545019420142023,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.3973,
      "step": 3997
    },
    {
      "epoch": 1.2548158028953666,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.3098,
      "step": 3998
    },
    {
      "epoch": 1.255129663776531,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.2877,
      "step": 3999
    },
    {
      "epoch": 1.2554435246576956,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.3905,
      "step": 4000
    },
    {
      "epoch": 1.25575738553886,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.4616,
      "step": 4001
    },
    {
      "epoch": 1.2560712464200243,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.4432,
      "step": 4002
    },
    {
      "epoch": 1.2563851073011887,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.2567,
      "step": 4003
    },
    {
      "epoch": 1.256698968182353,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.3767,
      "step": 4004
    },
    {
      "epoch": 1.2570128290635176,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.4998,
      "step": 4005
    },
    {
      "epoch": 1.257326689944682,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.332,
      "step": 4006
    },
    {
      "epoch": 1.2576405508258466,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.3372,
      "step": 4007
    },
    {
      "epoch": 1.257954411707011,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.5853,
      "step": 4008
    },
    {
      "epoch": 1.2582682725881753,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2254,
      "step": 4009
    },
    {
      "epoch": 1.2585821334693397,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.6269,
      "step": 4010
    },
    {
      "epoch": 1.258895994350504,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.6903,
      "step": 4011
    },
    {
      "epoch": 1.2592098552316686,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.6119,
      "step": 4012
    },
    {
      "epoch": 1.259523716112833,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.6418,
      "step": 4013
    },
    {
      "epoch": 1.2598375769939973,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.6182,
      "step": 4014
    },
    {
      "epoch": 1.260151437875162,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2947,
      "step": 4015
    },
    {
      "epoch": 1.2604652987563263,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 1.6339,
      "step": 4016
    },
    {
      "epoch": 1.2607791596374907,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.7531,
      "step": 4017
    },
    {
      "epoch": 1.261093020518655,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.2208,
      "step": 4018
    },
    {
      "epoch": 1.2614068813998196,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.9273,
      "step": 4019
    },
    {
      "epoch": 1.261720742280984,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 2.0857,
      "step": 4020
    },
    {
      "epoch": 1.2620346031621483,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.8421,
      "step": 4021
    },
    {
      "epoch": 1.262348464043313,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.8256,
      "step": 4022
    },
    {
      "epoch": 1.2626623249244773,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.8785,
      "step": 4023
    },
    {
      "epoch": 1.2629761858056416,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 2.2222,
      "step": 4024
    },
    {
      "epoch": 1.263290046686806,
      "grad_norm": 0.984375,
      "learning_rate": 0.0002,
      "loss": 2.7284,
      "step": 4025
    },
    {
      "epoch": 1.2636039075679704,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 2.0976,
      "step": 4026
    },
    {
      "epoch": 1.263917768449135,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.9994,
      "step": 4027
    },
    {
      "epoch": 1.2642316293302993,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 2.2444,
      "step": 4028
    },
    {
      "epoch": 1.264545490211464,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.826,
      "step": 4029
    },
    {
      "epoch": 1.2648593510926283,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.9897,
      "step": 4030
    },
    {
      "epoch": 1.2651732119737926,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.7658,
      "step": 4031
    },
    {
      "epoch": 1.265487072854957,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.3832,
      "step": 4032
    },
    {
      "epoch": 1.2658009337361213,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.6482,
      "step": 4033
    },
    {
      "epoch": 1.266114794617286,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.6645,
      "step": 4034
    },
    {
      "epoch": 1.2664286554984503,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.5989,
      "step": 4035
    },
    {
      "epoch": 1.2667425163796147,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 2.2411,
      "step": 4036
    },
    {
      "epoch": 1.2670563772607792,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.4773,
      "step": 4037
    },
    {
      "epoch": 1.2673702381419436,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.1318,
      "step": 4038
    },
    {
      "epoch": 1.267684099023108,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3812,
      "step": 4039
    },
    {
      "epoch": 1.2679979599042723,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.3225,
      "step": 4040
    },
    {
      "epoch": 1.268311820785437,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2517,
      "step": 4041
    },
    {
      "epoch": 1.2686256816666013,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.3175,
      "step": 4042
    },
    {
      "epoch": 1.2689395425477656,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.4699,
      "step": 4043
    },
    {
      "epoch": 1.2692534034289302,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.3492,
      "step": 4044
    },
    {
      "epoch": 1.2695672643100946,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.2974,
      "step": 4045
    },
    {
      "epoch": 1.269881125191259,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.3531,
      "step": 4046
    },
    {
      "epoch": 1.2701949860724233,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.4536,
      "step": 4047
    },
    {
      "epoch": 1.270508846953588,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.2499,
      "step": 4048
    },
    {
      "epoch": 1.2708227078347523,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.3348,
      "step": 4049
    },
    {
      "epoch": 1.2711365687159166,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.3379,
      "step": 4050
    },
    {
      "epoch": 1.2714504295970812,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.4797,
      "step": 4051
    },
    {
      "epoch": 1.2717642904782456,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.3381,
      "step": 4052
    },
    {
      "epoch": 1.27207815135941,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.4228,
      "step": 4053
    },
    {
      "epoch": 1.2723920122405743,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.4404,
      "step": 4054
    },
    {
      "epoch": 1.2727058731217387,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.4416,
      "step": 4055
    },
    {
      "epoch": 1.2730197340029032,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.3466,
      "step": 4056
    },
    {
      "epoch": 1.2733335948840676,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.3853,
      "step": 4057
    },
    {
      "epoch": 1.2736474557652322,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.3325,
      "step": 4058
    },
    {
      "epoch": 1.2739613166463966,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.5336,
      "step": 4059
    },
    {
      "epoch": 1.274275177527561,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.3639,
      "step": 4060
    },
    {
      "epoch": 1.2745890384087253,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.5649,
      "step": 4061
    },
    {
      "epoch": 1.2749028992898896,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.4593,
      "step": 4062
    },
    {
      "epoch": 1.2752167601710542,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.5844,
      "step": 4063
    },
    {
      "epoch": 1.2755306210522186,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.4457,
      "step": 4064
    },
    {
      "epoch": 1.275844481933383,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.7896,
      "step": 4065
    },
    {
      "epoch": 1.2761583428145475,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.6183,
      "step": 4066
    },
    {
      "epoch": 1.276472203695712,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.4078,
      "step": 4067
    },
    {
      "epoch": 1.2767860645768763,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.736,
      "step": 4068
    },
    {
      "epoch": 1.2770999254580406,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.7595,
      "step": 4069
    },
    {
      "epoch": 1.2774137863392052,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 2.2398,
      "step": 4070
    },
    {
      "epoch": 1.2777276472203696,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.6555,
      "step": 4071
    },
    {
      "epoch": 1.278041508101534,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.4183,
      "step": 4072
    },
    {
      "epoch": 1.2783553689826985,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.6922,
      "step": 4073
    },
    {
      "epoch": 1.2786692298638629,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.9127,
      "step": 4074
    },
    {
      "epoch": 1.2789830907450273,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 2.0626,
      "step": 4075
    },
    {
      "epoch": 1.2792969516261916,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.9806,
      "step": 4076
    },
    {
      "epoch": 1.279610812507356,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.1743,
      "step": 4077
    },
    {
      "epoch": 1.2799246733885206,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 2.2794,
      "step": 4078
    },
    {
      "epoch": 1.280238534269685,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.6562,
      "step": 4079
    },
    {
      "epoch": 1.2805523951508495,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.0937,
      "step": 4080
    },
    {
      "epoch": 1.2808662560320139,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.783,
      "step": 4081
    },
    {
      "epoch": 1.2811801169131782,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.6448,
      "step": 4082
    },
    {
      "epoch": 1.2814939777943426,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.6891,
      "step": 4083
    },
    {
      "epoch": 1.281807838675507,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.092,
      "step": 4084
    },
    {
      "epoch": 1.2821216995566715,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.905,
      "step": 4085
    },
    {
      "epoch": 1.282435560437836,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 2.2499,
      "step": 4086
    },
    {
      "epoch": 1.2827494213190003,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.5193,
      "step": 4087
    },
    {
      "epoch": 1.2830632822001649,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3801,
      "step": 4088
    },
    {
      "epoch": 1.2833771430813292,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.273,
      "step": 4089
    },
    {
      "epoch": 1.2836910039624936,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.4607,
      "step": 4090
    },
    {
      "epoch": 1.284004864843658,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.4262,
      "step": 4091
    },
    {
      "epoch": 1.2843187257248225,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.4388,
      "step": 4092
    },
    {
      "epoch": 1.284632586605987,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1754,
      "step": 4093
    },
    {
      "epoch": 1.2849464474871513,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.3231,
      "step": 4094
    },
    {
      "epoch": 1.2852603083683158,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.1401,
      "step": 4095
    },
    {
      "epoch": 1.2855741692494802,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2105,
      "step": 4096
    },
    {
      "epoch": 1.2858880301306446,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.1321,
      "step": 4097
    },
    {
      "epoch": 1.286201891011809,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3177,
      "step": 4098
    },
    {
      "epoch": 1.2865157518929735,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.4483,
      "step": 4099
    },
    {
      "epoch": 1.2868296127741379,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.3236,
      "step": 4100
    },
    {
      "epoch": 1.2871434736553022,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.3283,
      "step": 4101
    },
    {
      "epoch": 1.2874573345364668,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.4345,
      "step": 4102
    },
    {
      "epoch": 1.2877711954176312,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.4472,
      "step": 4103
    },
    {
      "epoch": 1.2880850562987955,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.2968,
      "step": 4104
    },
    {
      "epoch": 1.28839891717996,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.4617,
      "step": 4105
    },
    {
      "epoch": 1.2887127780611243,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.1371,
      "step": 4106
    },
    {
      "epoch": 1.2890266389422889,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.4406,
      "step": 4107
    },
    {
      "epoch": 1.2893404998234532,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.4313,
      "step": 4108
    },
    {
      "epoch": 1.2896543607046178,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.5082,
      "step": 4109
    },
    {
      "epoch": 1.2899682215857822,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.613,
      "step": 4110
    },
    {
      "epoch": 1.2902820824669465,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4634,
      "step": 4111
    },
    {
      "epoch": 1.290595943348111,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.3694,
      "step": 4112
    },
    {
      "epoch": 1.2909098042292753,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4812,
      "step": 4113
    },
    {
      "epoch": 1.2912236651104398,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.5116,
      "step": 4114
    },
    {
      "epoch": 1.2912236651104398,
      "eval_loss": 1.781996250152588,
      "eval_runtime": 123.3814,
      "eval_samples_per_second": 8.105,
      "eval_steps_per_second": 8.105,
      "step": 4114
    },
    {
      "epoch": 1.2912236651104398,
      "mmlu_eval_accuracy": 0.4049450626432859,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.18181818181818182,
      "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": 0.24390243902439024,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "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": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.42857142857142855,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.5666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.5,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.18181818181818182,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5232558139534884,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.3,
      "mmlu_eval_accuracy_nutrition": 0.48484848484848486,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5,
      "mmlu_eval_accuracy_us_foreign_policy": 0.36363636363636365,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.2858767965741598,
      "step": 4114
    },
    {
      "epoch": 1.2915375259916042,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.8054,
      "step": 4115
    },
    {
      "epoch": 1.2918513868727686,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.6913,
      "step": 4116
    },
    {
      "epoch": 1.2921652477539332,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.6834,
      "step": 4117
    },
    {
      "epoch": 1.2924791086350975,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.7491,
      "step": 4118
    },
    {
      "epoch": 1.2927929695162619,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.9833,
      "step": 4119
    },
    {
      "epoch": 1.2931068303974262,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.7453,
      "step": 4120
    },
    {
      "epoch": 1.2934206912785908,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 2.0319,
      "step": 4121
    },
    {
      "epoch": 1.2937345521597552,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.905,
      "step": 4122
    },
    {
      "epoch": 1.2940484130409196,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 2.3193,
      "step": 4123
    },
    {
      "epoch": 1.2943622739220841,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.8523,
      "step": 4124
    },
    {
      "epoch": 1.2946761348032485,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.8196,
      "step": 4125
    },
    {
      "epoch": 1.2949899956844129,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.7706,
      "step": 4126
    },
    {
      "epoch": 1.2953038565655772,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 1.8172,
      "step": 4127
    },
    {
      "epoch": 1.2956177174467418,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.6792,
      "step": 4128
    },
    {
      "epoch": 1.2959315783279062,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.4328,
      "step": 4129
    },
    {
      "epoch": 1.2962454392090705,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.1643,
      "step": 4130
    },
    {
      "epoch": 1.2965593000902351,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.7205,
      "step": 4131
    },
    {
      "epoch": 1.2968731609713995,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 2.1623,
      "step": 4132
    },
    {
      "epoch": 1.2971870218525638,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.8202,
      "step": 4133
    },
    {
      "epoch": 1.2975008827337282,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.5895,
      "step": 4134
    },
    {
      "epoch": 1.2978147436148926,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 2.074,
      "step": 4135
    },
    {
      "epoch": 1.2981286044960572,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0002,
      "loss": 2.431,
      "step": 4136
    },
    {
      "epoch": 1.2984424653772215,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.5394,
      "step": 4137
    },
    {
      "epoch": 1.298756326258386,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.3751,
      "step": 4138
    },
    {
      "epoch": 1.2990701871395505,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3041,
      "step": 4139
    },
    {
      "epoch": 1.2993840480207148,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2854,
      "step": 4140
    },
    {
      "epoch": 1.2996979089018792,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.3936,
      "step": 4141
    },
    {
      "epoch": 1.3000117697830436,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.4258,
      "step": 4142
    },
    {
      "epoch": 1.3003256306642081,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1228,
      "step": 4143
    },
    {
      "epoch": 1.3006394915453725,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4088,
      "step": 4144
    },
    {
      "epoch": 1.3009533524265369,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.257,
      "step": 4145
    },
    {
      "epoch": 1.3012672133077015,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.3949,
      "step": 4146
    },
    {
      "epoch": 1.3015810741888658,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.3965,
      "step": 4147
    },
    {
      "epoch": 1.3018949350700302,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.2891,
      "step": 4148
    },
    {
      "epoch": 1.3022087959511945,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.2491,
      "step": 4149
    },
    {
      "epoch": 1.3025226568323591,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.3742,
      "step": 4150
    },
    {
      "epoch": 1.3028365177135235,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.3521,
      "step": 4151
    },
    {
      "epoch": 1.3031503785946879,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.3397,
      "step": 4152
    },
    {
      "epoch": 1.3034642394758524,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.3382,
      "step": 4153
    },
    {
      "epoch": 1.3037781003570168,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.415,
      "step": 4154
    },
    {
      "epoch": 1.3040919612381812,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.2975,
      "step": 4155
    },
    {
      "epoch": 1.3044058221193455,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.2389,
      "step": 4156
    },
    {
      "epoch": 1.3047196830005099,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.2805,
      "step": 4157
    },
    {
      "epoch": 1.3050335438816745,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.3648,
      "step": 4158
    },
    {
      "epoch": 1.3053474047628388,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.5424,
      "step": 4159
    },
    {
      "epoch": 1.3056612656440034,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.5212,
      "step": 4160
    },
    {
      "epoch": 1.3059751265251678,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.3956,
      "step": 4161
    },
    {
      "epoch": 1.3062889874063321,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.5542,
      "step": 4162
    },
    {
      "epoch": 1.3066028482874965,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.4505,
      "step": 4163
    },
    {
      "epoch": 1.3069167091686609,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.8976,
      "step": 4164
    },
    {
      "epoch": 1.3072305700498255,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.5576,
      "step": 4165
    },
    {
      "epoch": 1.3075444309309898,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.588,
      "step": 4166
    },
    {
      "epoch": 1.3078582918121542,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.5278,
      "step": 4167
    },
    {
      "epoch": 1.3081721526933188,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.1306,
      "step": 4168
    },
    {
      "epoch": 1.3084860135744831,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.8992,
      "step": 4169
    },
    {
      "epoch": 1.3087998744556475,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.1338,
      "step": 4170
    },
    {
      "epoch": 1.3091137353368119,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.6033,
      "step": 4171
    },
    {
      "epoch": 1.3094275962179764,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 2.4472,
      "step": 4172
    },
    {
      "epoch": 1.3097414570991408,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.4998,
      "step": 4173
    },
    {
      "epoch": 1.3100553179803052,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.0096,
      "step": 4174
    },
    {
      "epoch": 1.3103691788614698,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.1268,
      "step": 4175
    },
    {
      "epoch": 1.3106830397426341,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.8187,
      "step": 4176
    },
    {
      "epoch": 1.3109969006237985,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.6616,
      "step": 4177
    },
    {
      "epoch": 1.3113107615049628,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 2.2469,
      "step": 4178
    },
    {
      "epoch": 1.3116246223861274,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.246,
      "step": 4179
    },
    {
      "epoch": 1.3119384832672918,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.279,
      "step": 4180
    },
    {
      "epoch": 1.3122523441484562,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.8124,
      "step": 4181
    },
    {
      "epoch": 1.3125662050296207,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.6443,
      "step": 4182
    },
    {
      "epoch": 1.312880065910785,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.5571,
      "step": 4183
    },
    {
      "epoch": 1.3131939267919495,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.577,
      "step": 4184
    },
    {
      "epoch": 1.3135077876731138,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.8116,
      "step": 4185
    },
    {
      "epoch": 1.3138216485542782,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002,
      "loss": 2.5815,
      "step": 4186
    },
    {
      "epoch": 1.3141355094354428,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.5853,
      "step": 4187
    },
    {
      "epoch": 1.3144493703166071,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4784,
      "step": 4188
    },
    {
      "epoch": 1.3147632311977717,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.3305,
      "step": 4189
    },
    {
      "epoch": 1.315077092078936,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.253,
      "step": 4190
    },
    {
      "epoch": 1.3153909529601004,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.2798,
      "step": 4191
    },
    {
      "epoch": 1.3157048138412648,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.273,
      "step": 4192
    },
    {
      "epoch": 1.3160186747224292,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.3309,
      "step": 4193
    },
    {
      "epoch": 1.3163325356035938,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.4338,
      "step": 4194
    },
    {
      "epoch": 1.3166463964847581,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.2822,
      "step": 4195
    },
    {
      "epoch": 1.3169602573659225,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.2953,
      "step": 4196
    },
    {
      "epoch": 1.317274118247087,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.3072,
      "step": 4197
    },
    {
      "epoch": 1.3175879791282514,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3306,
      "step": 4198
    },
    {
      "epoch": 1.3179018400094158,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.361,
      "step": 4199
    },
    {
      "epoch": 1.3182157008905802,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.2953,
      "step": 4200
    },
    {
      "epoch": 1.3185295617717447,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.3355,
      "step": 4201
    },
    {
      "epoch": 1.318843422652909,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.2873,
      "step": 4202
    },
    {
      "epoch": 1.3191572835340735,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.3111,
      "step": 4203
    },
    {
      "epoch": 1.319471144415238,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.2623,
      "step": 4204
    },
    {
      "epoch": 1.3197850052964024,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.3187,
      "step": 4205
    },
    {
      "epoch": 1.3200988661775668,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2645,
      "step": 4206
    },
    {
      "epoch": 1.3204127270587311,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.5755,
      "step": 4207
    },
    {
      "epoch": 1.3207265879398957,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.4094,
      "step": 4208
    },
    {
      "epoch": 1.32104044882106,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.5062,
      "step": 4209
    },
    {
      "epoch": 1.3213543097022244,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.5129,
      "step": 4210
    },
    {
      "epoch": 1.321668170583389,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.4351,
      "step": 4211
    },
    {
      "epoch": 1.3219820314645534,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.5621,
      "step": 4212
    },
    {
      "epoch": 1.3222958923457178,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.2832,
      "step": 4213
    },
    {
      "epoch": 1.3226097532268821,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.686,
      "step": 4214
    },
    {
      "epoch": 1.3229236141080465,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.8661,
      "step": 4215
    },
    {
      "epoch": 1.323237474989211,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.4074,
      "step": 4216
    },
    {
      "epoch": 1.3235513358703754,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.8661,
      "step": 4217
    },
    {
      "epoch": 1.3238651967515398,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.6044,
      "step": 4218
    },
    {
      "epoch": 1.3241790576327044,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.6835,
      "step": 4219
    },
    {
      "epoch": 1.3244929185138687,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.7726,
      "step": 4220
    },
    {
      "epoch": 1.324806779395033,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.81,
      "step": 4221
    },
    {
      "epoch": 1.3251206402761975,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.9177,
      "step": 4222
    },
    {
      "epoch": 1.325434501157362,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.8793,
      "step": 4223
    },
    {
      "epoch": 1.3257483620385264,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.9163,
      "step": 4224
    },
    {
      "epoch": 1.3260622229196908,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.823,
      "step": 4225
    },
    {
      "epoch": 1.3263760838008554,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.8248,
      "step": 4226
    },
    {
      "epoch": 1.3266899446820197,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 2.4049,
      "step": 4227
    },
    {
      "epoch": 1.327003805563184,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 2.1969,
      "step": 4228
    },
    {
      "epoch": 1.3273176664443485,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.1375,
      "step": 4229
    },
    {
      "epoch": 1.327631527325513,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 2.0477,
      "step": 4230
    },
    {
      "epoch": 1.3279453882066774,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.9783,
      "step": 4231
    },
    {
      "epoch": 1.3282592490878418,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.9565,
      "step": 4232
    },
    {
      "epoch": 1.3285731099690063,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 2.3455,
      "step": 4233
    },
    {
      "epoch": 1.3288869708501707,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.9415,
      "step": 4234
    },
    {
      "epoch": 1.329200831731335,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.5521,
      "step": 4235
    },
    {
      "epoch": 1.3295146926124994,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.1808,
      "step": 4236
    },
    {
      "epoch": 1.3298285534936638,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.6169,
      "step": 4237
    },
    {
      "epoch": 1.3301424143748284,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2474,
      "step": 4238
    },
    {
      "epoch": 1.3304562752559927,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3209,
      "step": 4239
    },
    {
      "epoch": 1.3307701361371573,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4461,
      "step": 4240
    },
    {
      "epoch": 1.3310839970183217,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.3047,
      "step": 4241
    },
    {
      "epoch": 1.331397857899486,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.3713,
      "step": 4242
    },
    {
      "epoch": 1.3317117187806504,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.2495,
      "step": 4243
    },
    {
      "epoch": 1.3320255796618148,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.3678,
      "step": 4244
    },
    {
      "epoch": 1.3323394405429794,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.3047,
      "step": 4245
    },
    {
      "epoch": 1.3326533014241437,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.3341,
      "step": 4246
    },
    {
      "epoch": 1.332967162305308,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.4346,
      "step": 4247
    },
    {
      "epoch": 1.3332810231864727,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.1887,
      "step": 4248
    },
    {
      "epoch": 1.333594884067637,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.223,
      "step": 4249
    },
    {
      "epoch": 1.3339087449488014,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.2228,
      "step": 4250
    },
    {
      "epoch": 1.3342226058299658,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.2542,
      "step": 4251
    },
    {
      "epoch": 1.3345364667111304,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.4315,
      "step": 4252
    },
    {
      "epoch": 1.3348503275922947,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.357,
      "step": 4253
    },
    {
      "epoch": 1.335164188473459,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.487,
      "step": 4254
    },
    {
      "epoch": 1.3354780493546237,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.3971,
      "step": 4255
    },
    {
      "epoch": 1.335791910235788,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.3167,
      "step": 4256
    },
    {
      "epoch": 1.3361057711169524,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.3869,
      "step": 4257
    },
    {
      "epoch": 1.3364196319981168,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.4561,
      "step": 4258
    },
    {
      "epoch": 1.3367334928792813,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.4991,
      "step": 4259
    },
    {
      "epoch": 1.3370473537604457,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.4555,
      "step": 4260
    },
    {
      "epoch": 1.33736121464161,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.3808,
      "step": 4261
    },
    {
      "epoch": 1.3376750755227746,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.8497,
      "step": 4262
    },
    {
      "epoch": 1.337988936403939,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.3665,
      "step": 4263
    },
    {
      "epoch": 1.3383027972851034,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.6702,
      "step": 4264
    },
    {
      "epoch": 1.3386166581662677,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.8136,
      "step": 4265
    },
    {
      "epoch": 1.338930519047432,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.9163,
      "step": 4266
    },
    {
      "epoch": 1.3392443799285967,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.6328,
      "step": 4267
    },
    {
      "epoch": 1.339558240809761,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.8287,
      "step": 4268
    },
    {
      "epoch": 1.3398721016909256,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.7631,
      "step": 4269
    },
    {
      "epoch": 1.34018596257209,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.9577,
      "step": 4270
    },
    {
      "epoch": 1.3404998234532544,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.6557,
      "step": 4271
    },
    {
      "epoch": 1.3408136843344187,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.4754,
      "step": 4272
    },
    {
      "epoch": 1.341127545215583,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.9046,
      "step": 4273
    },
    {
      "epoch": 1.3414414060967477,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.872,
      "step": 4274
    },
    {
      "epoch": 1.341755266977912,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.6745,
      "step": 4275
    },
    {
      "epoch": 1.3420691278590764,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.8666,
      "step": 4276
    },
    {
      "epoch": 1.342382988740241,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 2.2783,
      "step": 4277
    },
    {
      "epoch": 1.3426968496214053,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.5121,
      "step": 4278
    },
    {
      "epoch": 1.3430107105025697,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.7287,
      "step": 4279
    },
    {
      "epoch": 1.343324571383734,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.6171,
      "step": 4280
    },
    {
      "epoch": 1.3436384322648987,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.9601,
      "step": 4281
    },
    {
      "epoch": 1.343952293146063,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.6547,
      "step": 4282
    },
    {
      "epoch": 1.3442661540272274,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.9258,
      "step": 4283
    },
    {
      "epoch": 1.344580014908392,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.8691,
      "step": 4284
    },
    {
      "epoch": 1.3448938757895563,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.7277,
      "step": 4285
    },
    {
      "epoch": 1.3452077366707207,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 2.6264,
      "step": 4286
    },
    {
      "epoch": 1.345521597551885,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.6122,
      "step": 4287
    },
    {
      "epoch": 1.3458354584330494,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3902,
      "step": 4288
    },
    {
      "epoch": 1.346149319314214,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.2705,
      "step": 4289
    },
    {
      "epoch": 1.3464631801953784,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.4497,
      "step": 4290
    },
    {
      "epoch": 1.346777041076543,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2441,
      "step": 4291
    },
    {
      "epoch": 1.3470909019577073,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.3874,
      "step": 4292
    },
    {
      "epoch": 1.3474047628388717,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.3759,
      "step": 4293
    },
    {
      "epoch": 1.347718623720036,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.4092,
      "step": 4294
    },
    {
      "epoch": 1.3480324846012004,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.3588,
      "step": 4295
    },
    {
      "epoch": 1.348346345482365,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.5746,
      "step": 4296
    },
    {
      "epoch": 1.3486602063635293,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.3454,
      "step": 4297
    },
    {
      "epoch": 1.3489740672446937,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.4089,
      "step": 4298
    },
    {
      "epoch": 1.3492879281258583,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.51,
      "step": 4299
    },
    {
      "epoch": 1.3496017890070227,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.3044,
      "step": 4300
    },
    {
      "epoch": 1.349915649888187,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.3375,
      "step": 4301
    },
    {
      "epoch": 1.349915649888187,
      "eval_loss": 1.7903716564178467,
      "eval_runtime": 123.6262,
      "eval_samples_per_second": 8.089,
      "eval_steps_per_second": 8.089,
      "step": 4301
    },
    {
      "epoch": 1.349915649888187,
      "mmlu_eval_accuracy": 0.3908505406721757,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.35714285714285715,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3103448275862069,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.0,
      "mmlu_eval_accuracy_college_computer_science": 0.6363636363636364,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.18181818181818182,
      "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": 0.2926829268292683,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.25,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.5833333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5,
      "mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.68,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.5232558139534884,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.36363636363636365,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.36363636363636365,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.1934973221055671,
      "step": 4301
    },
    {
      "epoch": 1.3502295107693514,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.26,
      "step": 4302
    },
    {
      "epoch": 1.350543371650516,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.2986,
      "step": 4303
    },
    {
      "epoch": 1.3508572325316803,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.375,
      "step": 4304
    },
    {
      "epoch": 1.3511710934128447,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.4536,
      "step": 4305
    },
    {
      "epoch": 1.3514849542940093,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.4375,
      "step": 4306
    },
    {
      "epoch": 1.3517988151751736,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.3164,
      "step": 4307
    },
    {
      "epoch": 1.352112676056338,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.6206,
      "step": 4308
    },
    {
      "epoch": 1.3524265369375024,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.388,
      "step": 4309
    },
    {
      "epoch": 1.352740397818667,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.1809,
      "step": 4310
    },
    {
      "epoch": 1.3530542586998313,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.4941,
      "step": 4311
    },
    {
      "epoch": 1.3533681195809957,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.2362,
      "step": 4312
    },
    {
      "epoch": 1.3536819804621603,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.4381,
      "step": 4313
    },
    {
      "epoch": 1.3539958413433246,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.6817,
      "step": 4314
    },
    {
      "epoch": 1.354309702224489,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.4231,
      "step": 4315
    },
    {
      "epoch": 1.3546235631056533,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.6306,
      "step": 4316
    },
    {
      "epoch": 1.3549374239868177,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.8893,
      "step": 4317
    },
    {
      "epoch": 1.3552512848679823,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.791,
      "step": 4318
    },
    {
      "epoch": 1.3555651457491467,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.5052,
      "step": 4319
    },
    {
      "epoch": 1.3558790066303112,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.6438,
      "step": 4320
    },
    {
      "epoch": 1.3561928675114756,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.9806,
      "step": 4321
    },
    {
      "epoch": 1.35650672839264,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 2.0848,
      "step": 4322
    },
    {
      "epoch": 1.3568205892738043,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.9008,
      "step": 4323
    },
    {
      "epoch": 1.3571344501549687,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.3209,
      "step": 4324
    },
    {
      "epoch": 1.3574483110361333,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 2.3895,
      "step": 4325
    },
    {
      "epoch": 1.3577621719172976,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.8199,
      "step": 4326
    },
    {
      "epoch": 1.358076032798462,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.7933,
      "step": 4327
    },
    {
      "epoch": 1.3583898936796266,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.3329,
      "step": 4328
    },
    {
      "epoch": 1.358703754560791,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.1252,
      "step": 4329
    },
    {
      "epoch": 1.3590176154419553,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.8916,
      "step": 4330
    },
    {
      "epoch": 1.3593314763231197,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.862,
      "step": 4331
    },
    {
      "epoch": 1.3596453372042843,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.8936,
      "step": 4332
    },
    {
      "epoch": 1.3599591980854486,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 2.2088,
      "step": 4333
    },
    {
      "epoch": 1.360273058966613,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.6653,
      "step": 4334
    },
    {
      "epoch": 1.3605869198477776,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.849,
      "step": 4335
    },
    {
      "epoch": 1.360900780728942,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 2.4726,
      "step": 4336
    },
    {
      "epoch": 1.3612146416101063,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.5145,
      "step": 4337
    },
    {
      "epoch": 1.3615285024912707,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3967,
      "step": 4338
    },
    {
      "epoch": 1.3618423633724352,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.394,
      "step": 4339
    },
    {
      "epoch": 1.3621562242535996,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.4337,
      "step": 4340
    },
    {
      "epoch": 1.362470085134764,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.5217,
      "step": 4341
    },
    {
      "epoch": 1.3627839460159286,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.3281,
      "step": 4342
    },
    {
      "epoch": 1.363097806897093,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.2755,
      "step": 4343
    },
    {
      "epoch": 1.3634116677782573,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.3414,
      "step": 4344
    },
    {
      "epoch": 1.3637255286594216,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.285,
      "step": 4345
    },
    {
      "epoch": 1.364039389540586,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.3532,
      "step": 4346
    },
    {
      "epoch": 1.3643532504217506,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.3468,
      "step": 4347
    },
    {
      "epoch": 1.364667111302915,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.2533,
      "step": 4348
    },
    {
      "epoch": 1.3649809721840795,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.4034,
      "step": 4349
    },
    {
      "epoch": 1.365294833065244,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.332,
      "step": 4350
    },
    {
      "epoch": 1.3656086939464083,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.3745,
      "step": 4351
    },
    {
      "epoch": 1.3659225548275726,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.3052,
      "step": 4352
    },
    {
      "epoch": 1.366236415708737,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.4312,
      "step": 4353
    },
    {
      "epoch": 1.3665502765899016,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.4191,
      "step": 4354
    },
    {
      "epoch": 1.366864137471066,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.4128,
      "step": 4355
    },
    {
      "epoch": 1.3671779983522303,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.2835,
      "step": 4356
    },
    {
      "epoch": 1.367491859233395,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.4531,
      "step": 4357
    },
    {
      "epoch": 1.3678057201145593,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.4206,
      "step": 4358
    },
    {
      "epoch": 1.3681195809957236,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.4837,
      "step": 4359
    },
    {
      "epoch": 1.368433441876888,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.7016,
      "step": 4360
    },
    {
      "epoch": 1.3687473027580526,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.2667,
      "step": 4361
    },
    {
      "epoch": 1.369061163639217,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.4391,
      "step": 4362
    },
    {
      "epoch": 1.3693750245203813,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.2737,
      "step": 4363
    },
    {
      "epoch": 1.3696888854015459,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.52,
      "step": 4364
    },
    {
      "epoch": 1.3700027462827102,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.5624,
      "step": 4365
    },
    {
      "epoch": 1.3703166071638746,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.5594,
      "step": 4366
    },
    {
      "epoch": 1.370630468045039,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.571,
      "step": 4367
    },
    {
      "epoch": 1.3709443289262033,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.8231,
      "step": 4368
    },
    {
      "epoch": 1.371258189807368,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.6716,
      "step": 4369
    },
    {
      "epoch": 1.3715720506885323,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.8874,
      "step": 4370
    },
    {
      "epoch": 1.3718859115696969,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.8814,
      "step": 4371
    },
    {
      "epoch": 1.3721997724508612,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.0901,
      "step": 4372
    },
    {
      "epoch": 1.3725136333320256,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 2.1328,
      "step": 4373
    },
    {
      "epoch": 1.37282749421319,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.1256,
      "step": 4374
    },
    {
      "epoch": 1.3731413550943543,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.0031,
      "step": 4375
    },
    {
      "epoch": 1.373455215975519,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 1.8179,
      "step": 4376
    },
    {
      "epoch": 1.3737690768566833,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.0201,
      "step": 4377
    },
    {
      "epoch": 1.3740829377378476,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 2.0034,
      "step": 4378
    },
    {
      "epoch": 1.3743967986190122,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.3234,
      "step": 4379
    },
    {
      "epoch": 1.3747106595001766,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.5664,
      "step": 4380
    },
    {
      "epoch": 1.375024520381341,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.8863,
      "step": 4381
    },
    {
      "epoch": 1.3753383812625053,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.3873,
      "step": 4382
    },
    {
      "epoch": 1.3756522421436699,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.6511,
      "step": 4383
    },
    {
      "epoch": 1.3759661030248342,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 2.1379,
      "step": 4384
    },
    {
      "epoch": 1.3762799639059986,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.1523,
      "step": 4385
    },
    {
      "epoch": 1.3765938247871632,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 2.4014,
      "step": 4386
    },
    {
      "epoch": 1.3769076856683276,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.3388,
      "step": 4387
    },
    {
      "epoch": 1.377221546549492,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.2743,
      "step": 4388
    },
    {
      "epoch": 1.3775354074306563,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2867,
      "step": 4389
    },
    {
      "epoch": 1.3778492683118209,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3837,
      "step": 4390
    },
    {
      "epoch": 1.3781631291929852,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3365,
      "step": 4391
    },
    {
      "epoch": 1.3784769900741496,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2984,
      "step": 4392
    },
    {
      "epoch": 1.3787908509553142,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.3257,
      "step": 4393
    },
    {
      "epoch": 1.3791047118364785,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.1303,
      "step": 4394
    },
    {
      "epoch": 1.379418572717643,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.3418,
      "step": 4395
    },
    {
      "epoch": 1.3797324335988073,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.4456,
      "step": 4396
    },
    {
      "epoch": 1.3800462944799716,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2782,
      "step": 4397
    },
    {
      "epoch": 1.3803601553611362,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.2839,
      "step": 4398
    },
    {
      "epoch": 1.3806740162423006,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.4818,
      "step": 4399
    },
    {
      "epoch": 1.3809878771234652,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.3646,
      "step": 4400
    },
    {
      "epoch": 1.3813017380046295,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.3834,
      "step": 4401
    },
    {
      "epoch": 1.3816155988857939,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.2856,
      "step": 4402
    },
    {
      "epoch": 1.3819294597669582,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.336,
      "step": 4403
    },
    {
      "epoch": 1.3822433206481226,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.351,
      "step": 4404
    },
    {
      "epoch": 1.3825571815292872,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.5375,
      "step": 4405
    },
    {
      "epoch": 1.3828710424104516,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.2959,
      "step": 4406
    },
    {
      "epoch": 1.383184903291616,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.2444,
      "step": 4407
    },
    {
      "epoch": 1.3834987641727805,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.275,
      "step": 4408
    },
    {
      "epoch": 1.3838126250539449,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.4684,
      "step": 4409
    },
    {
      "epoch": 1.3841264859351092,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.3096,
      "step": 4410
    },
    {
      "epoch": 1.3844403468162736,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.5674,
      "step": 4411
    },
    {
      "epoch": 1.3847542076974382,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.3748,
      "step": 4412
    },
    {
      "epoch": 1.3850680685786025,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.4567,
      "step": 4413
    },
    {
      "epoch": 1.385381929459767,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.2749,
      "step": 4414
    },
    {
      "epoch": 1.3856957903409315,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.4029,
      "step": 4415
    },
    {
      "epoch": 1.3860096512220959,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.6521,
      "step": 4416
    },
    {
      "epoch": 1.3863235121032602,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.7173,
      "step": 4417
    },
    {
      "epoch": 1.3866373729844246,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.4553,
      "step": 4418
    },
    {
      "epoch": 1.386951233865589,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.8413,
      "step": 4419
    },
    {
      "epoch": 1.3872650947467535,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.807,
      "step": 4420
    },
    {
      "epoch": 1.3875789556279179,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.684,
      "step": 4421
    },
    {
      "epoch": 1.3878928165090825,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.9317,
      "step": 4422
    },
    {
      "epoch": 1.3882066773902468,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.7097,
      "step": 4423
    },
    {
      "epoch": 1.3885205382714112,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.3175,
      "step": 4424
    },
    {
      "epoch": 1.3888343991525756,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.8468,
      "step": 4425
    },
    {
      "epoch": 1.38914826003374,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.85,
      "step": 4426
    },
    {
      "epoch": 1.3894621209149045,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.0743,
      "step": 4427
    },
    {
      "epoch": 1.3897759817960689,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 2.0869,
      "step": 4428
    },
    {
      "epoch": 1.3900898426772332,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.5633,
      "step": 4429
    },
    {
      "epoch": 1.3904037035583978,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.1579,
      "step": 4430
    },
    {
      "epoch": 1.3907175644395622,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.4854,
      "step": 4431
    },
    {
      "epoch": 1.3910314253207265,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.6308,
      "step": 4432
    },
    {
      "epoch": 1.391345286201891,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.7707,
      "step": 4433
    },
    {
      "epoch": 1.3916591470830555,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.8306,
      "step": 4434
    },
    {
      "epoch": 1.3919730079642199,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.0241,
      "step": 4435
    },
    {
      "epoch": 1.3922868688453842,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 2.3202,
      "step": 4436
    },
    {
      "epoch": 1.3926007297265488,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.4694,
      "step": 4437
    },
    {
      "epoch": 1.3929145906077132,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.3261,
      "step": 4438
    },
    {
      "epoch": 1.3932284514888775,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.4208,
      "step": 4439
    },
    {
      "epoch": 1.393542312370042,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4576,
      "step": 4440
    },
    {
      "epoch": 1.3938561732512065,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3911,
      "step": 4441
    },
    {
      "epoch": 1.3941700341323708,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.3504,
      "step": 4442
    },
    {
      "epoch": 1.3944838950135352,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.2102,
      "step": 4443
    },
    {
      "epoch": 1.3947977558946998,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.3065,
      "step": 4444
    },
    {
      "epoch": 1.3951116167758641,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.3616,
      "step": 4445
    },
    {
      "epoch": 1.3954254776570285,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.3182,
      "step": 4446
    },
    {
      "epoch": 1.3957393385381929,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.4715,
      "step": 4447
    },
    {
      "epoch": 1.3960531994193572,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.259,
      "step": 4448
    },
    {
      "epoch": 1.3963670603005218,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.4684,
      "step": 4449
    },
    {
      "epoch": 1.3966809211816862,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.4605,
      "step": 4450
    },
    {
      "epoch": 1.3969947820628508,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.2664,
      "step": 4451
    },
    {
      "epoch": 1.3973086429440151,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.3576,
      "step": 4452
    },
    {
      "epoch": 1.3976225038251795,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.2497,
      "step": 4453
    },
    {
      "epoch": 1.3979363647063439,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.2739,
      "step": 4454
    },
    {
      "epoch": 1.3982502255875082,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.4105,
      "step": 4455
    },
    {
      "epoch": 1.3985640864686728,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.293,
      "step": 4456
    },
    {
      "epoch": 1.3988779473498372,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.4529,
      "step": 4457
    },
    {
      "epoch": 1.3991918082310015,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.2133,
      "step": 4458
    },
    {
      "epoch": 1.3995056691121661,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.559,
      "step": 4459
    },
    {
      "epoch": 1.3998195299933305,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.2974,
      "step": 4460
    },
    {
      "epoch": 1.4001333908744948,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.4939,
      "step": 4461
    },
    {
      "epoch": 1.4004472517556592,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.4051,
      "step": 4462
    },
    {
      "epoch": 1.4007611126368238,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.4685,
      "step": 4463
    },
    {
      "epoch": 1.4010749735179882,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.5184,
      "step": 4464
    },
    {
      "epoch": 1.4013888343991525,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.9703,
      "step": 4465
    },
    {
      "epoch": 1.401702695280317,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.9405,
      "step": 4466
    },
    {
      "epoch": 1.4020165561614815,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 1.9013,
      "step": 4467
    },
    {
      "epoch": 1.4023304170426458,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.8151,
      "step": 4468
    },
    {
      "epoch": 1.4026442779238102,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 2.0511,
      "step": 4469
    },
    {
      "epoch": 1.4029581388049748,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.1633,
      "step": 4470
    },
    {
      "epoch": 1.4032719996861391,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 2.0867,
      "step": 4471
    },
    {
      "epoch": 1.4035858605673035,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.6026,
      "step": 4472
    },
    {
      "epoch": 1.403899721448468,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.9118,
      "step": 4473
    },
    {
      "epoch": 1.4042135823296324,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.4391,
      "step": 4474
    },
    {
      "epoch": 1.4045274432107968,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.8097,
      "step": 4475
    },
    {
      "epoch": 1.4048413040919612,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 2.2578,
      "step": 4476
    },
    {
      "epoch": 1.4051551649731255,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 2.0483,
      "step": 4477
    },
    {
      "epoch": 1.4054690258542901,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002,
      "loss": 2.2424,
      "step": 4478
    },
    {
      "epoch": 1.4057828867354545,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 2.3727,
      "step": 4479
    },
    {
      "epoch": 1.406096747616619,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.8564,
      "step": 4480
    },
    {
      "epoch": 1.4064106084977834,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.9299,
      "step": 4481
    },
    {
      "epoch": 1.4067244693789478,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 2.0507,
      "step": 4482
    },
    {
      "epoch": 1.4070383302601122,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.2124,
      "step": 4483
    },
    {
      "epoch": 1.4073521911412765,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.0349,
      "step": 4484
    },
    {
      "epoch": 1.407666052022441,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 2.1551,
      "step": 4485
    },
    {
      "epoch": 1.4079799129036055,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 2.9484,
      "step": 4486
    },
    {
      "epoch": 1.4082937737847698,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.3464,
      "step": 4487
    },
    {
      "epoch": 1.4086076346659344,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4638,
      "step": 4488
    },
    {
      "epoch": 1.4086076346659344,
      "eval_loss": 1.7950125932693481,
      "eval_runtime": 122.7541,
      "eval_samples_per_second": 8.146,
      "eval_steps_per_second": 8.146,
      "step": 4488
    },
    {
      "epoch": 1.4086076346659344,
      "mmlu_eval_accuracy": 0.41516172277955254,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.18181818181818182,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.3333333333333333,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.6333333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5116279069767442,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.3333333333333333,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.2162521599574778,
      "step": 4488
    },
    {
      "epoch": 1.4089214955470988,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.4172,
      "step": 4489
    },
    {
      "epoch": 1.4092353564282631,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3756,
      "step": 4490
    },
    {
      "epoch": 1.4095492173094275,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3264,
      "step": 4491
    },
    {
      "epoch": 1.409863078190592,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.3612,
      "step": 4492
    },
    {
      "epoch": 1.4101769390717565,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.3923,
      "step": 4493
    },
    {
      "epoch": 1.4104907999529208,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.452,
      "step": 4494
    },
    {
      "epoch": 1.4108046608340854,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2037,
      "step": 4495
    },
    {
      "epoch": 1.4111185217152498,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.4086,
      "step": 4496
    },
    {
      "epoch": 1.4114323825964141,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3997,
      "step": 4497
    },
    {
      "epoch": 1.4117462434775785,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.4006,
      "step": 4498
    },
    {
      "epoch": 1.4120601043587429,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.4248,
      "step": 4499
    },
    {
      "epoch": 1.4123739652399074,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.2983,
      "step": 4500
    },
    {
      "epoch": 1.4126878261210718,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.2411,
      "step": 4501
    },
    {
      "epoch": 1.4130016870022364,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.404,
      "step": 4502
    },
    {
      "epoch": 1.4133155478834007,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.2068,
      "step": 4503
    },
    {
      "epoch": 1.413629408764565,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.4444,
      "step": 4504
    },
    {
      "epoch": 1.4139432696457295,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.3189,
      "step": 4505
    },
    {
      "epoch": 1.4142571305268938,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.4255,
      "step": 4506
    },
    {
      "epoch": 1.4145709914080584,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.2476,
      "step": 4507
    },
    {
      "epoch": 1.4148848522892228,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.5722,
      "step": 4508
    },
    {
      "epoch": 1.4151987131703871,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.182,
      "step": 4509
    },
    {
      "epoch": 1.4155125740515517,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.5858,
      "step": 4510
    },
    {
      "epoch": 1.415826434932716,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.3466,
      "step": 4511
    },
    {
      "epoch": 1.4161402958138805,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.2869,
      "step": 4512
    },
    {
      "epoch": 1.4164541566950448,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.662,
      "step": 4513
    },
    {
      "epoch": 1.4167680175762094,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.4312,
      "step": 4514
    },
    {
      "epoch": 1.4170818784573738,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.5378,
      "step": 4515
    },
    {
      "epoch": 1.4173957393385381,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.405,
      "step": 4516
    },
    {
      "epoch": 1.4177096002197027,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.5854,
      "step": 4517
    },
    {
      "epoch": 1.418023461100867,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.5027,
      "step": 4518
    },
    {
      "epoch": 1.4183373219820314,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.8634,
      "step": 4519
    },
    {
      "epoch": 1.4186511828631958,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.2005,
      "step": 4520
    },
    {
      "epoch": 1.4189650437443604,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.8067,
      "step": 4521
    },
    {
      "epoch": 1.4192789046255248,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 2.3058,
      "step": 4522
    },
    {
      "epoch": 1.4195927655066891,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.7765,
      "step": 4523
    },
    {
      "epoch": 1.4199066263878537,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.8979,
      "step": 4524
    },
    {
      "epoch": 1.420220487269018,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 2.0376,
      "step": 4525
    },
    {
      "epoch": 1.4205343481501824,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.2349,
      "step": 4526
    },
    {
      "epoch": 1.4208482090313468,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.8554,
      "step": 4527
    },
    {
      "epoch": 1.4211620699125111,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 2.308,
      "step": 4528
    },
    {
      "epoch": 1.4214759307936757,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.1704,
      "step": 4529
    },
    {
      "epoch": 1.42178979167484,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 2.1974,
      "step": 4530
    },
    {
      "epoch": 1.4221036525560047,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 2.3392,
      "step": 4531
    },
    {
      "epoch": 1.422417513437169,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 2.5634,
      "step": 4532
    },
    {
      "epoch": 1.4227313743183334,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.7705,
      "step": 4533
    },
    {
      "epoch": 1.4230452351994978,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.6408,
      "step": 4534
    },
    {
      "epoch": 1.4233590960806621,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 2.1068,
      "step": 4535
    },
    {
      "epoch": 1.4236729569618267,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002,
      "loss": 2.5958,
      "step": 4536
    },
    {
      "epoch": 1.423986817842991,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.6286,
      "step": 4537
    },
    {
      "epoch": 1.4243006787241554,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.322,
      "step": 4538
    },
    {
      "epoch": 1.42461453960532,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.6027,
      "step": 4539
    },
    {
      "epoch": 1.4249284004864844,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3281,
      "step": 4540
    },
    {
      "epoch": 1.4252422613676488,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4585,
      "step": 4541
    },
    {
      "epoch": 1.4255561222488131,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3164,
      "step": 4542
    },
    {
      "epoch": 1.4258699831299777,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4013,
      "step": 4543
    },
    {
      "epoch": 1.426183844011142,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.4631,
      "step": 4544
    },
    {
      "epoch": 1.4264977048923064,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.4571,
      "step": 4545
    },
    {
      "epoch": 1.426811565773471,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.2838,
      "step": 4546
    },
    {
      "epoch": 1.4271254266546354,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.2789,
      "step": 4547
    },
    {
      "epoch": 1.4274392875357997,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.4884,
      "step": 4548
    },
    {
      "epoch": 1.427753148416964,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.3723,
      "step": 4549
    },
    {
      "epoch": 1.4280670092981287,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.4656,
      "step": 4550
    },
    {
      "epoch": 1.428380870179293,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.3133,
      "step": 4551
    },
    {
      "epoch": 1.4286947310604574,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.4014,
      "step": 4552
    },
    {
      "epoch": 1.429008591941622,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.455,
      "step": 4553
    },
    {
      "epoch": 1.4293224528227864,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.3893,
      "step": 4554
    },
    {
      "epoch": 1.4296363137039507,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.4411,
      "step": 4555
    },
    {
      "epoch": 1.429950174585115,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.3577,
      "step": 4556
    },
    {
      "epoch": 1.4302640354662794,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2991,
      "step": 4557
    },
    {
      "epoch": 1.430577896347444,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.4277,
      "step": 4558
    },
    {
      "epoch": 1.4308917572286084,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.6938,
      "step": 4559
    },
    {
      "epoch": 1.431205618109773,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.2626,
      "step": 4560
    },
    {
      "epoch": 1.4315194789909373,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.3176,
      "step": 4561
    },
    {
      "epoch": 1.4318333398721017,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.3864,
      "step": 4562
    },
    {
      "epoch": 1.432147200753266,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.3965,
      "step": 4563
    },
    {
      "epoch": 1.4324610616344304,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.2969,
      "step": 4564
    },
    {
      "epoch": 1.432774922515595,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.523,
      "step": 4565
    },
    {
      "epoch": 1.4330887833967594,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.2708,
      "step": 4566
    },
    {
      "epoch": 1.4334026442779237,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.7182,
      "step": 4567
    },
    {
      "epoch": 1.4337165051590883,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.9066,
      "step": 4568
    },
    {
      "epoch": 1.4340303660402527,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.7612,
      "step": 4569
    },
    {
      "epoch": 1.434344226921417,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.8798,
      "step": 4570
    },
    {
      "epoch": 1.4346580878025814,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.8699,
      "step": 4571
    },
    {
      "epoch": 1.434971948683746,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.9965,
      "step": 4572
    },
    {
      "epoch": 1.4352858095649104,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.0068,
      "step": 4573
    },
    {
      "epoch": 1.4355996704460747,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.344,
      "step": 4574
    },
    {
      "epoch": 1.4359135313272393,
      "grad_norm": 0.75,
      "learning_rate": 0.0002,
      "loss": 2.4216,
      "step": 4575
    },
    {
      "epoch": 1.4362273922084037,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.9731,
      "step": 4576
    },
    {
      "epoch": 1.436541253089568,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 2.6413,
      "step": 4577
    },
    {
      "epoch": 1.4368551139707324,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.869,
      "step": 4578
    },
    {
      "epoch": 1.4371689748518968,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.9478,
      "step": 4579
    },
    {
      "epoch": 1.4374828357330613,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.5531,
      "step": 4580
    },
    {
      "epoch": 1.4377966966142257,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.9357,
      "step": 4581
    },
    {
      "epoch": 1.4381105574953903,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.9204,
      "step": 4582
    },
    {
      "epoch": 1.4384244183765547,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.7352,
      "step": 4583
    },
    {
      "epoch": 1.438738279257719,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.7622,
      "step": 4584
    },
    {
      "epoch": 1.4390521401388834,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.2262,
      "step": 4585
    },
    {
      "epoch": 1.4393660010200477,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 2.381,
      "step": 4586
    },
    {
      "epoch": 1.4396798619012123,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.5675,
      "step": 4587
    },
    {
      "epoch": 1.4399937227823767,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3227,
      "step": 4588
    },
    {
      "epoch": 1.440307583663541,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.503,
      "step": 4589
    },
    {
      "epoch": 1.4406214445447056,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4612,
      "step": 4590
    },
    {
      "epoch": 1.44093530542587,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.381,
      "step": 4591
    },
    {
      "epoch": 1.4412491663070344,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.5395,
      "step": 4592
    },
    {
      "epoch": 1.4415630271881987,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.4429,
      "step": 4593
    },
    {
      "epoch": 1.4418768880693633,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.4253,
      "step": 4594
    },
    {
      "epoch": 1.4421907489505277,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.3615,
      "step": 4595
    },
    {
      "epoch": 1.442504609831692,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.1712,
      "step": 4596
    },
    {
      "epoch": 1.4428184707128566,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.3181,
      "step": 4597
    },
    {
      "epoch": 1.443132331594021,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.2507,
      "step": 4598
    },
    {
      "epoch": 1.4434461924751854,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.4418,
      "step": 4599
    },
    {
      "epoch": 1.4437600533563497,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.3556,
      "step": 4600
    },
    {
      "epoch": 1.4440739142375143,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.3598,
      "step": 4601
    },
    {
      "epoch": 1.4443877751186787,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.4568,
      "step": 4602
    },
    {
      "epoch": 1.444701635999843,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.3942,
      "step": 4603
    },
    {
      "epoch": 1.4450154968810076,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.2559,
      "step": 4604
    },
    {
      "epoch": 1.445329357762172,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.3777,
      "step": 4605
    },
    {
      "epoch": 1.4456432186433363,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.3852,
      "step": 4606
    },
    {
      "epoch": 1.4459570795245007,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.4648,
      "step": 4607
    },
    {
      "epoch": 1.446270940405665,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4135,
      "step": 4608
    },
    {
      "epoch": 1.4465848012868296,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.6244,
      "step": 4609
    },
    {
      "epoch": 1.446898662167994,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.3172,
      "step": 4610
    },
    {
      "epoch": 1.4472125230491586,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.6917,
      "step": 4611
    },
    {
      "epoch": 1.447526383930323,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.4561,
      "step": 4612
    },
    {
      "epoch": 1.4478402448114873,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.6565,
      "step": 4613
    },
    {
      "epoch": 1.4481541056926517,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.8942,
      "step": 4614
    },
    {
      "epoch": 1.448467966573816,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.7879,
      "step": 4615
    },
    {
      "epoch": 1.4487818274549806,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.9602,
      "step": 4616
    },
    {
      "epoch": 1.449095688336145,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.8152,
      "step": 4617
    },
    {
      "epoch": 1.4494095492173094,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.0924,
      "step": 4618
    },
    {
      "epoch": 1.449723410098474,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.757,
      "step": 4619
    },
    {
      "epoch": 1.4500372709796383,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.6861,
      "step": 4620
    },
    {
      "epoch": 1.4503511318608027,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.6873,
      "step": 4621
    },
    {
      "epoch": 1.450664992741967,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 2.0753,
      "step": 4622
    },
    {
      "epoch": 1.4509788536231316,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.788,
      "step": 4623
    },
    {
      "epoch": 1.451292714504296,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 2.0385,
      "step": 4624
    },
    {
      "epoch": 1.4516065753854603,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.82,
      "step": 4625
    },
    {
      "epoch": 1.451920436266625,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 2.1059,
      "step": 4626
    },
    {
      "epoch": 1.4522342971477893,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 2.2155,
      "step": 4627
    },
    {
      "epoch": 1.4525481580289537,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.1566,
      "step": 4628
    },
    {
      "epoch": 1.452862018910118,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.7559,
      "step": 4629
    },
    {
      "epoch": 1.4531758797912824,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.2951,
      "step": 4630
    },
    {
      "epoch": 1.453489740672447,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.9572,
      "step": 4631
    },
    {
      "epoch": 1.4538036015536113,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.035,
      "step": 4632
    },
    {
      "epoch": 1.454117462434776,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.622,
      "step": 4633
    },
    {
      "epoch": 1.4544313233159403,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.8148,
      "step": 4634
    },
    {
      "epoch": 1.4547451841971046,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 2.4329,
      "step": 4635
    },
    {
      "epoch": 1.455059045078269,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.4666,
      "step": 4636
    },
    {
      "epoch": 1.4553729059594334,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.5192,
      "step": 4637
    },
    {
      "epoch": 1.455686766840598,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4045,
      "step": 4638
    },
    {
      "epoch": 1.4560006277217623,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.3559,
      "step": 4639
    },
    {
      "epoch": 1.4563144886029267,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3047,
      "step": 4640
    },
    {
      "epoch": 1.4566283494840913,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.4302,
      "step": 4641
    },
    {
      "epoch": 1.4569422103652556,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3603,
      "step": 4642
    },
    {
      "epoch": 1.45725607124642,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.3478,
      "step": 4643
    },
    {
      "epoch": 1.4575699321275843,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.4566,
      "step": 4644
    },
    {
      "epoch": 1.457883793008749,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.4345,
      "step": 4645
    },
    {
      "epoch": 1.4581976538899133,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.3806,
      "step": 4646
    },
    {
      "epoch": 1.4585115147710777,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.4406,
      "step": 4647
    },
    {
      "epoch": 1.4588253756522422,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.3077,
      "step": 4648
    },
    {
      "epoch": 1.4591392365334066,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.4456,
      "step": 4649
    },
    {
      "epoch": 1.459453097414571,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.4145,
      "step": 4650
    },
    {
      "epoch": 1.4597669582957353,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.3238,
      "step": 4651
    },
    {
      "epoch": 1.4600808191769,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.1412,
      "step": 4652
    },
    {
      "epoch": 1.4603946800580643,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.2406,
      "step": 4653
    },
    {
      "epoch": 1.4607085409392286,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.4524,
      "step": 4654
    },
    {
      "epoch": 1.4610224018203932,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.2858,
      "step": 4655
    },
    {
      "epoch": 1.4613362627015576,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2985,
      "step": 4656
    },
    {
      "epoch": 1.461650123582722,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.6286,
      "step": 4657
    },
    {
      "epoch": 1.4619639844638863,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.687,
      "step": 4658
    },
    {
      "epoch": 1.4622778453450507,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.3283,
      "step": 4659
    },
    {
      "epoch": 1.4625917062262153,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.3415,
      "step": 4660
    },
    {
      "epoch": 1.4629055671073796,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.2517,
      "step": 4661
    },
    {
      "epoch": 1.4632194279885442,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.3651,
      "step": 4662
    },
    {
      "epoch": 1.4635332888697086,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.68,
      "step": 4663
    },
    {
      "epoch": 1.463847149750873,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.8188,
      "step": 4664
    },
    {
      "epoch": 1.4641610106320373,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.6716,
      "step": 4665
    },
    {
      "epoch": 1.4644748715132017,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.6055,
      "step": 4666
    },
    {
      "epoch": 1.4647887323943662,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.7878,
      "step": 4667
    },
    {
      "epoch": 1.4651025932755306,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.5551,
      "step": 4668
    },
    {
      "epoch": 1.465416454156695,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.8907,
      "step": 4669
    },
    {
      "epoch": 1.4657303150378596,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.9346,
      "step": 4670
    },
    {
      "epoch": 1.466044175919024,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.6004,
      "step": 4671
    },
    {
      "epoch": 1.4663580368001883,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.8251,
      "step": 4672
    },
    {
      "epoch": 1.4666718976813526,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 2.1775,
      "step": 4673
    },
    {
      "epoch": 1.4669857585625172,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.7053,
      "step": 4674
    },
    {
      "epoch": 1.4672996194436816,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.1193,
      "step": 4675
    },
    {
      "epoch": 1.4672996194436816,
      "eval_loss": 1.7845886945724487,
      "eval_runtime": 123.49,
      "eval_samples_per_second": 8.098,
      "eval_steps_per_second": 8.098,
      "step": 4675
    },
    {
      "epoch": 1.4672996194436816,
      "mmlu_eval_accuracy": 0.39873369989462604,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3448275862068966,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.22727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.18181818181818182,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.25,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.3333333333333333,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.55,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.4230769230769231,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.68,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5232558139534884,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.48484848484848486,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.2857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.29411764705882354,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
      "mmlu_loss": 1.18201911997398,
      "step": 4675
    },
    {
      "epoch": 1.467613480324846,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.9196,
      "step": 4676
    },
    {
      "epoch": 1.4679273412060105,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.9435,
      "step": 4677
    },
    {
      "epoch": 1.468241202087175,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.6398,
      "step": 4678
    },
    {
      "epoch": 1.4685550629683393,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.7233,
      "step": 4679
    },
    {
      "epoch": 1.4688689238495036,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002,
      "loss": 2.2635,
      "step": 4680
    },
    {
      "epoch": 1.4691827847306682,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002,
      "loss": 1.9,
      "step": 4681
    },
    {
      "epoch": 1.4694966456118326,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.0136,
      "step": 4682
    },
    {
      "epoch": 1.469810506492997,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.9494,
      "step": 4683
    },
    {
      "epoch": 1.4701243673741615,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.9991,
      "step": 4684
    },
    {
      "epoch": 1.4704382282553259,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.1223,
      "step": 4685
    },
    {
      "epoch": 1.4707520891364902,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0002,
      "loss": 2.8974,
      "step": 4686
    },
    {
      "epoch": 1.4710659500176546,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.4895,
      "step": 4687
    },
    {
      "epoch": 1.471379810898819,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.2938,
      "step": 4688
    },
    {
      "epoch": 1.4716936717799836,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.5057,
      "step": 4689
    },
    {
      "epoch": 1.472007532661148,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.4302,
      "step": 4690
    },
    {
      "epoch": 1.4723213935423125,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.4823,
      "step": 4691
    },
    {
      "epoch": 1.4726352544234769,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.4628,
      "step": 4692
    },
    {
      "epoch": 1.4729491153046412,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.4033,
      "step": 4693
    },
    {
      "epoch": 1.4732629761858056,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.2811,
      "step": 4694
    },
    {
      "epoch": 1.47357683706697,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.2457,
      "step": 4695
    },
    {
      "epoch": 1.4738906979481345,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.3602,
      "step": 4696
    },
    {
      "epoch": 1.474204558829299,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.427,
      "step": 4697
    },
    {
      "epoch": 1.4745184197104633,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.2397,
      "step": 4698
    },
    {
      "epoch": 1.4748322805916279,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.2726,
      "step": 4699
    },
    {
      "epoch": 1.4751461414727922,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.5702,
      "step": 4700
    },
    {
      "epoch": 1.4754600023539566,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.4863,
      "step": 4701
    },
    {
      "epoch": 1.475773863235121,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.6425,
      "step": 4702
    },
    {
      "epoch": 1.4760877241162855,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.2094,
      "step": 4703
    },
    {
      "epoch": 1.47640158499745,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.3114,
      "step": 4704
    },
    {
      "epoch": 1.4767154458786143,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.3582,
      "step": 4705
    },
    {
      "epoch": 1.4770293067597788,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.367,
      "step": 4706
    },
    {
      "epoch": 1.4773431676409432,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.264,
      "step": 4707
    },
    {
      "epoch": 1.4776570285221076,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.5233,
      "step": 4708
    },
    {
      "epoch": 1.477970889403272,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.388,
      "step": 4709
    },
    {
      "epoch": 1.4782847502844363,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.4282,
      "step": 4710
    },
    {
      "epoch": 1.4785986111656009,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.3807,
      "step": 4711
    },
    {
      "epoch": 1.4789124720467652,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.6471,
      "step": 4712
    },
    {
      "epoch": 1.4792263329279298,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.6978,
      "step": 4713
    },
    {
      "epoch": 1.4795401938090942,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.7307,
      "step": 4714
    },
    {
      "epoch": 1.4798540546902585,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.4917,
      "step": 4715
    },
    {
      "epoch": 1.480167915571423,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.6797,
      "step": 4716
    },
    {
      "epoch": 1.4804817764525873,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.882,
      "step": 4717
    },
    {
      "epoch": 1.4807956373337519,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.7856,
      "step": 4718
    },
    {
      "epoch": 1.4811094982149162,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.4896,
      "step": 4719
    },
    {
      "epoch": 1.4814233590960806,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.7564,
      "step": 4720
    },
    {
      "epoch": 1.4817372199772452,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.6041,
      "step": 4721
    },
    {
      "epoch": 1.4820510808584095,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.6861,
      "step": 4722
    },
    {
      "epoch": 1.482364941739574,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.6922,
      "step": 4723
    },
    {
      "epoch": 1.4826788026207383,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 2.1272,
      "step": 4724
    },
    {
      "epoch": 1.4829926635019028,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.8977,
      "step": 4725
    },
    {
      "epoch": 1.4833065243830672,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 2.0812,
      "step": 4726
    },
    {
      "epoch": 1.4836203852642316,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.1398,
      "step": 4727
    },
    {
      "epoch": 1.4839342461453962,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 2.179,
      "step": 4728
    },
    {
      "epoch": 1.4842481070265605,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.8119,
      "step": 4729
    },
    {
      "epoch": 1.4845619679077249,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 2.6156,
      "step": 4730
    },
    {
      "epoch": 1.4848758287888892,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.9482,
      "step": 4731
    },
    {
      "epoch": 1.4851896896700538,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.5828,
      "step": 4732
    },
    {
      "epoch": 1.4855035505512182,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.5911,
      "step": 4733
    },
    {
      "epoch": 1.4858174114323826,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.7881,
      "step": 4734
    },
    {
      "epoch": 1.4861312723135471,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.8778,
      "step": 4735
    },
    {
      "epoch": 1.4864451331947115,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 2.6713,
      "step": 4736
    },
    {
      "epoch": 1.4867589940758759,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.5121,
      "step": 4737
    },
    {
      "epoch": 1.4870728549570402,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3319,
      "step": 4738
    },
    {
      "epoch": 1.4873867158382046,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.267,
      "step": 4739
    },
    {
      "epoch": 1.4877005767193692,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.5039,
      "step": 4740
    },
    {
      "epoch": 1.4880144376005335,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.1809,
      "step": 4741
    },
    {
      "epoch": 1.4883282984816981,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.2295,
      "step": 4742
    },
    {
      "epoch": 1.4886421593628625,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3372,
      "step": 4743
    },
    {
      "epoch": 1.4889560202440268,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2749,
      "step": 4744
    },
    {
      "epoch": 1.4892698811251912,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.1847,
      "step": 4745
    },
    {
      "epoch": 1.4895837420063556,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.4192,
      "step": 4746
    },
    {
      "epoch": 1.4898976028875202,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.283,
      "step": 4747
    },
    {
      "epoch": 1.4902114637686845,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.4341,
      "step": 4748
    },
    {
      "epoch": 1.4905253246498489,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.3149,
      "step": 4749
    },
    {
      "epoch": 1.4908391855310135,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.4504,
      "step": 4750
    },
    {
      "epoch": 1.4911530464121778,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.2171,
      "step": 4751
    },
    {
      "epoch": 1.4914669072933422,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.3384,
      "step": 4752
    },
    {
      "epoch": 1.4917807681745066,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.4283,
      "step": 4753
    },
    {
      "epoch": 1.4920946290556711,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.3798,
      "step": 4754
    },
    {
      "epoch": 1.4924084899368355,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.5737,
      "step": 4755
    },
    {
      "epoch": 1.4927223508179999,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.5947,
      "step": 4756
    },
    {
      "epoch": 1.4930362116991645,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4335,
      "step": 4757
    },
    {
      "epoch": 1.4933500725803288,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.3925,
      "step": 4758
    },
    {
      "epoch": 1.4936639334614932,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.2464,
      "step": 4759
    },
    {
      "epoch": 1.4939777943426575,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.5044,
      "step": 4760
    },
    {
      "epoch": 1.4942916552238221,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.5591,
      "step": 4761
    },
    {
      "epoch": 1.4946055161049865,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.8971,
      "step": 4762
    },
    {
      "epoch": 1.4949193769861508,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.3473,
      "step": 4763
    },
    {
      "epoch": 1.4952332378673154,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.2611,
      "step": 4764
    },
    {
      "epoch": 1.4955470987484798,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.7825,
      "step": 4765
    },
    {
      "epoch": 1.4958609596296442,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.8171,
      "step": 4766
    },
    {
      "epoch": 1.4961748205108085,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.7894,
      "step": 4767
    },
    {
      "epoch": 1.4964886813919729,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.8907,
      "step": 4768
    },
    {
      "epoch": 1.4968025422731375,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 2.2738,
      "step": 4769
    },
    {
      "epoch": 1.4971164031543018,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.6945,
      "step": 4770
    },
    {
      "epoch": 1.4974302640354662,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 2.0429,
      "step": 4771
    },
    {
      "epoch": 1.4977441249166308,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 2.356,
      "step": 4772
    },
    {
      "epoch": 1.4980579857977951,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.576,
      "step": 4773
    },
    {
      "epoch": 1.4983718466789595,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.0138,
      "step": 4774
    },
    {
      "epoch": 1.4986857075601239,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.8323,
      "step": 4775
    },
    {
      "epoch": 1.4989995684412885,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.9431,
      "step": 4776
    },
    {
      "epoch": 1.4993134293224528,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 2.0335,
      "step": 4777
    },
    {
      "epoch": 1.4996272902036172,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 2.0601,
      "step": 4778
    },
    {
      "epoch": 1.4999411510847818,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.2626,
      "step": 4779
    },
    {
      "epoch": 1.5002550119659461,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.7814,
      "step": 4780
    },
    {
      "epoch": 1.5005688728471105,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 2.3872,
      "step": 4781
    },
    {
      "epoch": 1.5008827337282749,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.5471,
      "step": 4782
    },
    {
      "epoch": 1.5011965946094392,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.8402,
      "step": 4783
    },
    {
      "epoch": 1.5015104554906038,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 2.095,
      "step": 4784
    },
    {
      "epoch": 1.5018243163717682,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 2.2912,
      "step": 4785
    },
    {
      "epoch": 1.5021381772529327,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002,
      "loss": 2.6753,
      "step": 4786
    },
    {
      "epoch": 1.5024520381340971,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.626,
      "step": 4787
    },
    {
      "epoch": 1.5027658990152615,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3211,
      "step": 4788
    },
    {
      "epoch": 1.5030797598964258,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3371,
      "step": 4789
    },
    {
      "epoch": 1.5033936207775902,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.4466,
      "step": 4790
    },
    {
      "epoch": 1.5037074816587548,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.4366,
      "step": 4791
    },
    {
      "epoch": 1.5040213425399191,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.3662,
      "step": 4792
    },
    {
      "epoch": 1.5043352034210837,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.5024,
      "step": 4793
    },
    {
      "epoch": 1.504649064302248,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3876,
      "step": 4794
    },
    {
      "epoch": 1.5049629251834125,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.3244,
      "step": 4795
    },
    {
      "epoch": 1.5052767860645768,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.2709,
      "step": 4796
    },
    {
      "epoch": 1.5055906469457412,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.2322,
      "step": 4797
    },
    {
      "epoch": 1.5059045078269058,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.4231,
      "step": 4798
    },
    {
      "epoch": 1.5062183687080701,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.3405,
      "step": 4799
    },
    {
      "epoch": 1.5065322295892347,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.2256,
      "step": 4800
    },
    {
      "epoch": 1.506846090470399,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.3446,
      "step": 4801
    },
    {
      "epoch": 1.5071599513515634,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.4563,
      "step": 4802
    },
    {
      "epoch": 1.5074738122327278,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.1671,
      "step": 4803
    },
    {
      "epoch": 1.5077876731138922,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.3043,
      "step": 4804
    },
    {
      "epoch": 1.5081015339950565,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.5666,
      "step": 4805
    },
    {
      "epoch": 1.5084153948762211,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.4813,
      "step": 4806
    },
    {
      "epoch": 1.5087292557573857,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.3034,
      "step": 4807
    },
    {
      "epoch": 1.50904311663855,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.4337,
      "step": 4808
    },
    {
      "epoch": 1.5093569775197144,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.5607,
      "step": 4809
    },
    {
      "epoch": 1.5096708384008788,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.5443,
      "step": 4810
    },
    {
      "epoch": 1.5099846992820432,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.595,
      "step": 4811
    },
    {
      "epoch": 1.5102985601632075,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.5278,
      "step": 4812
    },
    {
      "epoch": 1.510612421044372,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.5109,
      "step": 4813
    },
    {
      "epoch": 1.5109262819255365,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.5616,
      "step": 4814
    },
    {
      "epoch": 1.511240142806701,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.3786,
      "step": 4815
    },
    {
      "epoch": 1.5115540036878654,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.815,
      "step": 4816
    },
    {
      "epoch": 1.5118678645690298,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.4484,
      "step": 4817
    },
    {
      "epoch": 1.5121817254501941,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.7374,
      "step": 4818
    },
    {
      "epoch": 1.5124955863313585,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.615,
      "step": 4819
    },
    {
      "epoch": 1.512809447212523,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.6979,
      "step": 4820
    },
    {
      "epoch": 1.5131233080936874,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 2.2839,
      "step": 4821
    },
    {
      "epoch": 1.513437168974852,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 2.0101,
      "step": 4822
    },
    {
      "epoch": 1.5137510298560164,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.6714,
      "step": 4823
    },
    {
      "epoch": 1.5140648907371808,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.9403,
      "step": 4824
    },
    {
      "epoch": 1.5143787516183451,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.8463,
      "step": 4825
    },
    {
      "epoch": 1.5146926124995095,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.6704,
      "step": 4826
    },
    {
      "epoch": 1.515006473380674,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.7315,
      "step": 4827
    },
    {
      "epoch": 1.5153203342618384,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.6438,
      "step": 4828
    },
    {
      "epoch": 1.515634195143003,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.3559,
      "step": 4829
    },
    {
      "epoch": 1.5159480560241674,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.0338,
      "step": 4830
    },
    {
      "epoch": 1.5162619169053317,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.916,
      "step": 4831
    },
    {
      "epoch": 1.516575777786496,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.73,
      "step": 4832
    },
    {
      "epoch": 1.5168896386676605,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.8197,
      "step": 4833
    },
    {
      "epoch": 1.5172034995488248,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.7585,
      "step": 4834
    },
    {
      "epoch": 1.5175173604299894,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 1.8572,
      "step": 4835
    },
    {
      "epoch": 1.5178312213111538,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 2.4985,
      "step": 4836
    },
    {
      "epoch": 1.5181450821923184,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.5137,
      "step": 4837
    },
    {
      "epoch": 1.5184589430734827,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2961,
      "step": 4838
    },
    {
      "epoch": 1.518772803954647,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.4312,
      "step": 4839
    },
    {
      "epoch": 1.5190866648358115,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.4462,
      "step": 4840
    },
    {
      "epoch": 1.5194005257169758,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.3613,
      "step": 4841
    },
    {
      "epoch": 1.5197143865981404,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3979,
      "step": 4842
    },
    {
      "epoch": 1.5200282474793048,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.4604,
      "step": 4843
    },
    {
      "epoch": 1.5203421083604693,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.3047,
      "step": 4844
    },
    {
      "epoch": 1.5206559692416337,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.2199,
      "step": 4845
    },
    {
      "epoch": 1.520969830122798,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.489,
      "step": 4846
    },
    {
      "epoch": 1.5212836910039624,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.501,
      "step": 4847
    },
    {
      "epoch": 1.5215975518851268,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.4074,
      "step": 4848
    },
    {
      "epoch": 1.5219114127662914,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.4732,
      "step": 4849
    },
    {
      "epoch": 1.5222252736474557,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.2671,
      "step": 4850
    },
    {
      "epoch": 1.5225391345286203,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.2669,
      "step": 4851
    },
    {
      "epoch": 1.5228529954097847,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.3934,
      "step": 4852
    },
    {
      "epoch": 1.523166856290949,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.3603,
      "step": 4853
    },
    {
      "epoch": 1.5234807171721134,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.3657,
      "step": 4854
    },
    {
      "epoch": 1.5237945780532778,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3856,
      "step": 4855
    },
    {
      "epoch": 1.5241084389344421,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.3651,
      "step": 4856
    },
    {
      "epoch": 1.5244222998156067,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.4138,
      "step": 4857
    },
    {
      "epoch": 1.5247361606967713,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.5296,
      "step": 4858
    },
    {
      "epoch": 1.5250500215779357,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.4891,
      "step": 4859
    },
    {
      "epoch": 1.5253638824591,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.4156,
      "step": 4860
    },
    {
      "epoch": 1.5256777433402644,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.4966,
      "step": 4861
    },
    {
      "epoch": 1.5259916042214288,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.6988,
      "step": 4862
    },
    {
      "epoch": 1.5259916042214288,
      "eval_loss": 1.7795552015304565,
      "eval_runtime": 122.9144,
      "eval_samples_per_second": 8.136,
      "eval_steps_per_second": 8.136,
      "step": 4862
    },
    {
      "epoch": 1.5259916042214288,
      "mmlu_eval_accuracy": 0.40338534054114106,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.35714285714285715,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.2727272727272727,
      "mmlu_eval_accuracy_computer_security": 0.18181818181818182,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.21875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.5666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.38461538461538464,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.25,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5813953488372093,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.42424242424242425,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.2857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.29411764705882354,
      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.148823298742614,
      "step": 4862
    },
    {
      "epoch": 1.5263054651025931,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.4722,
      "step": 4863
    },
    {
      "epoch": 1.5266193259837577,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.3712,
      "step": 4864
    },
    {
      "epoch": 1.526933186864922,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.6423,
      "step": 4865
    },
    {
      "epoch": 1.5272470477460867,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.7239,
      "step": 4866
    },
    {
      "epoch": 1.527560908627251,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.5874,
      "step": 4867
    },
    {
      "epoch": 1.5278747695084154,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.0921,
      "step": 4868
    },
    {
      "epoch": 1.5281886303895797,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.8612,
      "step": 4869
    },
    {
      "epoch": 1.5285024912707441,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.812,
      "step": 4870
    },
    {
      "epoch": 1.5288163521519087,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 2.0088,
      "step": 4871
    },
    {
      "epoch": 1.529130213033073,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.0431,
      "step": 4872
    },
    {
      "epoch": 1.5294440739142376,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.3642,
      "step": 4873
    },
    {
      "epoch": 1.529757934795402,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.7782,
      "step": 4874
    },
    {
      "epoch": 1.5300717956765664,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.857,
      "step": 4875
    },
    {
      "epoch": 1.5303856565577307,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.7847,
      "step": 4876
    },
    {
      "epoch": 1.530699517438895,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.0318,
      "step": 4877
    },
    {
      "epoch": 1.5310133783200597,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.0986,
      "step": 4878
    },
    {
      "epoch": 1.531327239201224,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.8357,
      "step": 4879
    },
    {
      "epoch": 1.5316411000823886,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 2.2329,
      "step": 4880
    },
    {
      "epoch": 1.531954960963553,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.9576,
      "step": 4881
    },
    {
      "epoch": 1.5322688218447174,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.7508,
      "step": 4882
    },
    {
      "epoch": 1.5325826827258817,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 2.1047,
      "step": 4883
    },
    {
      "epoch": 1.532896543607046,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 2.0002,
      "step": 4884
    },
    {
      "epoch": 1.5332104044882104,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.0798,
      "step": 4885
    },
    {
      "epoch": 1.533524265369375,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 2.3123,
      "step": 4886
    },
    {
      "epoch": 1.5338381262505396,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.4414,
      "step": 4887
    },
    {
      "epoch": 1.534151987131704,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3801,
      "step": 4888
    },
    {
      "epoch": 1.5344658480128683,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3742,
      "step": 4889
    },
    {
      "epoch": 1.5347797088940327,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.2834,
      "step": 4890
    },
    {
      "epoch": 1.535093569775197,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.4661,
      "step": 4891
    },
    {
      "epoch": 1.5354074306563614,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.4743,
      "step": 4892
    },
    {
      "epoch": 1.535721291537526,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.3571,
      "step": 4893
    },
    {
      "epoch": 1.5360351524186904,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.3627,
      "step": 4894
    },
    {
      "epoch": 1.536349013299855,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.2071,
      "step": 4895
    },
    {
      "epoch": 1.5366628741810193,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.279,
      "step": 4896
    },
    {
      "epoch": 1.5369767350621837,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.3222,
      "step": 4897
    },
    {
      "epoch": 1.537290595943348,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.2858,
      "step": 4898
    },
    {
      "epoch": 1.5376044568245124,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.4648,
      "step": 4899
    },
    {
      "epoch": 1.537918317705677,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2091,
      "step": 4900
    },
    {
      "epoch": 1.5382321785868414,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.3099,
      "step": 4901
    },
    {
      "epoch": 1.538546039468006,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.2137,
      "step": 4902
    },
    {
      "epoch": 1.5388599003491703,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.294,
      "step": 4903
    },
    {
      "epoch": 1.5391737612303347,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.459,
      "step": 4904
    },
    {
      "epoch": 1.539487622111499,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.3275,
      "step": 4905
    },
    {
      "epoch": 1.5398014829926634,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.2145,
      "step": 4906
    },
    {
      "epoch": 1.540115343873828,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.2768,
      "step": 4907
    },
    {
      "epoch": 1.5404292047549923,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.3009,
      "step": 4908
    },
    {
      "epoch": 1.540743065636157,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.3927,
      "step": 4909
    },
    {
      "epoch": 1.5410569265173213,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.4026,
      "step": 4910
    },
    {
      "epoch": 1.5413707873984857,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.4493,
      "step": 4911
    },
    {
      "epoch": 1.54168464827965,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.5503,
      "step": 4912
    },
    {
      "epoch": 1.5419985091608144,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.3989,
      "step": 4913
    },
    {
      "epoch": 1.5423123700419787,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.4274,
      "step": 4914
    },
    {
      "epoch": 1.5426262309231433,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.3415,
      "step": 4915
    },
    {
      "epoch": 1.5429400918043077,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.4149,
      "step": 4916
    },
    {
      "epoch": 1.5432539526854723,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.6919,
      "step": 4917
    },
    {
      "epoch": 1.5435678135666366,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.7856,
      "step": 4918
    },
    {
      "epoch": 1.543881674447801,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.6884,
      "step": 4919
    },
    {
      "epoch": 1.5441955353289654,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.724,
      "step": 4920
    },
    {
      "epoch": 1.5445093962101297,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.9534,
      "step": 4921
    },
    {
      "epoch": 1.5448232570912943,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.2285,
      "step": 4922
    },
    {
      "epoch": 1.5451371179724587,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.8947,
      "step": 4923
    },
    {
      "epoch": 1.5454509788536233,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.6949,
      "step": 4924
    },
    {
      "epoch": 1.5457648397347876,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 2.0834,
      "step": 4925
    },
    {
      "epoch": 1.546078700615952,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 2.296,
      "step": 4926
    },
    {
      "epoch": 1.5463925614971163,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.245,
      "step": 4927
    },
    {
      "epoch": 1.5467064223782807,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.1643,
      "step": 4928
    },
    {
      "epoch": 1.5470202832594453,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 2.3892,
      "step": 4929
    },
    {
      "epoch": 1.5473341441406097,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 2.2535,
      "step": 4930
    },
    {
      "epoch": 1.5476480050217742,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 2.2135,
      "step": 4931
    },
    {
      "epoch": 1.5479618659029386,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.8875,
      "step": 4932
    },
    {
      "epoch": 1.548275726784103,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.284,
      "step": 4933
    },
    {
      "epoch": 1.5485895876652673,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.9798,
      "step": 4934
    },
    {
      "epoch": 1.5489034485464317,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 2.0749,
      "step": 4935
    },
    {
      "epoch": 1.549217309427596,
      "grad_norm": 0.859375,
      "learning_rate": 0.0002,
      "loss": 2.5011,
      "step": 4936
    },
    {
      "epoch": 1.5495311703087606,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.4444,
      "step": 4937
    },
    {
      "epoch": 1.5498450311899252,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.3725,
      "step": 4938
    },
    {
      "epoch": 1.5501588920710896,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.136,
      "step": 4939
    },
    {
      "epoch": 1.550472752952254,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4387,
      "step": 4940
    },
    {
      "epoch": 1.5507866138334183,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3535,
      "step": 4941
    },
    {
      "epoch": 1.5511004747145827,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2157,
      "step": 4942
    },
    {
      "epoch": 1.551414335595747,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.4201,
      "step": 4943
    },
    {
      "epoch": 1.5517281964769116,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.4822,
      "step": 4944
    },
    {
      "epoch": 1.552042057358076,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.3504,
      "step": 4945
    },
    {
      "epoch": 1.5523559182392406,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.2893,
      "step": 4946
    },
    {
      "epoch": 1.552669779120405,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.2906,
      "step": 4947
    },
    {
      "epoch": 1.5529836400015693,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.312,
      "step": 4948
    },
    {
      "epoch": 1.5532975008827337,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.3696,
      "step": 4949
    },
    {
      "epoch": 1.553611361763898,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.4176,
      "step": 4950
    },
    {
      "epoch": 1.5539252226450626,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.4041,
      "step": 4951
    },
    {
      "epoch": 1.554239083526227,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.3403,
      "step": 4952
    },
    {
      "epoch": 1.5545529444073916,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.2961,
      "step": 4953
    },
    {
      "epoch": 1.554866805288556,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.3889,
      "step": 4954
    },
    {
      "epoch": 1.5551806661697203,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.5129,
      "step": 4955
    },
    {
      "epoch": 1.5554945270508846,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.2571,
      "step": 4956
    },
    {
      "epoch": 1.555808387932049,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.3474,
      "step": 4957
    },
    {
      "epoch": 1.5561222488132136,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.3583,
      "step": 4958
    },
    {
      "epoch": 1.556436109694378,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.4329,
      "step": 4959
    },
    {
      "epoch": 1.5567499705755425,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.4391,
      "step": 4960
    },
    {
      "epoch": 1.557063831456707,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.4708,
      "step": 4961
    },
    {
      "epoch": 1.5573776923378713,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4421,
      "step": 4962
    },
    {
      "epoch": 1.5576915532190356,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.4067,
      "step": 4963
    },
    {
      "epoch": 1.5580054141002,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.7318,
      "step": 4964
    },
    {
      "epoch": 1.5583192749813644,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.8052,
      "step": 4965
    },
    {
      "epoch": 1.558633135862529,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.6521,
      "step": 4966
    },
    {
      "epoch": 1.5589469967436933,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.8888,
      "step": 4967
    },
    {
      "epoch": 1.5592608576248579,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.9253,
      "step": 4968
    },
    {
      "epoch": 1.5595747185060223,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.6776,
      "step": 4969
    },
    {
      "epoch": 1.5598885793871866,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.9312,
      "step": 4970
    },
    {
      "epoch": 1.560202440268351,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.9008,
      "step": 4971
    },
    {
      "epoch": 1.5605163011495153,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.8292,
      "step": 4972
    },
    {
      "epoch": 1.56083016203068,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.8828,
      "step": 4973
    },
    {
      "epoch": 1.5611440229118443,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 2.0352,
      "step": 4974
    },
    {
      "epoch": 1.5614578837930089,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.9757,
      "step": 4975
    },
    {
      "epoch": 1.5617717446741732,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.2297,
      "step": 4976
    },
    {
      "epoch": 1.5620856055553376,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.9935,
      "step": 4977
    },
    {
      "epoch": 1.562399466436502,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 1.9956,
      "step": 4978
    },
    {
      "epoch": 1.5627133273176663,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 2.1654,
      "step": 4979
    },
    {
      "epoch": 1.563027188198831,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.2317,
      "step": 4980
    },
    {
      "epoch": 1.5633410490799953,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.6525,
      "step": 4981
    },
    {
      "epoch": 1.5636549099611599,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.9219,
      "step": 4982
    },
    {
      "epoch": 1.5639687708423242,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.8591,
      "step": 4983
    },
    {
      "epoch": 1.5642826317234886,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 2.1601,
      "step": 4984
    },
    {
      "epoch": 1.564596492604653,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.0061,
      "step": 4985
    },
    {
      "epoch": 1.5649103534858173,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 2.063,
      "step": 4986
    },
    {
      "epoch": 1.5652242143669817,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.7017,
      "step": 4987
    },
    {
      "epoch": 1.5655380752481463,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3287,
      "step": 4988
    },
    {
      "epoch": 1.5658519361293108,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2995,
      "step": 4989
    },
    {
      "epoch": 1.5661657970104752,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.3732,
      "step": 4990
    },
    {
      "epoch": 1.5664796578916396,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3189,
      "step": 4991
    },
    {
      "epoch": 1.566793518772804,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.4269,
      "step": 4992
    },
    {
      "epoch": 1.5671073796539683,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3134,
      "step": 4993
    },
    {
      "epoch": 1.5674212405351327,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.3434,
      "step": 4994
    },
    {
      "epoch": 1.5677351014162972,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.3721,
      "step": 4995
    },
    {
      "epoch": 1.5680489622974616,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.3956,
      "step": 4996
    },
    {
      "epoch": 1.5683628231786262,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.4549,
      "step": 4997
    },
    {
      "epoch": 1.5686766840597905,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.2518,
      "step": 4998
    },
    {
      "epoch": 1.568990544940955,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3455,
      "step": 4999
    },
    {
      "epoch": 1.5693044058221193,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.3074,
      "step": 5000
    },
    {
      "epoch": 1.5696182667032836,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.2701,
      "step": 5001
    },
    {
      "epoch": 1.5699321275844482,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.242,
      "step": 5002
    },
    {
      "epoch": 1.5702459884656126,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.2864,
      "step": 5003
    },
    {
      "epoch": 1.5705598493467772,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.3041,
      "step": 5004
    },
    {
      "epoch": 1.5708737102279415,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.3437,
      "step": 5005
    },
    {
      "epoch": 1.571187571109106,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.3458,
      "step": 5006
    },
    {
      "epoch": 1.5715014319902703,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.4165,
      "step": 5007
    },
    {
      "epoch": 1.5718152928714346,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.3758,
      "step": 5008
    },
    {
      "epoch": 1.5721291537525992,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.5677,
      "step": 5009
    },
    {
      "epoch": 1.5724430146337636,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.3738,
      "step": 5010
    },
    {
      "epoch": 1.5727568755149282,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.1974,
      "step": 5011
    },
    {
      "epoch": 1.5730707363960925,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.5345,
      "step": 5012
    },
    {
      "epoch": 1.5733845972772569,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.4702,
      "step": 5013
    },
    {
      "epoch": 1.5736984581584212,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.5037,
      "step": 5014
    },
    {
      "epoch": 1.5740123190395856,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.3174,
      "step": 5015
    },
    {
      "epoch": 1.57432617992075,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.7725,
      "step": 5016
    },
    {
      "epoch": 1.5746400408019146,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.7217,
      "step": 5017
    },
    {
      "epoch": 1.5749539016830791,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.1253,
      "step": 5018
    },
    {
      "epoch": 1.5752677625642435,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.7848,
      "step": 5019
    },
    {
      "epoch": 1.5755816234454079,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.8598,
      "step": 5020
    },
    {
      "epoch": 1.5758954843265722,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.9197,
      "step": 5021
    },
    {
      "epoch": 1.5762093452077366,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 1.8669,
      "step": 5022
    },
    {
      "epoch": 1.576523206088901,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.5558,
      "step": 5023
    },
    {
      "epoch": 1.5768370669700655,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 2.0621,
      "step": 5024
    },
    {
      "epoch": 1.57715092785123,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.0386,
      "step": 5025
    },
    {
      "epoch": 1.5774647887323945,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.8108,
      "step": 5026
    },
    {
      "epoch": 1.5777786496135588,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.9032,
      "step": 5027
    },
    {
      "epoch": 1.5780925104947232,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 2.4514,
      "step": 5028
    },
    {
      "epoch": 1.5784063713758876,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.7561,
      "step": 5029
    },
    {
      "epoch": 1.578720232257052,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.263,
      "step": 5030
    },
    {
      "epoch": 1.5790340931382165,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.9355,
      "step": 5031
    },
    {
      "epoch": 1.5793479540193809,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.7209,
      "step": 5032
    },
    {
      "epoch": 1.5796618149005455,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.574,
      "step": 5033
    },
    {
      "epoch": 1.5799756757817098,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.972,
      "step": 5034
    },
    {
      "epoch": 1.5802895366628742,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 2.0734,
      "step": 5035
    },
    {
      "epoch": 1.5806033975440386,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 2.6693,
      "step": 5036
    },
    {
      "epoch": 1.580917258425203,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.3638,
      "step": 5037
    },
    {
      "epoch": 1.5812311193063675,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.2417,
      "step": 5038
    },
    {
      "epoch": 1.5815449801875319,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2349,
      "step": 5039
    },
    {
      "epoch": 1.5818588410686965,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3286,
      "step": 5040
    },
    {
      "epoch": 1.5821727019498608,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.3434,
      "step": 5041
    },
    {
      "epoch": 1.5824865628310252,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4358,
      "step": 5042
    },
    {
      "epoch": 1.5828004237121895,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.3198,
      "step": 5043
    },
    {
      "epoch": 1.583114284593354,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.3207,
      "step": 5044
    },
    {
      "epoch": 1.5834281454745183,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.3239,
      "step": 5045
    },
    {
      "epoch": 1.5837420063556829,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.5401,
      "step": 5046
    },
    {
      "epoch": 1.5840558672368472,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4877,
      "step": 5047
    },
    {
      "epoch": 1.5843697281180118,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.289,
      "step": 5048
    },
    {
      "epoch": 1.5846835889991762,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.3372,
      "step": 5049
    },
    {
      "epoch": 1.5846835889991762,
      "eval_loss": 1.7925392389297485,
      "eval_runtime": 123.6463,
      "eval_samples_per_second": 8.088,
      "eval_steps_per_second": 8.088,
      "step": 5049
    },
    {
      "epoch": 1.5846835889991762,
      "mmlu_eval_accuracy": 0.4240702509811583,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.5625,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.6363636363636364,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "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": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.21875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.5666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.25,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5697674418604651,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.42424242424242425,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.28823529411764703,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.5,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.1857343875626034,
      "step": 5049
    },
    {
      "epoch": 1.5849974498803405,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.257,
      "step": 5050
    },
    {
      "epoch": 1.5853113107615049,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.3983,
      "step": 5051
    },
    {
      "epoch": 1.5856251716426693,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.3132,
      "step": 5052
    },
    {
      "epoch": 1.5859390325238338,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.2861,
      "step": 5053
    },
    {
      "epoch": 1.5862528934049982,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.4179,
      "step": 5054
    },
    {
      "epoch": 1.5865667542861628,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.3597,
      "step": 5055
    },
    {
      "epoch": 1.5868806151673271,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.2883,
      "step": 5056
    },
    {
      "epoch": 1.5871944760484915,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.4195,
      "step": 5057
    },
    {
      "epoch": 1.5875083369296559,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.4079,
      "step": 5058
    },
    {
      "epoch": 1.5878221978108202,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.633,
      "step": 5059
    },
    {
      "epoch": 1.5881360586919848,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.2581,
      "step": 5060
    },
    {
      "epoch": 1.5884499195731492,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.4433,
      "step": 5061
    },
    {
      "epoch": 1.5887637804543138,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.3953,
      "step": 5062
    },
    {
      "epoch": 1.5890776413354781,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.3274,
      "step": 5063
    },
    {
      "epoch": 1.5893915022166425,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.3063,
      "step": 5064
    },
    {
      "epoch": 1.5897053630978069,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.6481,
      "step": 5065
    },
    {
      "epoch": 1.5900192239789712,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.7481,
      "step": 5066
    },
    {
      "epoch": 1.5903330848601356,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.5678,
      "step": 5067
    },
    {
      "epoch": 1.5906469457413002,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.8494,
      "step": 5068
    },
    {
      "epoch": 1.5909608066224648,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.6816,
      "step": 5069
    },
    {
      "epoch": 1.5912746675036291,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.3967,
      "step": 5070
    },
    {
      "epoch": 1.5915885283847935,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.588,
      "step": 5071
    },
    {
      "epoch": 1.5919023892659578,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.4999,
      "step": 5072
    },
    {
      "epoch": 1.5922162501471222,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.7389,
      "step": 5073
    },
    {
      "epoch": 1.5925301110282866,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.0021,
      "step": 5074
    },
    {
      "epoch": 1.5928439719094512,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.7741,
      "step": 5075
    },
    {
      "epoch": 1.5931578327906155,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 2.0061,
      "step": 5076
    },
    {
      "epoch": 1.59347169367178,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 2.3159,
      "step": 5077
    },
    {
      "epoch": 1.5937855545529445,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 2.0496,
      "step": 5078
    },
    {
      "epoch": 1.5940994154341088,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.0805,
      "step": 5079
    },
    {
      "epoch": 1.5944132763152732,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 2.3991,
      "step": 5080
    },
    {
      "epoch": 1.5947271371964375,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.6801,
      "step": 5081
    },
    {
      "epoch": 1.5950409980776021,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 2.0074,
      "step": 5082
    },
    {
      "epoch": 1.5953548589587665,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.8482,
      "step": 5083
    },
    {
      "epoch": 1.595668719839931,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.6638,
      "step": 5084
    },
    {
      "epoch": 1.5959825807210954,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.5203,
      "step": 5085
    },
    {
      "epoch": 1.5962964416022598,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0002,
      "loss": 2.508,
      "step": 5086
    },
    {
      "epoch": 1.5966103024834242,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.4785,
      "step": 5087
    },
    {
      "epoch": 1.5969241633645885,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2313,
      "step": 5088
    },
    {
      "epoch": 1.5972380242457531,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.3661,
      "step": 5089
    },
    {
      "epoch": 1.5975518851269175,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4161,
      "step": 5090
    },
    {
      "epoch": 1.597865746008082,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3296,
      "step": 5091
    },
    {
      "epoch": 1.5981796068892464,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.2616,
      "step": 5092
    },
    {
      "epoch": 1.5984934677704108,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.3501,
      "step": 5093
    },
    {
      "epoch": 1.5988073286515752,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.3956,
      "step": 5094
    },
    {
      "epoch": 1.5991211895327395,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.366,
      "step": 5095
    },
    {
      "epoch": 1.5994350504139039,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.4729,
      "step": 5096
    },
    {
      "epoch": 1.5997489112950685,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.3622,
      "step": 5097
    },
    {
      "epoch": 1.600062772176233,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.2904,
      "step": 5098
    },
    {
      "epoch": 1.6003766330573974,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.337,
      "step": 5099
    },
    {
      "epoch": 1.6006904939385618,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.3583,
      "step": 5100
    },
    {
      "epoch": 1.6010043548197261,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.3341,
      "step": 5101
    },
    {
      "epoch": 1.6013182157008905,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.3488,
      "step": 5102
    },
    {
      "epoch": 1.6016320765820549,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3594,
      "step": 5103
    },
    {
      "epoch": 1.6019459374632194,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.2418,
      "step": 5104
    },
    {
      "epoch": 1.6022597983443838,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.423,
      "step": 5105
    },
    {
      "epoch": 1.6025736592255484,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.5248,
      "step": 5106
    },
    {
      "epoch": 1.6028875201067128,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.3286,
      "step": 5107
    },
    {
      "epoch": 1.6032013809878771,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.406,
      "step": 5108
    },
    {
      "epoch": 1.6035152418690415,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.239,
      "step": 5109
    },
    {
      "epoch": 1.6038291027502058,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.4122,
      "step": 5110
    },
    {
      "epoch": 1.6041429636313704,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.6443,
      "step": 5111
    },
    {
      "epoch": 1.6044568245125348,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.4452,
      "step": 5112
    },
    {
      "epoch": 1.6047706853936994,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.2717,
      "step": 5113
    },
    {
      "epoch": 1.6050845462748637,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.4244,
      "step": 5114
    },
    {
      "epoch": 1.605398407156028,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.4921,
      "step": 5115
    },
    {
      "epoch": 1.6057122680371925,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.7972,
      "step": 5116
    },
    {
      "epoch": 1.6060261289183568,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.6309,
      "step": 5117
    },
    {
      "epoch": 1.6063399897995214,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.6465,
      "step": 5118
    },
    {
      "epoch": 1.6066538506806858,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.7099,
      "step": 5119
    },
    {
      "epoch": 1.6069677115618504,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.0473,
      "step": 5120
    },
    {
      "epoch": 1.6072815724430147,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.8316,
      "step": 5121
    },
    {
      "epoch": 1.607595433324179,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 1.9689,
      "step": 5122
    },
    {
      "epoch": 1.6079092942053435,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.8618,
      "step": 5123
    },
    {
      "epoch": 1.6082231550865078,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.8294,
      "step": 5124
    },
    {
      "epoch": 1.6085370159676722,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002,
      "loss": 2.2998,
      "step": 5125
    },
    {
      "epoch": 1.6088508768488368,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.9971,
      "step": 5126
    },
    {
      "epoch": 1.6091647377300011,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.8556,
      "step": 5127
    },
    {
      "epoch": 1.6094785986111657,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.8845,
      "step": 5128
    },
    {
      "epoch": 1.60979245949233,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.8863,
      "step": 5129
    },
    {
      "epoch": 1.6101063203734944,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 1.8496,
      "step": 5130
    },
    {
      "epoch": 1.6104201812546588,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 2.2089,
      "step": 5131
    },
    {
      "epoch": 1.6107340421358232,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.9057,
      "step": 5132
    },
    {
      "epoch": 1.6110479030169877,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.6346,
      "step": 5133
    },
    {
      "epoch": 1.611361763898152,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.6142,
      "step": 5134
    },
    {
      "epoch": 1.6116756247793167,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 2.1235,
      "step": 5135
    },
    {
      "epoch": 1.611989485660481,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 2.7421,
      "step": 5136
    },
    {
      "epoch": 1.6123033465416454,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.3866,
      "step": 5137
    },
    {
      "epoch": 1.6126172074228098,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1569,
      "step": 5138
    },
    {
      "epoch": 1.6129310683039741,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3513,
      "step": 5139
    },
    {
      "epoch": 1.6132449291851387,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.4872,
      "step": 5140
    },
    {
      "epoch": 1.613558790066303,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.4158,
      "step": 5141
    },
    {
      "epoch": 1.6138726509474677,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.2856,
      "step": 5142
    },
    {
      "epoch": 1.614186511828632,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.4552,
      "step": 5143
    },
    {
      "epoch": 1.6145003727097964,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.3369,
      "step": 5144
    },
    {
      "epoch": 1.6148142335909608,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.4883,
      "step": 5145
    },
    {
      "epoch": 1.6151280944721251,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.3296,
      "step": 5146
    },
    {
      "epoch": 1.6154419553532895,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.231,
      "step": 5147
    },
    {
      "epoch": 1.615755816234454,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3855,
      "step": 5148
    },
    {
      "epoch": 1.6160696771156187,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.3475,
      "step": 5149
    },
    {
      "epoch": 1.616383537996783,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.1748,
      "step": 5150
    },
    {
      "epoch": 1.6166973988779474,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.4136,
      "step": 5151
    },
    {
      "epoch": 1.6170112597591118,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.227,
      "step": 5152
    },
    {
      "epoch": 1.6173251206402761,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.2219,
      "step": 5153
    },
    {
      "epoch": 1.6176389815214405,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.3647,
      "step": 5154
    },
    {
      "epoch": 1.617952842402605,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.3262,
      "step": 5155
    },
    {
      "epoch": 1.6182667032837694,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.39,
      "step": 5156
    },
    {
      "epoch": 1.618580564164934,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.299,
      "step": 5157
    },
    {
      "epoch": 1.6188944250460984,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.294,
      "step": 5158
    },
    {
      "epoch": 1.6192082859272627,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.3709,
      "step": 5159
    },
    {
      "epoch": 1.619522146808427,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.5415,
      "step": 5160
    },
    {
      "epoch": 1.6198360076895915,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.688,
      "step": 5161
    },
    {
      "epoch": 1.620149868570756,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.6558,
      "step": 5162
    },
    {
      "epoch": 1.6204637294519204,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.7029,
      "step": 5163
    },
    {
      "epoch": 1.620777590333085,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.5837,
      "step": 5164
    },
    {
      "epoch": 1.6210914512142494,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.7758,
      "step": 5165
    },
    {
      "epoch": 1.6214053120954137,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.6964,
      "step": 5166
    },
    {
      "epoch": 1.621719172976578,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.4762,
      "step": 5167
    },
    {
      "epoch": 1.6220330338577424,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.5627,
      "step": 5168
    },
    {
      "epoch": 1.622346894738907,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.9072,
      "step": 5169
    },
    {
      "epoch": 1.6226607556200714,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.5664,
      "step": 5170
    },
    {
      "epoch": 1.622974616501236,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.5761,
      "step": 5171
    },
    {
      "epoch": 1.6232884773824003,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 1.9573,
      "step": 5172
    },
    {
      "epoch": 1.6236023382635647,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.001,
      "step": 5173
    },
    {
      "epoch": 1.623916199144729,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.3235,
      "step": 5174
    },
    {
      "epoch": 1.6242300600258934,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 2.2949,
      "step": 5175
    },
    {
      "epoch": 1.6245439209070578,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 2.0129,
      "step": 5176
    },
    {
      "epoch": 1.6248577817882224,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 2.0571,
      "step": 5177
    },
    {
      "epoch": 1.6251716426693867,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.3679,
      "step": 5178
    },
    {
      "epoch": 1.6254855035505513,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.7976,
      "step": 5179
    },
    {
      "epoch": 1.6257993644317157,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.1299,
      "step": 5180
    },
    {
      "epoch": 1.62611322531288,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.7216,
      "step": 5181
    },
    {
      "epoch": 1.6264270861940444,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.518,
      "step": 5182
    },
    {
      "epoch": 1.6267409470752088,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.1603,
      "step": 5183
    },
    {
      "epoch": 1.6270548079563734,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.9139,
      "step": 5184
    },
    {
      "epoch": 1.6273686688375377,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 2.3273,
      "step": 5185
    },
    {
      "epoch": 1.6276825297187023,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 2.4695,
      "step": 5186
    },
    {
      "epoch": 1.6279963905998667,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.4228,
      "step": 5187
    },
    {
      "epoch": 1.628310251481031,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3978,
      "step": 5188
    },
    {
      "epoch": 1.6286241123621954,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3138,
      "step": 5189
    },
    {
      "epoch": 1.6289379732433598,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.39,
      "step": 5190
    },
    {
      "epoch": 1.6292518341245243,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.4673,
      "step": 5191
    },
    {
      "epoch": 1.6295656950056887,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.3023,
      "step": 5192
    },
    {
      "epoch": 1.6298795558868533,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2163,
      "step": 5193
    },
    {
      "epoch": 1.6301934167680177,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.27,
      "step": 5194
    },
    {
      "epoch": 1.630507277649182,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.4509,
      "step": 5195
    },
    {
      "epoch": 1.6308211385303464,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.344,
      "step": 5196
    },
    {
      "epoch": 1.6311349994115107,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.4739,
      "step": 5197
    },
    {
      "epoch": 1.631448860292675,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.329,
      "step": 5198
    },
    {
      "epoch": 1.6317627211738397,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.3776,
      "step": 5199
    },
    {
      "epoch": 1.6320765820550043,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.3536,
      "step": 5200
    },
    {
      "epoch": 1.6323904429361686,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.2967,
      "step": 5201
    },
    {
      "epoch": 1.632704303817333,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.2148,
      "step": 5202
    },
    {
      "epoch": 1.6330181646984974,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.1922,
      "step": 5203
    },
    {
      "epoch": 1.6333320255796617,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.4999,
      "step": 5204
    },
    {
      "epoch": 1.633645886460826,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.2931,
      "step": 5205
    },
    {
      "epoch": 1.6339597473419907,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.418,
      "step": 5206
    },
    {
      "epoch": 1.634273608223155,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.4777,
      "step": 5207
    },
    {
      "epoch": 1.6345874691043196,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.3601,
      "step": 5208
    },
    {
      "epoch": 1.634901329985484,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.3026,
      "step": 5209
    },
    {
      "epoch": 1.6352151908666483,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4258,
      "step": 5210
    },
    {
      "epoch": 1.6355290517478127,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.3424,
      "step": 5211
    },
    {
      "epoch": 1.635842912628977,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.6434,
      "step": 5212
    },
    {
      "epoch": 1.6361567735101417,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.4022,
      "step": 5213
    },
    {
      "epoch": 1.636470634391306,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.5548,
      "step": 5214
    },
    {
      "epoch": 1.6367844952724706,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 1.4238,
      "step": 5215
    },
    {
      "epoch": 1.637098356153635,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.6629,
      "step": 5216
    },
    {
      "epoch": 1.6374122170347993,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.7608,
      "step": 5217
    },
    {
      "epoch": 1.6377260779159637,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.7776,
      "step": 5218
    },
    {
      "epoch": 1.638039938797128,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.5472,
      "step": 5219
    },
    {
      "epoch": 1.6383537996782926,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.7492,
      "step": 5220
    },
    {
      "epoch": 1.638667660559457,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 2.0288,
      "step": 5221
    },
    {
      "epoch": 1.6389815214406216,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.2096,
      "step": 5222
    },
    {
      "epoch": 1.639295382321786,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 2.0986,
      "step": 5223
    },
    {
      "epoch": 1.6396092432029503,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.6296,
      "step": 5224
    },
    {
      "epoch": 1.6399231040841147,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.9282,
      "step": 5225
    },
    {
      "epoch": 1.640236964965279,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.9752,
      "step": 5226
    },
    {
      "epoch": 1.6405508258464434,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 2.299,
      "step": 5227
    },
    {
      "epoch": 1.640864686727608,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.8258,
      "step": 5228
    },
    {
      "epoch": 1.6411785476087726,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.6818,
      "step": 5229
    },
    {
      "epoch": 1.641492408489937,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.6708,
      "step": 5230
    },
    {
      "epoch": 1.6418062693711013,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.13,
      "step": 5231
    },
    {
      "epoch": 1.6421201302522657,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.9442,
      "step": 5232
    },
    {
      "epoch": 1.64243399113343,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.9665,
      "step": 5233
    },
    {
      "epoch": 1.6427478520145944,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.8521,
      "step": 5234
    },
    {
      "epoch": 1.643061712895759,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.8763,
      "step": 5235
    },
    {
      "epoch": 1.6433755737769233,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 2.6208,
      "step": 5236
    },
    {
      "epoch": 1.6433755737769233,
      "eval_loss": 1.7919257879257202,
      "eval_runtime": 146.8008,
      "eval_samples_per_second": 6.812,
      "eval_steps_per_second": 6.812,
      "step": 5236
    },
    {
      "epoch": 1.6433755737769233,
      "mmlu_eval_accuracy": 0.4046188198456904,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.6363636363636364,
      "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": 0.2727272727272727,
      "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": 0.2682926829268293,
      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.25,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "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": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.5833333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.68,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5581395348837209,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.5151515151515151,
      "mmlu_eval_accuracy_philosophy": 0.38235294117647056,
      "mmlu_eval_accuracy_prehistory": 0.37142857142857144,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.5789473684210527,
      "mmlu_loss": 1.0532882166679665,
      "step": 5236
    },
    {
      "epoch": 1.643689434658088,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.5944,
      "step": 5237
    },
    {
      "epoch": 1.6440032955392523,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3256,
      "step": 5238
    },
    {
      "epoch": 1.6443171564204166,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.4149,
      "step": 5239
    },
    {
      "epoch": 1.644631017301581,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.32,
      "step": 5240
    },
    {
      "epoch": 1.6449448781827454,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.472,
      "step": 5241
    },
    {
      "epoch": 1.64525873906391,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.4399,
      "step": 5242
    },
    {
      "epoch": 1.6455725999450743,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.4195,
      "step": 5243
    },
    {
      "epoch": 1.645886460826239,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3585,
      "step": 5244
    },
    {
      "epoch": 1.6462003217074033,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.5199,
      "step": 5245
    },
    {
      "epoch": 1.6465141825885676,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.4195,
      "step": 5246
    },
    {
      "epoch": 1.646828043469732,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.3701,
      "step": 5247
    },
    {
      "epoch": 1.6471419043508964,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.3268,
      "step": 5248
    },
    {
      "epoch": 1.647455765232061,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.2806,
      "step": 5249
    },
    {
      "epoch": 1.6477696261132253,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.4121,
      "step": 5250
    },
    {
      "epoch": 1.64808348699439,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.2581,
      "step": 5251
    },
    {
      "epoch": 1.6483973478755543,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.4176,
      "step": 5252
    },
    {
      "epoch": 1.6487112087567186,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.6702,
      "step": 5253
    },
    {
      "epoch": 1.649025069637883,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.374,
      "step": 5254
    },
    {
      "epoch": 1.6493389305190473,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.1493,
      "step": 5255
    },
    {
      "epoch": 1.6496527914002117,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.4382,
      "step": 5256
    },
    {
      "epoch": 1.6499666522813763,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.4023,
      "step": 5257
    },
    {
      "epoch": 1.6502805131625407,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.5046,
      "step": 5258
    },
    {
      "epoch": 1.6505943740437052,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.4378,
      "step": 5259
    },
    {
      "epoch": 1.6509082349248696,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.6153,
      "step": 5260
    },
    {
      "epoch": 1.651222095806034,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.4205,
      "step": 5261
    },
    {
      "epoch": 1.6515359566871983,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.6033,
      "step": 5262
    },
    {
      "epoch": 1.6518498175683627,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.8266,
      "step": 5263
    },
    {
      "epoch": 1.6521636784495273,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.611,
      "step": 5264
    },
    {
      "epoch": 1.6524775393306916,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.5343,
      "step": 5265
    },
    {
      "epoch": 1.6527914002118562,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.091,
      "step": 5266
    },
    {
      "epoch": 1.6531052610930206,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.8864,
      "step": 5267
    },
    {
      "epoch": 1.653419121974185,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.8913,
      "step": 5268
    },
    {
      "epoch": 1.6537329828553493,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.6002,
      "step": 5269
    },
    {
      "epoch": 1.6540468437365137,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.7484,
      "step": 5270
    },
    {
      "epoch": 1.6543607046176783,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 2.0815,
      "step": 5271
    },
    {
      "epoch": 1.6546745654988426,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 2.0333,
      "step": 5272
    },
    {
      "epoch": 1.6549884263800072,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.9842,
      "step": 5273
    },
    {
      "epoch": 1.6553022872611716,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 2.1703,
      "step": 5274
    },
    {
      "epoch": 1.655616148142336,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 2.2295,
      "step": 5275
    },
    {
      "epoch": 1.6559300090235003,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.9882,
      "step": 5276
    },
    {
      "epoch": 1.6562438699046647,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 2.1352,
      "step": 5277
    },
    {
      "epoch": 1.656557730785829,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 2.1654,
      "step": 5278
    },
    {
      "epoch": 1.6568715916669936,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.0885,
      "step": 5279
    },
    {
      "epoch": 1.6571854525481582,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.9674,
      "step": 5280
    },
    {
      "epoch": 1.6574993134293226,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002,
      "loss": 2.0082,
      "step": 5281
    },
    {
      "epoch": 1.657813174310487,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 2.0707,
      "step": 5282
    },
    {
      "epoch": 1.6581270351916513,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.8177,
      "step": 5283
    },
    {
      "epoch": 1.6584408960728156,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.8653,
      "step": 5284
    },
    {
      "epoch": 1.65875475695398,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.8444,
      "step": 5285
    },
    {
      "epoch": 1.6590686178351446,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 2.4233,
      "step": 5286
    },
    {
      "epoch": 1.659382478716309,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4098,
      "step": 5287
    },
    {
      "epoch": 1.6596963395974735,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4043,
      "step": 5288
    },
    {
      "epoch": 1.660010200478638,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.3667,
      "step": 5289
    },
    {
      "epoch": 1.6603240613598023,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.4895,
      "step": 5290
    },
    {
      "epoch": 1.6606379222409666,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3081,
      "step": 5291
    },
    {
      "epoch": 1.660951783122131,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.408,
      "step": 5292
    },
    {
      "epoch": 1.6612656440032956,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.4971,
      "step": 5293
    },
    {
      "epoch": 1.66157950488446,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.4166,
      "step": 5294
    },
    {
      "epoch": 1.6618933657656245,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.5222,
      "step": 5295
    },
    {
      "epoch": 1.6622072266467889,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.3026,
      "step": 5296
    },
    {
      "epoch": 1.6625210875279532,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.3182,
      "step": 5297
    },
    {
      "epoch": 1.6628349484091176,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.4146,
      "step": 5298
    },
    {
      "epoch": 1.663148809290282,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.3603,
      "step": 5299
    },
    {
      "epoch": 1.6634626701714466,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.4746,
      "step": 5300
    },
    {
      "epoch": 1.663776531052611,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.2972,
      "step": 5301
    },
    {
      "epoch": 1.6640903919337755,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.2751,
      "step": 5302
    },
    {
      "epoch": 1.6644042528149399,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.345,
      "step": 5303
    },
    {
      "epoch": 1.6647181136961042,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.3429,
      "step": 5304
    },
    {
      "epoch": 1.6650319745772686,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.188,
      "step": 5305
    },
    {
      "epoch": 1.665345835458433,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.2375,
      "step": 5306
    },
    {
      "epoch": 1.6656596963395973,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4213,
      "step": 5307
    },
    {
      "epoch": 1.665973557220762,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.1308,
      "step": 5308
    },
    {
      "epoch": 1.6662874181019265,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3391,
      "step": 5309
    },
    {
      "epoch": 1.6666012789830909,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.5578,
      "step": 5310
    },
    {
      "epoch": 1.6669151398642552,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.7623,
      "step": 5311
    },
    {
      "epoch": 1.6672290007454196,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.6784,
      "step": 5312
    },
    {
      "epoch": 1.667542861626584,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.4349,
      "step": 5313
    },
    {
      "epoch": 1.6678567225077483,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.6305,
      "step": 5314
    },
    {
      "epoch": 1.6681705833889129,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.8189,
      "step": 5315
    },
    {
      "epoch": 1.6684844442700772,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 2.0669,
      "step": 5316
    },
    {
      "epoch": 1.6687983051512418,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.725,
      "step": 5317
    },
    {
      "epoch": 1.6691121660324062,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.9341,
      "step": 5318
    },
    {
      "epoch": 1.6694260269135706,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.7933,
      "step": 5319
    },
    {
      "epoch": 1.669739887794735,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 1.5582,
      "step": 5320
    },
    {
      "epoch": 1.6700537486758993,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.1427,
      "step": 5321
    },
    {
      "epoch": 1.6703676095570639,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.9629,
      "step": 5322
    },
    {
      "epoch": 1.6706814704382282,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 2.0382,
      "step": 5323
    },
    {
      "epoch": 1.6709953313193928,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 2.1086,
      "step": 5324
    },
    {
      "epoch": 1.6713091922005572,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.6112,
      "step": 5325
    },
    {
      "epoch": 1.6716230530817215,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.9886,
      "step": 5326
    },
    {
      "epoch": 1.671936913962886,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.3365,
      "step": 5327
    },
    {
      "epoch": 1.6722507748440503,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.1497,
      "step": 5328
    },
    {
      "epoch": 1.6725646357252146,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 2.1876,
      "step": 5329
    },
    {
      "epoch": 1.6728784966063792,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0002,
      "loss": 1.9152,
      "step": 5330
    },
    {
      "epoch": 1.6731923574875438,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.9573,
      "step": 5331
    },
    {
      "epoch": 1.6735062183687082,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.8511,
      "step": 5332
    },
    {
      "epoch": 1.6738200792498725,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.4428,
      "step": 5333
    },
    {
      "epoch": 1.674133940131037,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.5779,
      "step": 5334
    },
    {
      "epoch": 1.6744478010122013,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 2.4024,
      "step": 5335
    },
    {
      "epoch": 1.6747616618933656,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 2.1432,
      "step": 5336
    },
    {
      "epoch": 1.6750755227745302,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.4491,
      "step": 5337
    },
    {
      "epoch": 1.6753893836556946,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4891,
      "step": 5338
    },
    {
      "epoch": 1.6757032445368591,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.4065,
      "step": 5339
    },
    {
      "epoch": 1.6760171054180235,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.3976,
      "step": 5340
    },
    {
      "epoch": 1.6763309662991879,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2961,
      "step": 5341
    },
    {
      "epoch": 1.6766448271803522,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4172,
      "step": 5342
    },
    {
      "epoch": 1.6769586880615166,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.343,
      "step": 5343
    },
    {
      "epoch": 1.6772725489426812,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.3408,
      "step": 5344
    },
    {
      "epoch": 1.6775864098238455,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.4977,
      "step": 5345
    },
    {
      "epoch": 1.6779002707050101,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.4584,
      "step": 5346
    },
    {
      "epoch": 1.6782141315861745,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.4354,
      "step": 5347
    },
    {
      "epoch": 1.6785279924673389,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.3733,
      "step": 5348
    },
    {
      "epoch": 1.6788418533485032,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.2364,
      "step": 5349
    },
    {
      "epoch": 1.6791557142296676,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.2784,
      "step": 5350
    },
    {
      "epoch": 1.6794695751108322,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.4247,
      "step": 5351
    },
    {
      "epoch": 1.6797834359919965,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2141,
      "step": 5352
    },
    {
      "epoch": 1.6800972968731611,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.3006,
      "step": 5353
    },
    {
      "epoch": 1.6804111577543255,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.3487,
      "step": 5354
    },
    {
      "epoch": 1.6807250186354898,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.2892,
      "step": 5355
    },
    {
      "epoch": 1.6810388795166542,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.3255,
      "step": 5356
    },
    {
      "epoch": 1.6813527403978186,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.6207,
      "step": 5357
    },
    {
      "epoch": 1.681666601278983,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.6852,
      "step": 5358
    },
    {
      "epoch": 1.6819804621601475,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.3062,
      "step": 5359
    },
    {
      "epoch": 1.682294323041312,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.4352,
      "step": 5360
    },
    {
      "epoch": 1.6826081839224765,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.4089,
      "step": 5361
    },
    {
      "epoch": 1.6829220448036408,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.4615,
      "step": 5362
    },
    {
      "epoch": 1.6832359056848052,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.7598,
      "step": 5363
    },
    {
      "epoch": 1.6835497665659696,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.9255,
      "step": 5364
    },
    {
      "epoch": 1.683863627447134,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.4791,
      "step": 5365
    },
    {
      "epoch": 1.6841774883282985,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.8529,
      "step": 5366
    },
    {
      "epoch": 1.6844913492094629,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.6055,
      "step": 5367
    },
    {
      "epoch": 1.6848052100906274,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.7601,
      "step": 5368
    },
    {
      "epoch": 1.6851190709717918,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 2.0651,
      "step": 5369
    },
    {
      "epoch": 1.6854329318529562,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.8065,
      "step": 5370
    },
    {
      "epoch": 1.6857467927341205,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.9603,
      "step": 5371
    },
    {
      "epoch": 1.686060653615285,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.8112,
      "step": 5372
    },
    {
      "epoch": 1.6863745144964495,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.5999,
      "step": 5373
    },
    {
      "epoch": 1.6866883753776138,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.1754,
      "step": 5374
    },
    {
      "epoch": 1.6870022362587784,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.7879,
      "step": 5375
    },
    {
      "epoch": 1.6873160971399428,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.9381,
      "step": 5376
    },
    {
      "epoch": 1.6876299580211072,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.6776,
      "step": 5377
    },
    {
      "epoch": 1.6879438189022715,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.0856,
      "step": 5378
    },
    {
      "epoch": 1.6882576797834359,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.8692,
      "step": 5379
    },
    {
      "epoch": 1.6885715406646005,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002,
      "loss": 2.1128,
      "step": 5380
    },
    {
      "epoch": 1.6888854015457648,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.2145,
      "step": 5381
    },
    {
      "epoch": 1.6891992624269294,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.8243,
      "step": 5382
    },
    {
      "epoch": 1.6895131233080938,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 1.8438,
      "step": 5383
    },
    {
      "epoch": 1.6898269841892581,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 2.1272,
      "step": 5384
    },
    {
      "epoch": 1.6901408450704225,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 2.3897,
      "step": 5385
    },
    {
      "epoch": 1.6904547059515869,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 2.5437,
      "step": 5386
    },
    {
      "epoch": 1.6907685668327512,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.4554,
      "step": 5387
    },
    {
      "epoch": 1.6910824277139158,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4041,
      "step": 5388
    },
    {
      "epoch": 1.6913962885950802,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3059,
      "step": 5389
    },
    {
      "epoch": 1.6917101494762448,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.4077,
      "step": 5390
    },
    {
      "epoch": 1.6920240103574091,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.4185,
      "step": 5391
    },
    {
      "epoch": 1.6923378712385735,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.4299,
      "step": 5392
    },
    {
      "epoch": 1.6926517321197379,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.3579,
      "step": 5393
    },
    {
      "epoch": 1.6929655930009022,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.4735,
      "step": 5394
    },
    {
      "epoch": 1.6932794538820668,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.3569,
      "step": 5395
    },
    {
      "epoch": 1.6935933147632312,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.5498,
      "step": 5396
    },
    {
      "epoch": 1.6939071756443957,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.429,
      "step": 5397
    },
    {
      "epoch": 1.69422103652556,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.4454,
      "step": 5398
    },
    {
      "epoch": 1.6945348974067245,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.2722,
      "step": 5399
    },
    {
      "epoch": 1.6948487582878888,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.2964,
      "step": 5400
    },
    {
      "epoch": 1.6951626191690532,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.3567,
      "step": 5401
    },
    {
      "epoch": 1.6954764800502178,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.3113,
      "step": 5402
    },
    {
      "epoch": 1.6957903409313821,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.2178,
      "step": 5403
    },
    {
      "epoch": 1.6961042018125467,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.377,
      "step": 5404
    },
    {
      "epoch": 1.696418062693711,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.3427,
      "step": 5405
    },
    {
      "epoch": 1.6967319235748755,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.4818,
      "step": 5406
    },
    {
      "epoch": 1.6970457844560398,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.3405,
      "step": 5407
    },
    {
      "epoch": 1.6973596453372042,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.3294,
      "step": 5408
    },
    {
      "epoch": 1.6976735062183685,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.4936,
      "step": 5409
    },
    {
      "epoch": 1.6979873670995331,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.282,
      "step": 5410
    },
    {
      "epoch": 1.6983012279806977,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.6205,
      "step": 5411
    },
    {
      "epoch": 1.698615088861862,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.457,
      "step": 5412
    },
    {
      "epoch": 1.6989289497430264,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.4012,
      "step": 5413
    },
    {
      "epoch": 1.6992428106241908,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.4042,
      "step": 5414
    },
    {
      "epoch": 1.6995566715053552,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.5957,
      "step": 5415
    },
    {
      "epoch": 1.6998705323865195,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.6471,
      "step": 5416
    },
    {
      "epoch": 1.7001843932676841,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.9019,
      "step": 5417
    },
    {
      "epoch": 1.7004982541488485,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.5745,
      "step": 5418
    },
    {
      "epoch": 1.700812115030013,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.845,
      "step": 5419
    },
    {
      "epoch": 1.7011259759111774,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 2.0496,
      "step": 5420
    },
    {
      "epoch": 1.7014398367923418,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.7222,
      "step": 5421
    },
    {
      "epoch": 1.7017536976735061,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.8778,
      "step": 5422
    },
    {
      "epoch": 1.7020675585546705,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.7047,
      "step": 5423
    },
    {
      "epoch": 1.7020675585546705,
      "eval_loss": 1.7857495546340942,
      "eval_runtime": 147.0108,
      "eval_samples_per_second": 6.802,
      "eval_steps_per_second": 6.802,
      "step": 5423
    },
    {
      "epoch": 1.7020675585546705,
      "mmlu_eval_accuracy": 0.4185827138010515,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3448275862068966,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "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": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5,
      "mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5813953488372093,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.25,
      "mmlu_eval_accuracy_nutrition": 0.48484848484848486,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.37142857142857144,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.3333333333333333,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.5,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.2777777777777778,
      "mmlu_eval_accuracy_world_religions": 0.5789473684210527,
      "mmlu_loss": 1.0690680153030419,
      "step": 5423
    },
    {
      "epoch": 1.702381419435835,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.8613,
      "step": 5424
    },
    {
      "epoch": 1.7026952803169995,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 2.518,
      "step": 5425
    },
    {
      "epoch": 1.703009141198164,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.6108,
      "step": 5426
    },
    {
      "epoch": 1.7033230020793284,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.6537,
      "step": 5427
    },
    {
      "epoch": 1.7036368629604928,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.6812,
      "step": 5428
    },
    {
      "epoch": 1.7039507238416571,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.6862,
      "step": 5429
    },
    {
      "epoch": 1.7042645847228215,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.8144,
      "step": 5430
    },
    {
      "epoch": 1.704578445603986,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.3003,
      "step": 5431
    },
    {
      "epoch": 1.7048923064851504,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.868,
      "step": 5432
    },
    {
      "epoch": 1.705206167366315,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 2.5762,
      "step": 5433
    },
    {
      "epoch": 1.7055200282474794,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 2.1479,
      "step": 5434
    },
    {
      "epoch": 1.7058338891286438,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.8356,
      "step": 5435
    },
    {
      "epoch": 1.7061477500098081,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 2.7924,
      "step": 5436
    },
    {
      "epoch": 1.7064616108909725,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.6025,
      "step": 5437
    },
    {
      "epoch": 1.7067754717721368,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3645,
      "step": 5438
    },
    {
      "epoch": 1.7070893326533014,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4154,
      "step": 5439
    },
    {
      "epoch": 1.707403193534466,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3476,
      "step": 5440
    },
    {
      "epoch": 1.7077170544156304,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.5175,
      "step": 5441
    },
    {
      "epoch": 1.7080309152967947,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.4335,
      "step": 5442
    },
    {
      "epoch": 1.708344776177959,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.4175,
      "step": 5443
    },
    {
      "epoch": 1.7086586370591235,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.4136,
      "step": 5444
    },
    {
      "epoch": 1.7089724979402878,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.4126,
      "step": 5445
    },
    {
      "epoch": 1.7092863588214524,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.3386,
      "step": 5446
    },
    {
      "epoch": 1.7096002197026168,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.2632,
      "step": 5447
    },
    {
      "epoch": 1.7099140805837814,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.5089,
      "step": 5448
    },
    {
      "epoch": 1.7102279414649457,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.294,
      "step": 5449
    },
    {
      "epoch": 1.71054180234611,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.2516,
      "step": 5450
    },
    {
      "epoch": 1.7108556632272744,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.2478,
      "step": 5451
    },
    {
      "epoch": 1.7111695241084388,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.2558,
      "step": 5452
    },
    {
      "epoch": 1.7114833849896034,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.4596,
      "step": 5453
    },
    {
      "epoch": 1.7117972458707678,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.5655,
      "step": 5454
    },
    {
      "epoch": 1.7121111067519323,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.5339,
      "step": 5455
    },
    {
      "epoch": 1.7124249676330967,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.2464,
      "step": 5456
    },
    {
      "epoch": 1.712738828514261,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.5458,
      "step": 5457
    },
    {
      "epoch": 1.7130526893954254,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.2654,
      "step": 5458
    },
    {
      "epoch": 1.7133665502765898,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.296,
      "step": 5459
    },
    {
      "epoch": 1.7136804111577544,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.342,
      "step": 5460
    },
    {
      "epoch": 1.7139942720389187,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.6868,
      "step": 5461
    },
    {
      "epoch": 1.7143081329200833,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.3321,
      "step": 5462
    },
    {
      "epoch": 1.7146219938012477,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 1.6589,
      "step": 5463
    },
    {
      "epoch": 1.714935854682412,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.5774,
      "step": 5464
    },
    {
      "epoch": 1.7152497155635764,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.8594,
      "step": 5465
    },
    {
      "epoch": 1.7155635764447408,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.6811,
      "step": 5466
    },
    {
      "epoch": 1.7158774373259051,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.6543,
      "step": 5467
    },
    {
      "epoch": 1.7161912982070697,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.7647,
      "step": 5468
    },
    {
      "epoch": 1.716505159088234,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.9218,
      "step": 5469
    },
    {
      "epoch": 1.7168190199693987,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.9351,
      "step": 5470
    },
    {
      "epoch": 1.717132880850563,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.817,
      "step": 5471
    },
    {
      "epoch": 1.7174467417317274,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.0335,
      "step": 5472
    },
    {
      "epoch": 1.7177606026128918,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 2.0642,
      "step": 5473
    },
    {
      "epoch": 1.7180744634940561,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.6929,
      "step": 5474
    },
    {
      "epoch": 1.7183883243752207,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.0268,
      "step": 5475
    },
    {
      "epoch": 1.718702185256385,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.1018,
      "step": 5476
    },
    {
      "epoch": 1.7190160461375497,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.7464,
      "step": 5477
    },
    {
      "epoch": 1.719329907018714,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.5394,
      "step": 5478
    },
    {
      "epoch": 1.7196437678998784,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 2.3166,
      "step": 5479
    },
    {
      "epoch": 1.7199576287810427,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.2199,
      "step": 5480
    },
    {
      "epoch": 1.720271489662207,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 2.2483,
      "step": 5481
    },
    {
      "epoch": 1.7205853505433717,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 1.9631,
      "step": 5482
    },
    {
      "epoch": 1.720899211424536,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.9375,
      "step": 5483
    },
    {
      "epoch": 1.7212130723057006,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.6763,
      "step": 5484
    },
    {
      "epoch": 1.721526933186865,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 2.4518,
      "step": 5485
    },
    {
      "epoch": 1.7218407940680294,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002,
      "loss": 2.7767,
      "step": 5486
    },
    {
      "epoch": 1.7221546549491937,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.5217,
      "step": 5487
    },
    {
      "epoch": 1.722468515830358,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3144,
      "step": 5488
    },
    {
      "epoch": 1.7227823767115225,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2063,
      "step": 5489
    },
    {
      "epoch": 1.723096237592687,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.4158,
      "step": 5490
    },
    {
      "epoch": 1.7234100984738516,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.3824,
      "step": 5491
    },
    {
      "epoch": 1.723723959355016,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.5642,
      "step": 5492
    },
    {
      "epoch": 1.7240378202361804,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.2813,
      "step": 5493
    },
    {
      "epoch": 1.7243516811173447,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.3315,
      "step": 5494
    },
    {
      "epoch": 1.724665541998509,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.3186,
      "step": 5495
    },
    {
      "epoch": 1.7249794028796734,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.2724,
      "step": 5496
    },
    {
      "epoch": 1.725293263760838,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.2774,
      "step": 5497
    },
    {
      "epoch": 1.7256071246420024,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.4014,
      "step": 5498
    },
    {
      "epoch": 1.725920985523167,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.4198,
      "step": 5499
    },
    {
      "epoch": 1.7262348464043313,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.4122,
      "step": 5500
    },
    {
      "epoch": 1.7265487072854957,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.4289,
      "step": 5501
    },
    {
      "epoch": 1.72686256816666,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.235,
      "step": 5502
    },
    {
      "epoch": 1.7271764290478244,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.2452,
      "step": 5503
    },
    {
      "epoch": 1.727490289928989,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.2983,
      "step": 5504
    },
    {
      "epoch": 1.7278041508101534,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.3754,
      "step": 5505
    },
    {
      "epoch": 1.728118011691318,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.5459,
      "step": 5506
    },
    {
      "epoch": 1.7284318725724823,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.4628,
      "step": 5507
    },
    {
      "epoch": 1.7287457334536467,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.0722,
      "step": 5508
    },
    {
      "epoch": 1.729059594334811,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.4004,
      "step": 5509
    },
    {
      "epoch": 1.7293734552159754,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.4633,
      "step": 5510
    },
    {
      "epoch": 1.72968731609714,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.489,
      "step": 5511
    },
    {
      "epoch": 1.7300011769783044,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 1.352,
      "step": 5512
    },
    {
      "epoch": 1.730315037859469,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.7752,
      "step": 5513
    },
    {
      "epoch": 1.7306288987406333,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.6925,
      "step": 5514
    },
    {
      "epoch": 1.7309427596217977,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.4145,
      "step": 5515
    },
    {
      "epoch": 1.731256620502962,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.7095,
      "step": 5516
    },
    {
      "epoch": 1.7315704813841264,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 1.5195,
      "step": 5517
    },
    {
      "epoch": 1.7318843422652908,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.5376,
      "step": 5518
    },
    {
      "epoch": 1.7321982031464553,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 2.0259,
      "step": 5519
    },
    {
      "epoch": 1.7325120640276197,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 2.0046,
      "step": 5520
    },
    {
      "epoch": 1.7328259249087843,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.8622,
      "step": 5521
    },
    {
      "epoch": 1.7331397857899487,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.7366,
      "step": 5522
    },
    {
      "epoch": 1.733453646671113,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.939,
      "step": 5523
    },
    {
      "epoch": 1.7337675075522774,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.8009,
      "step": 5524
    },
    {
      "epoch": 1.7340813684334417,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 2.1492,
      "step": 5525
    },
    {
      "epoch": 1.7343952293146063,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.9199,
      "step": 5526
    },
    {
      "epoch": 1.7347090901957707,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 2.056,
      "step": 5527
    },
    {
      "epoch": 1.7350229510769353,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 2.013,
      "step": 5528
    },
    {
      "epoch": 1.7353368119580996,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.7799,
      "step": 5529
    },
    {
      "epoch": 1.735650672839264,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.2011,
      "step": 5530
    },
    {
      "epoch": 1.7359645337204284,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 2.0116,
      "step": 5531
    },
    {
      "epoch": 1.7362783946015927,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 2.0485,
      "step": 5532
    },
    {
      "epoch": 1.7365922554827573,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.5143,
      "step": 5533
    },
    {
      "epoch": 1.7369061163639217,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.9354,
      "step": 5534
    },
    {
      "epoch": 1.7372199772450863,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 2.1758,
      "step": 5535
    },
    {
      "epoch": 1.7375338381262506,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 2.7749,
      "step": 5536
    },
    {
      "epoch": 1.737847699007415,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.5269,
      "step": 5537
    },
    {
      "epoch": 1.7381615598885793,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.383,
      "step": 5538
    },
    {
      "epoch": 1.7384754207697437,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.3237,
      "step": 5539
    },
    {
      "epoch": 1.738789281650908,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.2926,
      "step": 5540
    },
    {
      "epoch": 1.7391031425320727,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3281,
      "step": 5541
    },
    {
      "epoch": 1.7394170034132372,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.3902,
      "step": 5542
    },
    {
      "epoch": 1.7397308642944016,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.3967,
      "step": 5543
    },
    {
      "epoch": 1.740044725175566,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.221,
      "step": 5544
    },
    {
      "epoch": 1.7403585860567303,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.1903,
      "step": 5545
    },
    {
      "epoch": 1.7406724469378947,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.2487,
      "step": 5546
    },
    {
      "epoch": 1.740986307819059,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.4515,
      "step": 5547
    },
    {
      "epoch": 1.7413001687002236,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.1023,
      "step": 5548
    },
    {
      "epoch": 1.741614029581388,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.3426,
      "step": 5549
    },
    {
      "epoch": 1.7419278904625526,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.3032,
      "step": 5550
    },
    {
      "epoch": 1.742241751343717,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.3823,
      "step": 5551
    },
    {
      "epoch": 1.7425556122248813,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.3341,
      "step": 5552
    },
    {
      "epoch": 1.7428694731060457,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.3123,
      "step": 5553
    },
    {
      "epoch": 1.74318333398721,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.4137,
      "step": 5554
    },
    {
      "epoch": 1.7434971948683746,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.2661,
      "step": 5555
    },
    {
      "epoch": 1.743811055749539,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.3533,
      "step": 5556
    },
    {
      "epoch": 1.7441249166307036,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.2212,
      "step": 5557
    },
    {
      "epoch": 1.744438777511868,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.6135,
      "step": 5558
    },
    {
      "epoch": 1.7447526383930323,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.4766,
      "step": 5559
    },
    {
      "epoch": 1.7450664992741967,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.4571,
      "step": 5560
    },
    {
      "epoch": 1.745380360155361,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.3557,
      "step": 5561
    },
    {
      "epoch": 1.7456942210365256,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.6698,
      "step": 5562
    },
    {
      "epoch": 1.74600808191769,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.8339,
      "step": 5563
    },
    {
      "epoch": 1.7463219427988546,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 2.0184,
      "step": 5564
    },
    {
      "epoch": 1.746635803680019,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.9101,
      "step": 5565
    },
    {
      "epoch": 1.7469496645611833,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 1.8805,
      "step": 5566
    },
    {
      "epoch": 1.7472635254423476,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.7106,
      "step": 5567
    },
    {
      "epoch": 1.747577386323512,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.6501,
      "step": 5568
    },
    {
      "epoch": 1.7478912472046764,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.7959,
      "step": 5569
    },
    {
      "epoch": 1.748205108085841,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.0345,
      "step": 5570
    },
    {
      "epoch": 1.7485189689670055,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.6343,
      "step": 5571
    },
    {
      "epoch": 1.74883282984817,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.6024,
      "step": 5572
    },
    {
      "epoch": 1.7491466907293343,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.9988,
      "step": 5573
    },
    {
      "epoch": 1.7494605516104986,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.8459,
      "step": 5574
    },
    {
      "epoch": 1.749774412491663,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.3739,
      "step": 5575
    },
    {
      "epoch": 1.7500882733728274,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.9286,
      "step": 5576
    },
    {
      "epoch": 1.750402134253992,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.9319,
      "step": 5577
    },
    {
      "epoch": 1.7507159951351563,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.9208,
      "step": 5578
    },
    {
      "epoch": 1.7510298560163209,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.7093,
      "step": 5579
    },
    {
      "epoch": 1.7513437168974852,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002,
      "loss": 2.2733,
      "step": 5580
    },
    {
      "epoch": 1.7516575777786496,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.5373,
      "step": 5581
    },
    {
      "epoch": 1.751971438659814,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.7424,
      "step": 5582
    },
    {
      "epoch": 1.7522852995409783,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.4287,
      "step": 5583
    },
    {
      "epoch": 1.752599160422143,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.7389,
      "step": 5584
    },
    {
      "epoch": 1.7529130213033073,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0002,
      "loss": 2.3242,
      "step": 5585
    },
    {
      "epoch": 1.7532268821844719,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0002,
      "loss": 2.4223,
      "step": 5586
    },
    {
      "epoch": 1.7535407430656362,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.4838,
      "step": 5587
    },
    {
      "epoch": 1.7538546039468006,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2177,
      "step": 5588
    },
    {
      "epoch": 1.754168464827965,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.4128,
      "step": 5589
    },
    {
      "epoch": 1.7544823257091293,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3305,
      "step": 5590
    },
    {
      "epoch": 1.754796186590294,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.3185,
      "step": 5591
    },
    {
      "epoch": 1.7551100474714583,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.2726,
      "step": 5592
    },
    {
      "epoch": 1.7554239083526229,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.2291,
      "step": 5593
    },
    {
      "epoch": 1.7557377692337872,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.4381,
      "step": 5594
    },
    {
      "epoch": 1.7560516301149516,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.4414,
      "step": 5595
    },
    {
      "epoch": 1.756365490996116,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.5047,
      "step": 5596
    },
    {
      "epoch": 1.7566793518772803,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.4359,
      "step": 5597
    },
    {
      "epoch": 1.7569932127584447,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.5671,
      "step": 5598
    },
    {
      "epoch": 1.7573070736396093,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.3196,
      "step": 5599
    },
    {
      "epoch": 1.7576209345207736,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.1978,
      "step": 5600
    },
    {
      "epoch": 1.7579347954019382,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.4638,
      "step": 5601
    },
    {
      "epoch": 1.7582486562831026,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.3435,
      "step": 5602
    },
    {
      "epoch": 1.758562517164267,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.3371,
      "step": 5603
    },
    {
      "epoch": 1.7588763780454313,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.3211,
      "step": 5604
    },
    {
      "epoch": 1.7591902389265957,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.3467,
      "step": 5605
    },
    {
      "epoch": 1.7595040998077602,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.3954,
      "step": 5606
    },
    {
      "epoch": 1.7598179606889246,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.1211,
      "step": 5607
    },
    {
      "epoch": 1.7601318215700892,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.5554,
      "step": 5608
    },
    {
      "epoch": 1.7604456824512535,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.6266,
      "step": 5609
    },
    {
      "epoch": 1.760759543332418,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.5075,
      "step": 5610
    },
    {
      "epoch": 1.760759543332418,
      "eval_loss": 1.781874656677246,
      "eval_runtime": 122.787,
      "eval_samples_per_second": 8.144,
      "eval_steps_per_second": 8.144,
      "step": 5610
    },
    {
      "epoch": 1.760759543332418,
      "mmlu_eval_accuracy": 0.4155586197243395,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3448275862068966,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
      "mmlu_eval_accuracy_high_school_psychology": 0.5666666666666667,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.5,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5697674418604651,
      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.38235294117647056,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
      "mmlu_eval_accuracy_professional_psychology": 0.3333333333333333,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.5789473684210527,
      "mmlu_loss": 1.0053661084210148,
      "step": 5610
    },
    {
      "epoch": 1.7610734042135823,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.615,
      "step": 5611
    },
    {
      "epoch": 1.7613872650947466,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.3273,
      "step": 5612
    },
    {
      "epoch": 1.7617011259759112,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.3637,
      "step": 5613
    },
    {
      "epoch": 1.7620149868570756,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.7726,
      "step": 5614
    },
    {
      "epoch": 1.7623288477382402,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.8462,
      "step": 5615
    },
    {
      "epoch": 1.7626427086194045,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.7213,
      "step": 5616
    },
    {
      "epoch": 1.762956569500569,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.8882,
      "step": 5617
    },
    {
      "epoch": 1.7632704303817333,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.8901,
      "step": 5618
    },
    {
      "epoch": 1.7635842912628976,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 2.0592,
      "step": 5619
    },
    {
      "epoch": 1.763898152144062,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.1989,
      "step": 5620
    },
    {
      "epoch": 1.7642120130252266,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.9823,
      "step": 5621
    },
    {
      "epoch": 1.7645258739063912,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.0021,
      "step": 5622
    },
    {
      "epoch": 1.7648397347875555,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.9681,
      "step": 5623
    },
    {
      "epoch": 1.7651535956687199,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.6639,
      "step": 5624
    },
    {
      "epoch": 1.7654674565498842,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.869,
      "step": 5625
    },
    {
      "epoch": 1.7657813174310486,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 1.9022,
      "step": 5626
    },
    {
      "epoch": 1.766095178312213,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.6985,
      "step": 5627
    },
    {
      "epoch": 1.7664090391933776,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 2.3216,
      "step": 5628
    },
    {
      "epoch": 1.766722900074542,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.9844,
      "step": 5629
    },
    {
      "epoch": 1.7670367609557065,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.9624,
      "step": 5630
    },
    {
      "epoch": 1.7673506218368709,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.2566,
      "step": 5631
    },
    {
      "epoch": 1.7676644827180352,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 2.3533,
      "step": 5632
    },
    {
      "epoch": 1.7679783435991996,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.6872,
      "step": 5633
    },
    {
      "epoch": 1.768292204480364,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.8253,
      "step": 5634
    },
    {
      "epoch": 1.7686060653615285,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.9717,
      "step": 5635
    },
    {
      "epoch": 1.768919926242693,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 2.7643,
      "step": 5636
    },
    {
      "epoch": 1.7692337871238575,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.428,
      "step": 5637
    },
    {
      "epoch": 1.7695476480050218,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3254,
      "step": 5638
    },
    {
      "epoch": 1.7698615088861862,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.3439,
      "step": 5639
    },
    {
      "epoch": 1.7701753697673506,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.3656,
      "step": 5640
    },
    {
      "epoch": 1.770489230648515,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.4807,
      "step": 5641
    },
    {
      "epoch": 1.7708030915296795,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.362,
      "step": 5642
    },
    {
      "epoch": 1.7711169524108439,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.2893,
      "step": 5643
    },
    {
      "epoch": 1.7714308132920085,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.459,
      "step": 5644
    },
    {
      "epoch": 1.7717446741731728,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.3325,
      "step": 5645
    },
    {
      "epoch": 1.7720585350543372,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.4091,
      "step": 5646
    },
    {
      "epoch": 1.7723723959355016,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.3117,
      "step": 5647
    },
    {
      "epoch": 1.772686256816666,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.3572,
      "step": 5648
    },
    {
      "epoch": 1.7730001176978303,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.4819,
      "step": 5649
    },
    {
      "epoch": 1.7733139785789949,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.3822,
      "step": 5650
    },
    {
      "epoch": 1.7736278394601594,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.3218,
      "step": 5651
    },
    {
      "epoch": 1.7739417003413238,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.3498,
      "step": 5652
    },
    {
      "epoch": 1.7742555612224882,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.3485,
      "step": 5653
    },
    {
      "epoch": 1.7745694221036525,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.3077,
      "step": 5654
    },
    {
      "epoch": 1.774883282984817,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.3087,
      "step": 5655
    },
    {
      "epoch": 1.7751971438659813,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.3835,
      "step": 5656
    },
    {
      "epoch": 1.7755110047471458,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.4421,
      "step": 5657
    },
    {
      "epoch": 1.7758248656283102,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.367,
      "step": 5658
    },
    {
      "epoch": 1.7761387265094748,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.5125,
      "step": 5659
    },
    {
      "epoch": 1.7764525873906392,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.2472,
      "step": 5660
    },
    {
      "epoch": 1.7767664482718035,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.3469,
      "step": 5661
    },
    {
      "epoch": 1.7770803091529679,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.6867,
      "step": 5662
    },
    {
      "epoch": 1.7773941700341322,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 2.0652,
      "step": 5663
    },
    {
      "epoch": 1.7777080309152968,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.7862,
      "step": 5664
    },
    {
      "epoch": 1.7780218917964612,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.673,
      "step": 5665
    },
    {
      "epoch": 1.7783357526776258,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.5529,
      "step": 5666
    },
    {
      "epoch": 1.7786496135587901,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 1.889,
      "step": 5667
    },
    {
      "epoch": 1.7789634744399545,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.9459,
      "step": 5668
    },
    {
      "epoch": 1.7792773353211189,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.9317,
      "step": 5669
    },
    {
      "epoch": 1.7795911962022832,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.8915,
      "step": 5670
    },
    {
      "epoch": 1.7799050570834478,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.8185,
      "step": 5671
    },
    {
      "epoch": 1.7802189179646122,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.9135,
      "step": 5672
    },
    {
      "epoch": 1.7805327788457768,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.6733,
      "step": 5673
    },
    {
      "epoch": 1.7808466397269411,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.1443,
      "step": 5674
    },
    {
      "epoch": 1.7811605006081055,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.6653,
      "step": 5675
    },
    {
      "epoch": 1.7814743614892699,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.7433,
      "step": 5676
    },
    {
      "epoch": 1.7817882223704342,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.5781,
      "step": 5677
    },
    {
      "epoch": 1.7821020832515986,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 2.1697,
      "step": 5678
    },
    {
      "epoch": 1.7824159441327632,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 2.0052,
      "step": 5679
    },
    {
      "epoch": 1.7827298050139275,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.8674,
      "step": 5680
    },
    {
      "epoch": 1.7830436658950921,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.6011,
      "step": 5681
    },
    {
      "epoch": 1.7833575267762565,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.8661,
      "step": 5682
    },
    {
      "epoch": 1.7836713876574208,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 2.1494,
      "step": 5683
    },
    {
      "epoch": 1.7839852485385852,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.9949,
      "step": 5684
    },
    {
      "epoch": 1.7842991094197496,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.9779,
      "step": 5685
    },
    {
      "epoch": 1.7846129703009141,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0002,
      "loss": 2.2549,
      "step": 5686
    },
    {
      "epoch": 1.7849268311820785,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.7224,
      "step": 5687
    },
    {
      "epoch": 1.785240692063243,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0861,
      "step": 5688
    },
    {
      "epoch": 1.7855545529444075,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.3832,
      "step": 5689
    },
    {
      "epoch": 1.7858684138255718,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3256,
      "step": 5690
    },
    {
      "epoch": 1.7861822747067362,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.508,
      "step": 5691
    },
    {
      "epoch": 1.7864961355879005,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.3166,
      "step": 5692
    },
    {
      "epoch": 1.7868099964690651,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.2937,
      "step": 5693
    },
    {
      "epoch": 1.7871238573502295,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.3582,
      "step": 5694
    },
    {
      "epoch": 1.787437718231394,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.3818,
      "step": 5695
    },
    {
      "epoch": 1.7877515791125584,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.5635,
      "step": 5696
    },
    {
      "epoch": 1.7880654399937228,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.3364,
      "step": 5697
    },
    {
      "epoch": 1.7883793008748872,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.3078,
      "step": 5698
    },
    {
      "epoch": 1.7886931617560515,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.4968,
      "step": 5699
    },
    {
      "epoch": 1.789007022637216,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3104,
      "step": 5700
    },
    {
      "epoch": 1.7893208835183805,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.3153,
      "step": 5701
    },
    {
      "epoch": 1.789634744399545,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.3626,
      "step": 5702
    },
    {
      "epoch": 1.7899486052807094,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.3026,
      "step": 5703
    },
    {
      "epoch": 1.7902624661618738,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.2832,
      "step": 5704
    },
    {
      "epoch": 1.7905763270430382,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.2591,
      "step": 5705
    },
    {
      "epoch": 1.7908901879242025,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.3971,
      "step": 5706
    },
    {
      "epoch": 1.7912040488053669,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.5718,
      "step": 5707
    },
    {
      "epoch": 1.7915179096865315,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.195,
      "step": 5708
    },
    {
      "epoch": 1.7918317705676958,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2835,
      "step": 5709
    },
    {
      "epoch": 1.7921456314488604,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.4917,
      "step": 5710
    },
    {
      "epoch": 1.7924594923300248,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.5119,
      "step": 5711
    },
    {
      "epoch": 1.7927733532111891,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.7028,
      "step": 5712
    },
    {
      "epoch": 1.7930872140923535,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.6387,
      "step": 5713
    },
    {
      "epoch": 1.7934010749735179,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.6766,
      "step": 5714
    },
    {
      "epoch": 1.7937149358546824,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.3949,
      "step": 5715
    },
    {
      "epoch": 1.7940287967358468,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.7873,
      "step": 5716
    },
    {
      "epoch": 1.7943426576170114,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.6646,
      "step": 5717
    },
    {
      "epoch": 1.7946565184981758,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.5148,
      "step": 5718
    },
    {
      "epoch": 1.7949703793793401,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.6509,
      "step": 5719
    },
    {
      "epoch": 1.7952842402605045,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 2.1718,
      "step": 5720
    },
    {
      "epoch": 1.7955981011416688,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.6369,
      "step": 5721
    },
    {
      "epoch": 1.7959119620228334,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.6852,
      "step": 5722
    },
    {
      "epoch": 1.7962258229039978,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.8289,
      "step": 5723
    },
    {
      "epoch": 1.7965396837851624,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.9634,
      "step": 5724
    },
    {
      "epoch": 1.7968535446663267,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 1.9003,
      "step": 5725
    },
    {
      "epoch": 1.797167405547491,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 2.0595,
      "step": 5726
    },
    {
      "epoch": 1.7974812664286555,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.8105,
      "step": 5727
    },
    {
      "epoch": 1.7977951273098198,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.857,
      "step": 5728
    },
    {
      "epoch": 1.7981089881909842,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.4325,
      "step": 5729
    },
    {
      "epoch": 1.7984228490721488,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002,
      "loss": 2.6919,
      "step": 5730
    },
    {
      "epoch": 1.7987367099533131,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.9036,
      "step": 5731
    },
    {
      "epoch": 1.7990505708344777,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.8962,
      "step": 5732
    },
    {
      "epoch": 1.799364431715642,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.7343,
      "step": 5733
    },
    {
      "epoch": 1.7996782925968065,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.793,
      "step": 5734
    },
    {
      "epoch": 1.7999921534779708,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.994,
      "step": 5735
    },
    {
      "epoch": 1.8003060143591352,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 2.2049,
      "step": 5736
    },
    {
      "epoch": 1.8006198752402998,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.4931,
      "step": 5737
    },
    {
      "epoch": 1.8009337361214641,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3133,
      "step": 5738
    },
    {
      "epoch": 1.8012475970026287,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2941,
      "step": 5739
    },
    {
      "epoch": 1.801561457883793,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.3704,
      "step": 5740
    },
    {
      "epoch": 1.8018753187649574,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.5015,
      "step": 5741
    },
    {
      "epoch": 1.8021891796461218,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.4961,
      "step": 5742
    },
    {
      "epoch": 1.8025030405272862,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.4393,
      "step": 5743
    },
    {
      "epoch": 1.8028169014084507,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3249,
      "step": 5744
    },
    {
      "epoch": 1.803130762289615,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.3936,
      "step": 5745
    },
    {
      "epoch": 1.8034446231707797,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.1797,
      "step": 5746
    },
    {
      "epoch": 1.803758484051944,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.3464,
      "step": 5747
    },
    {
      "epoch": 1.8040723449331084,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.3781,
      "step": 5748
    },
    {
      "epoch": 1.8043862058142728,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.3861,
      "step": 5749
    },
    {
      "epoch": 1.8047000666954371,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.4047,
      "step": 5750
    },
    {
      "epoch": 1.8050139275766015,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.4029,
      "step": 5751
    },
    {
      "epoch": 1.805327788457766,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.3544,
      "step": 5752
    },
    {
      "epoch": 1.8056416493389307,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.517,
      "step": 5753
    },
    {
      "epoch": 1.805955510220095,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.7595,
      "step": 5754
    },
    {
      "epoch": 1.8062693711012594,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.2078,
      "step": 5755
    },
    {
      "epoch": 1.8065832319824238,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.5419,
      "step": 5756
    },
    {
      "epoch": 1.8068970928635881,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.236,
      "step": 5757
    },
    {
      "epoch": 1.8072109537447525,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3237,
      "step": 5758
    },
    {
      "epoch": 1.807524814625917,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.3794,
      "step": 5759
    },
    {
      "epoch": 1.8078386755070814,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.528,
      "step": 5760
    },
    {
      "epoch": 1.808152536388246,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.4909,
      "step": 5761
    },
    {
      "epoch": 1.8084663972694104,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.7566,
      "step": 5762
    },
    {
      "epoch": 1.8087802581505747,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.8292,
      "step": 5763
    },
    {
      "epoch": 1.8090941190317391,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.6946,
      "step": 5764
    },
    {
      "epoch": 1.8094079799129035,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.8144,
      "step": 5765
    },
    {
      "epoch": 1.809721840794068,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 2.0289,
      "step": 5766
    },
    {
      "epoch": 1.8100357016752324,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.7732,
      "step": 5767
    },
    {
      "epoch": 1.810349562556397,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.6602,
      "step": 5768
    },
    {
      "epoch": 1.8106634234375614,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 2.0523,
      "step": 5769
    },
    {
      "epoch": 1.8109772843187257,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.8063,
      "step": 5770
    },
    {
      "epoch": 1.81129114519989,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.9525,
      "step": 5771
    },
    {
      "epoch": 1.8116050060810545,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.7752,
      "step": 5772
    },
    {
      "epoch": 1.811918866962219,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.8529,
      "step": 5773
    },
    {
      "epoch": 1.8122327278433834,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.8923,
      "step": 5774
    },
    {
      "epoch": 1.812546588724548,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.8144,
      "step": 5775
    },
    {
      "epoch": 1.8128604496057124,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 2.1445,
      "step": 5776
    },
    {
      "epoch": 1.8131743104868767,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.9996,
      "step": 5777
    },
    {
      "epoch": 1.813488171368041,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.8569,
      "step": 5778
    },
    {
      "epoch": 1.8138020322492054,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.1226,
      "step": 5779
    },
    {
      "epoch": 1.8141158931303698,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 2.2404,
      "step": 5780
    },
    {
      "epoch": 1.8144297540115344,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.7292,
      "step": 5781
    },
    {
      "epoch": 1.814743614892699,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.7762,
      "step": 5782
    },
    {
      "epoch": 1.8150574757738633,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.6273,
      "step": 5783
    },
    {
      "epoch": 1.8153713366550277,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.7332,
      "step": 5784
    },
    {
      "epoch": 1.815685197536192,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.7667,
      "step": 5785
    },
    {
      "epoch": 1.8159990584173564,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 2.1648,
      "step": 5786
    },
    {
      "epoch": 1.8163129192985208,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.4316,
      "step": 5787
    },
    {
      "epoch": 1.8166267801796854,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1858,
      "step": 5788
    },
    {
      "epoch": 1.8169406410608497,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4101,
      "step": 5789
    },
    {
      "epoch": 1.8172545019420143,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.5071,
      "step": 5790
    },
    {
      "epoch": 1.8175683628231787,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2818,
      "step": 5791
    },
    {
      "epoch": 1.817882223704343,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.3911,
      "step": 5792
    },
    {
      "epoch": 1.8181960845855074,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.177,
      "step": 5793
    },
    {
      "epoch": 1.8185099454666718,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.3634,
      "step": 5794
    },
    {
      "epoch": 1.8188238063478364,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.3598,
      "step": 5795
    },
    {
      "epoch": 1.8191376672290007,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.3231,
      "step": 5796
    },
    {
      "epoch": 1.8194515281101653,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.4335,
      "step": 5797
    },
    {
      "epoch": 1.8194515281101653,
      "eval_loss": 1.782687783241272,
      "eval_runtime": 123.7187,
      "eval_samples_per_second": 8.083,
      "eval_steps_per_second": 8.083,
      "step": 5797
    },
    {
      "epoch": 1.8194515281101653,
      "mmlu_eval_accuracy": 0.41312694620958684,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.0,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.3125,
      "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": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.6333333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_world_history": 0.4230769230769231,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5116279069767442,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.32,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.4117647058823529,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "mmlu_eval_accuracy_professional_medicine": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
      "mmlu_eval_accuracy_sociology": 0.4090909090909091,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.114936444151643,
      "step": 5797
    },
    {
      "epoch": 1.8197653889913297,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.3649,
      "step": 5798
    },
    {
      "epoch": 1.820079249872494,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.288,
      "step": 5799
    },
    {
      "epoch": 1.8203931107536584,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.3097,
      "step": 5800
    },
    {
      "epoch": 1.8207069716348228,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.3494,
      "step": 5801
    },
    {
      "epoch": 1.8210208325159873,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.3934,
      "step": 5802
    },
    {
      "epoch": 1.8213346933971517,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.3018,
      "step": 5803
    },
    {
      "epoch": 1.8216485542783163,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.1474,
      "step": 5804
    },
    {
      "epoch": 1.8219624151594807,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.4874,
      "step": 5805
    },
    {
      "epoch": 1.822276276040645,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.2835,
      "step": 5806
    },
    {
      "epoch": 1.8225901369218094,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.4347,
      "step": 5807
    },
    {
      "epoch": 1.8229039978029737,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.4818,
      "step": 5808
    },
    {
      "epoch": 1.823217858684138,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.4124,
      "step": 5809
    },
    {
      "epoch": 1.8235317195653027,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.3921,
      "step": 5810
    },
    {
      "epoch": 1.823845580446467,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.56,
      "step": 5811
    },
    {
      "epoch": 1.8241594413276316,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.5578,
      "step": 5812
    },
    {
      "epoch": 1.824473302208796,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.551,
      "step": 5813
    },
    {
      "epoch": 1.8247871630899604,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.7328,
      "step": 5814
    },
    {
      "epoch": 1.8251010239711247,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.7849,
      "step": 5815
    },
    {
      "epoch": 1.825414884852289,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.6976,
      "step": 5816
    },
    {
      "epoch": 1.8257287457334537,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 2.0065,
      "step": 5817
    },
    {
      "epoch": 1.826042606614618,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.7382,
      "step": 5818
    },
    {
      "epoch": 1.8263564674957826,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.6435,
      "step": 5819
    },
    {
      "epoch": 1.826670328376947,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 2.4371,
      "step": 5820
    },
    {
      "epoch": 1.8269841892581113,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.9233,
      "step": 5821
    },
    {
      "epoch": 1.8272980501392757,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.8327,
      "step": 5822
    },
    {
      "epoch": 1.82761191102044,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 2.1253,
      "step": 5823
    },
    {
      "epoch": 1.8279257719016047,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.7072,
      "step": 5824
    },
    {
      "epoch": 1.828239632782769,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.7254,
      "step": 5825
    },
    {
      "epoch": 1.8285534936639336,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 2.0108,
      "step": 5826
    },
    {
      "epoch": 1.828867354545098,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 2.0992,
      "step": 5827
    },
    {
      "epoch": 1.8291812154262623,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 2.0665,
      "step": 5828
    },
    {
      "epoch": 1.8294950763074267,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.5506,
      "step": 5829
    },
    {
      "epoch": 1.829808937188591,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.9926,
      "step": 5830
    },
    {
      "epoch": 1.8301227980697554,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.9656,
      "step": 5831
    },
    {
      "epoch": 1.83043665895092,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.7434,
      "step": 5832
    },
    {
      "epoch": 1.8307505198320846,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.6967,
      "step": 5833
    },
    {
      "epoch": 1.831064380713249,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.7366,
      "step": 5834
    },
    {
      "epoch": 1.8313782415944133,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.6125,
      "step": 5835
    },
    {
      "epoch": 1.8316921024755777,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002,
      "loss": 2.7015,
      "step": 5836
    },
    {
      "epoch": 1.832005963356742,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.7805,
      "step": 5837
    },
    {
      "epoch": 1.8323198242379064,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.3908,
      "step": 5838
    },
    {
      "epoch": 1.832633685119071,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.4512,
      "step": 5839
    },
    {
      "epoch": 1.8329475460002354,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.287,
      "step": 5840
    },
    {
      "epoch": 1.8332614068814,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.6453,
      "step": 5841
    },
    {
      "epoch": 1.8335752677625643,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2642,
      "step": 5842
    },
    {
      "epoch": 1.8338891286437287,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3075,
      "step": 5843
    },
    {
      "epoch": 1.834202989524893,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.2219,
      "step": 5844
    },
    {
      "epoch": 1.8345168504060574,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.5635,
      "step": 5845
    },
    {
      "epoch": 1.834830711287222,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.3794,
      "step": 5846
    },
    {
      "epoch": 1.8351445721683863,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.5997,
      "step": 5847
    },
    {
      "epoch": 1.835458433049551,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.362,
      "step": 5848
    },
    {
      "epoch": 1.8357722939307153,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.3858,
      "step": 5849
    },
    {
      "epoch": 1.8360861548118796,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.3161,
      "step": 5850
    },
    {
      "epoch": 1.836400015693044,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.4395,
      "step": 5851
    },
    {
      "epoch": 1.8367138765742084,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.3102,
      "step": 5852
    },
    {
      "epoch": 1.837027737455373,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.3069,
      "step": 5853
    },
    {
      "epoch": 1.8373415983365373,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.4019,
      "step": 5854
    },
    {
      "epoch": 1.837655459217702,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.3886,
      "step": 5855
    },
    {
      "epoch": 1.8379693200988663,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.405,
      "step": 5856
    },
    {
      "epoch": 1.8382831809800306,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4266,
      "step": 5857
    },
    {
      "epoch": 1.838597041861195,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.5111,
      "step": 5858
    },
    {
      "epoch": 1.8389109027423594,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.378,
      "step": 5859
    },
    {
      "epoch": 1.8392247636235237,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.4416,
      "step": 5860
    },
    {
      "epoch": 1.8395386245046883,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.5043,
      "step": 5861
    },
    {
      "epoch": 1.8398524853858529,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.2472,
      "step": 5862
    },
    {
      "epoch": 1.8401663462670172,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.3677,
      "step": 5863
    },
    {
      "epoch": 1.8404802071481816,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.6571,
      "step": 5864
    },
    {
      "epoch": 1.840794068029346,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.6922,
      "step": 5865
    },
    {
      "epoch": 1.8411079289105103,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.5887,
      "step": 5866
    },
    {
      "epoch": 1.8414217897916747,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.7617,
      "step": 5867
    },
    {
      "epoch": 1.8417356506728393,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.8468,
      "step": 5868
    },
    {
      "epoch": 1.8420495115540036,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.8676,
      "step": 5869
    },
    {
      "epoch": 1.8423633724351682,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 2.0462,
      "step": 5870
    },
    {
      "epoch": 1.8426772333163326,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.8405,
      "step": 5871
    },
    {
      "epoch": 1.842991094197497,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.8106,
      "step": 5872
    },
    {
      "epoch": 1.8433049550786613,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.8007,
      "step": 5873
    },
    {
      "epoch": 1.8436188159598257,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 2.1436,
      "step": 5874
    },
    {
      "epoch": 1.8439326768409903,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 2.1204,
      "step": 5875
    },
    {
      "epoch": 1.8442465377221546,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 2.3496,
      "step": 5876
    },
    {
      "epoch": 1.8445603986033192,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.5854,
      "step": 5877
    },
    {
      "epoch": 1.8448742594844836,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.7404,
      "step": 5878
    },
    {
      "epoch": 1.845188120365648,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.9137,
      "step": 5879
    },
    {
      "epoch": 1.8455019812468123,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.9283,
      "step": 5880
    },
    {
      "epoch": 1.8458158421279767,
      "grad_norm": 0.953125,
      "learning_rate": 0.0002,
      "loss": 2.3969,
      "step": 5881
    },
    {
      "epoch": 1.846129703009141,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.6984,
      "step": 5882
    },
    {
      "epoch": 1.8464435638903056,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.6577,
      "step": 5883
    },
    {
      "epoch": 1.8467574247714702,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 2.3698,
      "step": 5884
    },
    {
      "epoch": 1.8470712856526346,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002,
      "loss": 2.6423,
      "step": 5885
    },
    {
      "epoch": 1.847385146533799,
      "grad_norm": 0.84375,
      "learning_rate": 0.0002,
      "loss": 2.5797,
      "step": 5886
    },
    {
      "epoch": 1.8476990074149633,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.6131,
      "step": 5887
    },
    {
      "epoch": 1.8480128682961277,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.3824,
      "step": 5888
    },
    {
      "epoch": 1.848326729177292,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.3644,
      "step": 5889
    },
    {
      "epoch": 1.8486405900584566,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.4024,
      "step": 5890
    },
    {
      "epoch": 1.848954450939621,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.3604,
      "step": 5891
    },
    {
      "epoch": 1.8492683118207855,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.3064,
      "step": 5892
    },
    {
      "epoch": 1.84958217270195,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.2609,
      "step": 5893
    },
    {
      "epoch": 1.8498960335831143,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.2902,
      "step": 5894
    },
    {
      "epoch": 1.8502098944642786,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3162,
      "step": 5895
    },
    {
      "epoch": 1.850523755345443,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.3751,
      "step": 5896
    },
    {
      "epoch": 1.8508376162266076,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.468,
      "step": 5897
    },
    {
      "epoch": 1.851151477107772,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3809,
      "step": 5898
    },
    {
      "epoch": 1.8514653379889365,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.2975,
      "step": 5899
    },
    {
      "epoch": 1.851779198870101,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.5673,
      "step": 5900
    },
    {
      "epoch": 1.8520930597512653,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.3654,
      "step": 5901
    },
    {
      "epoch": 1.8524069206324296,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.3557,
      "step": 5902
    },
    {
      "epoch": 1.852720781513594,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.2748,
      "step": 5903
    },
    {
      "epoch": 1.8530346423947586,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.3098,
      "step": 5904
    },
    {
      "epoch": 1.853348503275923,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.1865,
      "step": 5905
    },
    {
      "epoch": 1.8536623641570875,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.4136,
      "step": 5906
    },
    {
      "epoch": 1.8539762250382519,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.4703,
      "step": 5907
    },
    {
      "epoch": 1.8542900859194162,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.1209,
      "step": 5908
    },
    {
      "epoch": 1.8546039468005806,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.3871,
      "step": 5909
    },
    {
      "epoch": 1.854917807681745,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.6733,
      "step": 5910
    },
    {
      "epoch": 1.8552316685629093,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.4729,
      "step": 5911
    },
    {
      "epoch": 1.855545529444074,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.6418,
      "step": 5912
    },
    {
      "epoch": 1.8558593903252385,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.572,
      "step": 5913
    },
    {
      "epoch": 1.8561732512064029,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.5443,
      "step": 5914
    },
    {
      "epoch": 1.8564871120875672,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.6224,
      "step": 5915
    },
    {
      "epoch": 1.8568009729687316,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 1.7739,
      "step": 5916
    },
    {
      "epoch": 1.857114833849896,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.5957,
      "step": 5917
    },
    {
      "epoch": 1.8574286947310603,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.7727,
      "step": 5918
    },
    {
      "epoch": 1.857742555612225,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.8576,
      "step": 5919
    },
    {
      "epoch": 1.8580564164933893,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.9285,
      "step": 5920
    },
    {
      "epoch": 1.8583702773745538,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.8665,
      "step": 5921
    },
    {
      "epoch": 1.8586841382557182,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.9875,
      "step": 5922
    },
    {
      "epoch": 1.8589979991368826,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.8197,
      "step": 5923
    },
    {
      "epoch": 1.859311860018047,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.8642,
      "step": 5924
    },
    {
      "epoch": 1.8596257208992113,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.4036,
      "step": 5925
    },
    {
      "epoch": 1.8599395817803759,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.5666,
      "step": 5926
    },
    {
      "epoch": 1.8602534426615402,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 2.2278,
      "step": 5927
    },
    {
      "epoch": 1.8605673035427048,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.8289,
      "step": 5928
    },
    {
      "epoch": 1.8608811644238692,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 2.0646,
      "step": 5929
    },
    {
      "epoch": 1.8611950253050336,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.6013,
      "step": 5930
    },
    {
      "epoch": 1.861508886186198,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.7115,
      "step": 5931
    },
    {
      "epoch": 1.8618227470673623,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 2.0013,
      "step": 5932
    },
    {
      "epoch": 1.8621366079485269,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.7879,
      "step": 5933
    },
    {
      "epoch": 1.8624504688296912,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.1872,
      "step": 5934
    },
    {
      "epoch": 1.8627643297108558,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 2.1315,
      "step": 5935
    },
    {
      "epoch": 1.8630781905920202,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 2.6237,
      "step": 5936
    },
    {
      "epoch": 1.8633920514731845,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.8178,
      "step": 5937
    },
    {
      "epoch": 1.863705912354349,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2998,
      "step": 5938
    },
    {
      "epoch": 1.8640197732355133,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.3782,
      "step": 5939
    },
    {
      "epoch": 1.8643336341166776,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.399,
      "step": 5940
    },
    {
      "epoch": 1.8646474949978422,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2741,
      "step": 5941
    },
    {
      "epoch": 1.8649613558790066,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3048,
      "step": 5942
    },
    {
      "epoch": 1.8652752167601712,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.3547,
      "step": 5943
    },
    {
      "epoch": 1.8655890776413355,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.2876,
      "step": 5944
    },
    {
      "epoch": 1.8659029385224999,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.3567,
      "step": 5945
    },
    {
      "epoch": 1.8662167994036643,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.4003,
      "step": 5946
    },
    {
      "epoch": 1.8665306602848286,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.2482,
      "step": 5947
    },
    {
      "epoch": 1.8668445211659932,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.3462,
      "step": 5948
    },
    {
      "epoch": 1.8671583820471576,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.3661,
      "step": 5949
    },
    {
      "epoch": 1.8674722429283221,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.3655,
      "step": 5950
    },
    {
      "epoch": 1.8677861038094865,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.3092,
      "step": 5951
    },
    {
      "epoch": 1.8680999646906509,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.5488,
      "step": 5952
    },
    {
      "epoch": 1.8684138255718152,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.3023,
      "step": 5953
    },
    {
      "epoch": 1.8687276864529796,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.4053,
      "step": 5954
    },
    {
      "epoch": 1.8690415473341442,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.6521,
      "step": 5955
    },
    {
      "epoch": 1.8693554082153085,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.4736,
      "step": 5956
    },
    {
      "epoch": 1.8696692690964731,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.253,
      "step": 5957
    },
    {
      "epoch": 1.8699831299776375,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.459,
      "step": 5958
    },
    {
      "epoch": 1.8702969908588019,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.6717,
      "step": 5959
    },
    {
      "epoch": 1.8706108517399662,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.657,
      "step": 5960
    },
    {
      "epoch": 1.8709247126211306,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.5178,
      "step": 5961
    },
    {
      "epoch": 1.871238573502295,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.833,
      "step": 5962
    },
    {
      "epoch": 1.8715524343834595,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.6718,
      "step": 5963
    },
    {
      "epoch": 1.8718662952646241,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.6281,
      "step": 5964
    },
    {
      "epoch": 1.8721801561457885,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.4548,
      "step": 5965
    },
    {
      "epoch": 1.8724940170269528,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.3447,
      "step": 5966
    },
    {
      "epoch": 1.8728078779081172,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 2.0358,
      "step": 5967
    },
    {
      "epoch": 1.8731217387892816,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 2.1333,
      "step": 5968
    },
    {
      "epoch": 1.873435599670446,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.695,
      "step": 5969
    },
    {
      "epoch": 1.8737494605516105,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.848,
      "step": 5970
    },
    {
      "epoch": 1.8740633214327749,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 2.0354,
      "step": 5971
    },
    {
      "epoch": 1.8743771823139395,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.0644,
      "step": 5972
    },
    {
      "epoch": 1.8746910431951038,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.7386,
      "step": 5973
    },
    {
      "epoch": 1.8750049040762682,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.1882,
      "step": 5974
    },
    {
      "epoch": 1.8753187649574325,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.6669,
      "step": 5975
    },
    {
      "epoch": 1.875632625838597,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 2.5656,
      "step": 5976
    },
    {
      "epoch": 1.8759464867197615,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.1146,
      "step": 5977
    },
    {
      "epoch": 1.8762603476009259,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.2268,
      "step": 5978
    },
    {
      "epoch": 1.8765742084820904,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 2.1216,
      "step": 5979
    },
    {
      "epoch": 1.8768880693632548,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.2109,
      "step": 5980
    },
    {
      "epoch": 1.8772019302444192,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 2.2491,
      "step": 5981
    },
    {
      "epoch": 1.8775157911255835,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.7636,
      "step": 5982
    },
    {
      "epoch": 1.877829652006748,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.1231,
      "step": 5983
    },
    {
      "epoch": 1.8781435128879125,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.7568,
      "step": 5984
    },
    {
      "epoch": 1.8781435128879125,
      "eval_loss": 1.7855916023254395,
      "eval_runtime": 123.5616,
      "eval_samples_per_second": 8.093,
      "eval_steps_per_second": 8.093,
      "step": 5984
    },
    {
      "epoch": 1.8781435128879125,
      "mmlu_eval_accuracy": 0.42423091770400356,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.3125,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "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": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5,
      "mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "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": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
      "mmlu_eval_accuracy_miscellaneous": 0.5813953488372093,
      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
      "mmlu_eval_accuracy_moral_scenarios": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.5151515151515151,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.4090909090909091,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.3333333333333333,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.1574250328665383,
      "step": 5984
    },
    {
      "epoch": 1.8784573737690768,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 2.0926,
      "step": 5985
    },
    {
      "epoch": 1.8787712346502414,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 2.523,
      "step": 5986
    },
    {
      "epoch": 1.8790850955314058,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.4967,
      "step": 5987
    },
    {
      "epoch": 1.8793989564125702,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2295,
      "step": 5988
    },
    {
      "epoch": 1.8797128172937345,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2173,
      "step": 5989
    },
    {
      "epoch": 1.8800266781748989,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.5014,
      "step": 5990
    },
    {
      "epoch": 1.8803405390560632,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.3897,
      "step": 5991
    },
    {
      "epoch": 1.8806543999372278,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.1561,
      "step": 5992
    },
    {
      "epoch": 1.8809682608183924,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.3402,
      "step": 5993
    },
    {
      "epoch": 1.8812821216995568,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.4062,
      "step": 5994
    },
    {
      "epoch": 1.8815959825807211,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.4649,
      "step": 5995
    },
    {
      "epoch": 1.8819098434618855,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.3717,
      "step": 5996
    },
    {
      "epoch": 1.8822237043430499,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.4346,
      "step": 5997
    },
    {
      "epoch": 1.8825375652242142,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.3202,
      "step": 5998
    },
    {
      "epoch": 1.8828514261053788,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.3707,
      "step": 5999
    },
    {
      "epoch": 1.8831652869865432,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.3233,
      "step": 6000
    },
    {
      "epoch": 1.8834791478677078,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.2141,
      "step": 6001
    },
    {
      "epoch": 1.8837930087488721,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.3269,
      "step": 6002
    },
    {
      "epoch": 1.8841068696300365,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.3923,
      "step": 6003
    },
    {
      "epoch": 1.8844207305112008,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.2323,
      "step": 6004
    },
    {
      "epoch": 1.8847345913923652,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.424,
      "step": 6005
    },
    {
      "epoch": 1.8850484522735298,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.5137,
      "step": 6006
    },
    {
      "epoch": 1.8853623131546942,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.5626,
      "step": 6007
    },
    {
      "epoch": 1.8856761740358587,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.5102,
      "step": 6008
    },
    {
      "epoch": 1.885990034917023,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.2722,
      "step": 6009
    },
    {
      "epoch": 1.8863038957981875,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.6836,
      "step": 6010
    },
    {
      "epoch": 1.8866177566793518,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.611,
      "step": 6011
    },
    {
      "epoch": 1.8869316175605162,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.677,
      "step": 6012
    },
    {
      "epoch": 1.8872454784416808,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 2.1169,
      "step": 6013
    },
    {
      "epoch": 1.8875593393228451,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.4547,
      "step": 6014
    },
    {
      "epoch": 1.8878732002040097,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 1.7594,
      "step": 6015
    },
    {
      "epoch": 1.888187061085174,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.4907,
      "step": 6016
    },
    {
      "epoch": 1.8885009219663385,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.6983,
      "step": 6017
    },
    {
      "epoch": 1.8888147828475028,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.7239,
      "step": 6018
    },
    {
      "epoch": 1.8891286437286672,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.9725,
      "step": 6019
    },
    {
      "epoch": 1.8894425046098315,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.842,
      "step": 6020
    },
    {
      "epoch": 1.8897563654909961,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.9049,
      "step": 6021
    },
    {
      "epoch": 1.8900702263721605,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 1.6995,
      "step": 6022
    },
    {
      "epoch": 1.890384087253325,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002,
      "loss": 2.1398,
      "step": 6023
    },
    {
      "epoch": 1.8906979481344894,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 2.1568,
      "step": 6024
    },
    {
      "epoch": 1.8910118090156538,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002,
      "loss": 2.4192,
      "step": 6025
    },
    {
      "epoch": 1.8913256698968182,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002,
      "loss": 2.1758,
      "step": 6026
    },
    {
      "epoch": 1.8916395307779825,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.8815,
      "step": 6027
    },
    {
      "epoch": 1.891953391659147,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.5712,
      "step": 6028
    },
    {
      "epoch": 1.8922672525403115,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 2.0794,
      "step": 6029
    },
    {
      "epoch": 1.892581113421476,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 1.7872,
      "step": 6030
    },
    {
      "epoch": 1.8928949743026404,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.3885,
      "step": 6031
    },
    {
      "epoch": 1.8932088351838048,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.874,
      "step": 6032
    },
    {
      "epoch": 1.8935226960649691,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.9045,
      "step": 6033
    },
    {
      "epoch": 1.8938365569461335,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.6468,
      "step": 6034
    },
    {
      "epoch": 1.894150417827298,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.8801,
      "step": 6035
    },
    {
      "epoch": 1.8944642787084625,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002,
      "loss": 2.6754,
      "step": 6036
    },
    {
      "epoch": 1.894778139589627,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.2811,
      "step": 6037
    },
    {
      "epoch": 1.8950920004707914,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2129,
      "step": 6038
    },
    {
      "epoch": 1.8954058613519558,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.4311,
      "step": 6039
    },
    {
      "epoch": 1.8957197222331201,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2499,
      "step": 6040
    },
    {
      "epoch": 1.8960335831142845,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.3589,
      "step": 6041
    },
    {
      "epoch": 1.8963474439954489,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.2864,
      "step": 6042
    },
    {
      "epoch": 1.8966613048766134,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.5288,
      "step": 6043
    },
    {
      "epoch": 1.896975165757778,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.2375,
      "step": 6044
    },
    {
      "epoch": 1.8972890266389424,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3652,
      "step": 6045
    },
    {
      "epoch": 1.8976028875201068,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.3556,
      "step": 6046
    },
    {
      "epoch": 1.8979167484012711,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2775,
      "step": 6047
    },
    {
      "epoch": 1.8982306092824355,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.3814,
      "step": 6048
    },
    {
      "epoch": 1.8985444701635998,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.4055,
      "step": 6049
    },
    {
      "epoch": 1.8988583310447644,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.3125,
      "step": 6050
    },
    {
      "epoch": 1.8991721919259288,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.3607,
      "step": 6051
    },
    {
      "epoch": 1.8994860528070934,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.3836,
      "step": 6052
    },
    {
      "epoch": 1.8997999136882577,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.3888,
      "step": 6053
    },
    {
      "epoch": 1.900113774569422,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.2514,
      "step": 6054
    },
    {
      "epoch": 1.9004276354505865,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.4204,
      "step": 6055
    },
    {
      "epoch": 1.9007414963317508,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.3758,
      "step": 6056
    },
    {
      "epoch": 1.9010553572129154,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.2802,
      "step": 6057
    },
    {
      "epoch": 1.9013692180940798,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.3027,
      "step": 6058
    },
    {
      "epoch": 1.9016830789752444,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.3404,
      "step": 6059
    },
    {
      "epoch": 1.9019969398564087,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.1739,
      "step": 6060
    },
    {
      "epoch": 1.902310800737573,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.4307,
      "step": 6061
    },
    {
      "epoch": 1.9026246616187374,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.3368,
      "step": 6062
    },
    {
      "epoch": 1.9029385224999018,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.5696,
      "step": 6063
    },
    {
      "epoch": 1.9032523833810664,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.3868,
      "step": 6064
    },
    {
      "epoch": 1.9035662442622308,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.7312,
      "step": 6065
    },
    {
      "epoch": 1.9038801051433953,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.6214,
      "step": 6066
    },
    {
      "epoch": 1.9041939660245597,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.6447,
      "step": 6067
    },
    {
      "epoch": 1.904507826905724,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.844,
      "step": 6068
    },
    {
      "epoch": 1.9048216877868884,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.6066,
      "step": 6069
    },
    {
      "epoch": 1.9051355486680528,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.8824,
      "step": 6070
    },
    {
      "epoch": 1.9054494095492172,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 1.9519,
      "step": 6071
    },
    {
      "epoch": 1.9057632704303817,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.7469,
      "step": 6072
    },
    {
      "epoch": 1.906077131311546,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.9807,
      "step": 6073
    },
    {
      "epoch": 1.9063909921927107,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 2.1678,
      "step": 6074
    },
    {
      "epoch": 1.906704853073875,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 2.1662,
      "step": 6075
    },
    {
      "epoch": 1.9070187139550394,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.9905,
      "step": 6076
    },
    {
      "epoch": 1.9073325748362038,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 2.1265,
      "step": 6077
    },
    {
      "epoch": 1.9076464357173681,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 2.4555,
      "step": 6078
    },
    {
      "epoch": 1.9079602965985327,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 2.7651,
      "step": 6079
    },
    {
      "epoch": 1.908274157479697,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.8466,
      "step": 6080
    },
    {
      "epoch": 1.9085880183608617,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.9321,
      "step": 6081
    },
    {
      "epoch": 1.908901879242026,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 2.6196,
      "step": 6082
    },
    {
      "epoch": 1.9092157401231904,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 1.7671,
      "step": 6083
    },
    {
      "epoch": 1.9095296010043548,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.9945,
      "step": 6084
    },
    {
      "epoch": 1.9098434618855191,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.8311,
      "step": 6085
    },
    {
      "epoch": 1.9101573227666837,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 2.6576,
      "step": 6086
    },
    {
      "epoch": 1.910471183647848,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3709,
      "step": 6087
    },
    {
      "epoch": 1.9107850445290127,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3617,
      "step": 6088
    },
    {
      "epoch": 1.911098905410177,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.3939,
      "step": 6089
    },
    {
      "epoch": 1.9114127662913414,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2023,
      "step": 6090
    },
    {
      "epoch": 1.9117266271725057,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1513,
      "step": 6091
    },
    {
      "epoch": 1.91204048805367,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.3465,
      "step": 6092
    },
    {
      "epoch": 1.9123543489348345,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.3369,
      "step": 6093
    },
    {
      "epoch": 1.912668209815999,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.5491,
      "step": 6094
    },
    {
      "epoch": 1.9129820706971636,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.5203,
      "step": 6095
    },
    {
      "epoch": 1.913295931578328,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.4288,
      "step": 6096
    },
    {
      "epoch": 1.9136097924594924,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.3437,
      "step": 6097
    },
    {
      "epoch": 1.9139236533406567,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.5743,
      "step": 6098
    },
    {
      "epoch": 1.914237514221821,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.1673,
      "step": 6099
    },
    {
      "epoch": 1.9145513751029855,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.4265,
      "step": 6100
    },
    {
      "epoch": 1.91486523598415,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.2792,
      "step": 6101
    },
    {
      "epoch": 1.9151790968653144,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.3234,
      "step": 6102
    },
    {
      "epoch": 1.915492957746479,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.3086,
      "step": 6103
    },
    {
      "epoch": 1.9158068186276433,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.1944,
      "step": 6104
    },
    {
      "epoch": 1.9161206795088077,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.3692,
      "step": 6105
    },
    {
      "epoch": 1.916434540389972,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.3932,
      "step": 6106
    },
    {
      "epoch": 1.9167484012711364,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.4377,
      "step": 6107
    },
    {
      "epoch": 1.917062262152301,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.469,
      "step": 6108
    },
    {
      "epoch": 1.9173761230334654,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3331,
      "step": 6109
    },
    {
      "epoch": 1.91768998391463,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.3366,
      "step": 6110
    },
    {
      "epoch": 1.9180038447957943,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.1168,
      "step": 6111
    },
    {
      "epoch": 1.9183177056769587,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3462,
      "step": 6112
    },
    {
      "epoch": 1.918631566558123,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.4024,
      "step": 6113
    },
    {
      "epoch": 1.9189454274392874,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.5792,
      "step": 6114
    },
    {
      "epoch": 1.919259288320452,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.8491,
      "step": 6115
    },
    {
      "epoch": 1.9195731492016164,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.8856,
      "step": 6116
    },
    {
      "epoch": 1.919887010082781,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.7271,
      "step": 6117
    },
    {
      "epoch": 1.9202008709639453,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.8991,
      "step": 6118
    },
    {
      "epoch": 1.9205147318451097,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.7717,
      "step": 6119
    },
    {
      "epoch": 1.920828592726274,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.1364,
      "step": 6120
    },
    {
      "epoch": 1.9211424536074384,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.7342,
      "step": 6121
    },
    {
      "epoch": 1.9214563144886028,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.7976,
      "step": 6122
    },
    {
      "epoch": 1.9217701753697674,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.7846,
      "step": 6123
    },
    {
      "epoch": 1.922084036250932,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 2.0553,
      "step": 6124
    },
    {
      "epoch": 1.9223978971320963,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.7303,
      "step": 6125
    },
    {
      "epoch": 1.9227117580132607,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 2.1597,
      "step": 6126
    },
    {
      "epoch": 1.923025618894425,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.9155,
      "step": 6127
    },
    {
      "epoch": 1.9233394797755894,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 1.9295,
      "step": 6128
    },
    {
      "epoch": 1.9236533406567538,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 2.2244,
      "step": 6129
    },
    {
      "epoch": 1.9239672015379183,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.6728,
      "step": 6130
    },
    {
      "epoch": 1.9242810624190827,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.9492,
      "step": 6131
    },
    {
      "epoch": 1.9245949233002473,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.8253,
      "step": 6132
    },
    {
      "epoch": 1.9249087841814116,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 1.8183,
      "step": 6133
    },
    {
      "epoch": 1.925222645062576,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.7386,
      "step": 6134
    },
    {
      "epoch": 1.9255365059437404,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.9165,
      "step": 6135
    },
    {
      "epoch": 1.9258503668249047,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0002,
      "loss": 3.2698,
      "step": 6136
    },
    {
      "epoch": 1.9261642277060693,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.5735,
      "step": 6137
    },
    {
      "epoch": 1.9264780885872337,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.42,
      "step": 6138
    },
    {
      "epoch": 1.9267919494683983,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.321,
      "step": 6139
    },
    {
      "epoch": 1.9271058103495626,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.3791,
      "step": 6140
    },
    {
      "epoch": 1.927419671230727,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.4568,
      "step": 6141
    },
    {
      "epoch": 1.9277335321118914,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.3315,
      "step": 6142
    },
    {
      "epoch": 1.9280473929930557,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.347,
      "step": 6143
    },
    {
      "epoch": 1.9283612538742203,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.3535,
      "step": 6144
    },
    {
      "epoch": 1.9286751147553847,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.261,
      "step": 6145
    },
    {
      "epoch": 1.9289889756365493,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.3203,
      "step": 6146
    },
    {
      "epoch": 1.9293028365177136,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.4292,
      "step": 6147
    },
    {
      "epoch": 1.929616697398878,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.1916,
      "step": 6148
    },
    {
      "epoch": 1.9299305582800423,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3454,
      "step": 6149
    },
    {
      "epoch": 1.9302444191612067,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.26,
      "step": 6150
    },
    {
      "epoch": 1.930558280042371,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.4168,
      "step": 6151
    },
    {
      "epoch": 1.9308721409235357,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.4485,
      "step": 6152
    },
    {
      "epoch": 1.9311860018047,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.3304,
      "step": 6153
    },
    {
      "epoch": 1.9314998626858646,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.4071,
      "step": 6154
    },
    {
      "epoch": 1.931813723567029,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.4118,
      "step": 6155
    },
    {
      "epoch": 1.9321275844481933,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.4228,
      "step": 6156
    },
    {
      "epoch": 1.9324414453293577,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.2848,
      "step": 6157
    },
    {
      "epoch": 1.932755306210522,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.3849,
      "step": 6158
    },
    {
      "epoch": 1.9330691670916866,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.4587,
      "step": 6159
    },
    {
      "epoch": 1.933383027972851,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.4131,
      "step": 6160
    },
    {
      "epoch": 1.9336968888540156,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.3038,
      "step": 6161
    },
    {
      "epoch": 1.93401074973518,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.6321,
      "step": 6162
    },
    {
      "epoch": 1.9343246106163443,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.6609,
      "step": 6163
    },
    {
      "epoch": 1.9346384714975087,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.617,
      "step": 6164
    },
    {
      "epoch": 1.934952332378673,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.6255,
      "step": 6165
    },
    {
      "epoch": 1.9352661932598376,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.6832,
      "step": 6166
    },
    {
      "epoch": 1.935580054141002,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.4293,
      "step": 6167
    },
    {
      "epoch": 1.9358939150221666,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.4334,
      "step": 6168
    },
    {
      "epoch": 1.936207775903331,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.6668,
      "step": 6169
    },
    {
      "epoch": 1.9365216367844953,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.9475,
      "step": 6170
    },
    {
      "epoch": 1.9368354976656597,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.7741,
      "step": 6171
    },
    {
      "epoch": 1.9368354976656597,
      "eval_loss": 1.78682541847229,
      "eval_runtime": 152.5449,
      "eval_samples_per_second": 6.555,
      "eval_steps_per_second": 6.555,
      "step": 6171
    },
    {
      "epoch": 1.9368354976656597,
      "mmlu_eval_accuracy": 0.4197257727194544,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "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": 0.3181818181818182,
      "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": 0.2926829268292683,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.25,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
      "mmlu_eval_accuracy_high_school_geography": 0.6363636363636364,
      "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": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.65,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
      "mmlu_eval_accuracy_miscellaneous": 0.5,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.3,
      "mmlu_eval_accuracy_nutrition": 0.42424242424242425,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.213808760309749,
      "step": 6171
    },
    {
      "epoch": 1.937149358546824,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.8713,
      "step": 6172
    },
    {
      "epoch": 1.9374632194279884,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.7965,
      "step": 6173
    },
    {
      "epoch": 1.937777080309153,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.7159,
      "step": 6174
    },
    {
      "epoch": 1.9380909411903176,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.9154,
      "step": 6175
    },
    {
      "epoch": 1.938404802071482,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.9675,
      "step": 6176
    },
    {
      "epoch": 1.9387186629526463,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 2.4266,
      "step": 6177
    },
    {
      "epoch": 1.9390325238338106,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 2.2141,
      "step": 6178
    },
    {
      "epoch": 1.939346384714975,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.9954,
      "step": 6179
    },
    {
      "epoch": 1.9396602455961394,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.7545,
      "step": 6180
    },
    {
      "epoch": 1.939974106477304,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 2.0048,
      "step": 6181
    },
    {
      "epoch": 1.9402879673584683,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.7197,
      "step": 6182
    },
    {
      "epoch": 1.940601828239633,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 1.9448,
      "step": 6183
    },
    {
      "epoch": 1.9409156891207973,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.8629,
      "step": 6184
    },
    {
      "epoch": 1.9412295500019616,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 1.8311,
      "step": 6185
    },
    {
      "epoch": 1.941543410883126,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 2.6935,
      "step": 6186
    },
    {
      "epoch": 1.9418572717642903,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.483,
      "step": 6187
    },
    {
      "epoch": 1.942171132645455,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2082,
      "step": 6188
    },
    {
      "epoch": 1.9424849935266193,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3822,
      "step": 6189
    },
    {
      "epoch": 1.9427988544077839,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.3582,
      "step": 6190
    },
    {
      "epoch": 1.9431127152889482,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.295,
      "step": 6191
    },
    {
      "epoch": 1.9434265761701126,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2375,
      "step": 6192
    },
    {
      "epoch": 1.943740437051277,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.2288,
      "step": 6193
    },
    {
      "epoch": 1.9440542979324413,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.3575,
      "step": 6194
    },
    {
      "epoch": 1.944368158813606,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.5366,
      "step": 6195
    },
    {
      "epoch": 1.9446820196947703,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.2837,
      "step": 6196
    },
    {
      "epoch": 1.9449958805759349,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.3292,
      "step": 6197
    },
    {
      "epoch": 1.9453097414570992,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.3007,
      "step": 6198
    },
    {
      "epoch": 1.9456236023382636,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.3888,
      "step": 6199
    },
    {
      "epoch": 1.945937463219428,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.3016,
      "step": 6200
    },
    {
      "epoch": 1.9462513241005923,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.254,
      "step": 6201
    },
    {
      "epoch": 1.9465651849817567,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.414,
      "step": 6202
    },
    {
      "epoch": 1.9468790458629213,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.2443,
      "step": 6203
    },
    {
      "epoch": 1.9471929067440858,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.3593,
      "step": 6204
    },
    {
      "epoch": 1.9475067676252502,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3608,
      "step": 6205
    },
    {
      "epoch": 1.9478206285064146,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.4143,
      "step": 6206
    },
    {
      "epoch": 1.948134489387579,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.1969,
      "step": 6207
    },
    {
      "epoch": 1.9484483502687433,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3597,
      "step": 6208
    },
    {
      "epoch": 1.9487622111499077,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.3017,
      "step": 6209
    },
    {
      "epoch": 1.9490760720310722,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.3243,
      "step": 6210
    },
    {
      "epoch": 1.9493899329122366,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.4498,
      "step": 6211
    },
    {
      "epoch": 1.9497037937934012,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.5024,
      "step": 6212
    },
    {
      "epoch": 1.9500176546745656,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.7039,
      "step": 6213
    },
    {
      "epoch": 1.95033151555573,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.73,
      "step": 6214
    },
    {
      "epoch": 1.9506453764368943,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.7362,
      "step": 6215
    },
    {
      "epoch": 1.9509592373180586,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.7638,
      "step": 6216
    },
    {
      "epoch": 1.9512730981992232,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.896,
      "step": 6217
    },
    {
      "epoch": 1.9515869590803876,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.7783,
      "step": 6218
    },
    {
      "epoch": 1.9519008199615522,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.5481,
      "step": 6219
    },
    {
      "epoch": 1.9522146808427165,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.6818,
      "step": 6220
    },
    {
      "epoch": 1.952528541723881,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 1.8717,
      "step": 6221
    },
    {
      "epoch": 1.9528424026050453,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.8411,
      "step": 6222
    },
    {
      "epoch": 1.9531562634862096,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 2.1213,
      "step": 6223
    },
    {
      "epoch": 1.9534701243673742,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.0931,
      "step": 6224
    },
    {
      "epoch": 1.9537839852485386,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.8087,
      "step": 6225
    },
    {
      "epoch": 1.9540978461297032,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.9759,
      "step": 6226
    },
    {
      "epoch": 1.9544117070108675,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 2.0105,
      "step": 6227
    },
    {
      "epoch": 1.954725567892032,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 2.4152,
      "step": 6228
    },
    {
      "epoch": 1.9550394287731963,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 2.1308,
      "step": 6229
    },
    {
      "epoch": 1.9553532896543606,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.7342,
      "step": 6230
    },
    {
      "epoch": 1.955667150535525,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.3799,
      "step": 6231
    },
    {
      "epoch": 1.9559810114166896,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 2.0526,
      "step": 6232
    },
    {
      "epoch": 1.956294872297854,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.8936,
      "step": 6233
    },
    {
      "epoch": 1.9566087331790185,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.8216,
      "step": 6234
    },
    {
      "epoch": 1.9569225940601829,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002,
      "loss": 2.5648,
      "step": 6235
    },
    {
      "epoch": 1.9572364549413472,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 2.8273,
      "step": 6236
    },
    {
      "epoch": 1.9575503158225116,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.4938,
      "step": 6237
    },
    {
      "epoch": 1.957864176703676,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4521,
      "step": 6238
    },
    {
      "epoch": 1.9581780375848405,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2935,
      "step": 6239
    },
    {
      "epoch": 1.958491898466005,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.2917,
      "step": 6240
    },
    {
      "epoch": 1.9588057593471695,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.3696,
      "step": 6241
    },
    {
      "epoch": 1.9591196202283339,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.4037,
      "step": 6242
    },
    {
      "epoch": 1.9594334811094982,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.3007,
      "step": 6243
    },
    {
      "epoch": 1.9597473419906626,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.3639,
      "step": 6244
    },
    {
      "epoch": 1.960061202871827,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.3895,
      "step": 6245
    },
    {
      "epoch": 1.9603750637529915,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.2628,
      "step": 6246
    },
    {
      "epoch": 1.960688924634156,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.3607,
      "step": 6247
    },
    {
      "epoch": 1.9610027855153205,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.4322,
      "step": 6248
    },
    {
      "epoch": 1.9613166463964848,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.2913,
      "step": 6249
    },
    {
      "epoch": 1.9616305072776492,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.241,
      "step": 6250
    },
    {
      "epoch": 1.9619443681588136,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.3618,
      "step": 6251
    },
    {
      "epoch": 1.962258229039978,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.2447,
      "step": 6252
    },
    {
      "epoch": 1.9625720899211423,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.2828,
      "step": 6253
    },
    {
      "epoch": 1.9628859508023069,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.2431,
      "step": 6254
    },
    {
      "epoch": 1.9631998116834715,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.298,
      "step": 6255
    },
    {
      "epoch": 1.9635136725646358,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.2924,
      "step": 6256
    },
    {
      "epoch": 1.9638275334458002,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.6099,
      "step": 6257
    },
    {
      "epoch": 1.9641413943269646,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.3266,
      "step": 6258
    },
    {
      "epoch": 1.964455255208129,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.3105,
      "step": 6259
    },
    {
      "epoch": 1.9647691160892933,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.4488,
      "step": 6260
    },
    {
      "epoch": 1.9650829769704579,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.7142,
      "step": 6261
    },
    {
      "epoch": 1.9653968378516222,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.5408,
      "step": 6262
    },
    {
      "epoch": 1.9657106987327868,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.6264,
      "step": 6263
    },
    {
      "epoch": 1.9660245596139512,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.5745,
      "step": 6264
    },
    {
      "epoch": 1.9663384204951155,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 2.0871,
      "step": 6265
    },
    {
      "epoch": 1.96665228137628,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.5642,
      "step": 6266
    },
    {
      "epoch": 1.9669661422574443,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.6649,
      "step": 6267
    },
    {
      "epoch": 1.9672800031386088,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.9547,
      "step": 6268
    },
    {
      "epoch": 1.9675938640197732,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.791,
      "step": 6269
    },
    {
      "epoch": 1.9679077249009378,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 2.2883,
      "step": 6270
    },
    {
      "epoch": 1.9682215857821022,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 2.1999,
      "step": 6271
    },
    {
      "epoch": 1.9685354466632665,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 2.0373,
      "step": 6272
    },
    {
      "epoch": 1.9688493075444309,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.9539,
      "step": 6273
    },
    {
      "epoch": 1.9691631684255952,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 2.3226,
      "step": 6274
    },
    {
      "epoch": 1.9694770293067598,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.5914,
      "step": 6275
    },
    {
      "epoch": 1.9697908901879242,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.8675,
      "step": 6276
    },
    {
      "epoch": 1.9701047510690888,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.9848,
      "step": 6277
    },
    {
      "epoch": 1.9704186119502531,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.6943,
      "step": 6278
    },
    {
      "epoch": 1.9707324728314175,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.2691,
      "step": 6279
    },
    {
      "epoch": 1.9710463337125819,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.1909,
      "step": 6280
    },
    {
      "epoch": 1.9713601945937462,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.7719,
      "step": 6281
    },
    {
      "epoch": 1.9716740554749106,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.9302,
      "step": 6282
    },
    {
      "epoch": 1.9719879163560752,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 2.1265,
      "step": 6283
    },
    {
      "epoch": 1.9723017772372395,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.4069,
      "step": 6284
    },
    {
      "epoch": 1.9726156381184041,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 2.089,
      "step": 6285
    },
    {
      "epoch": 1.9729294989995685,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0002,
      "loss": 2.8715,
      "step": 6286
    },
    {
      "epoch": 1.9732433598807329,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.5183,
      "step": 6287
    },
    {
      "epoch": 1.9735572207618972,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.378,
      "step": 6288
    },
    {
      "epoch": 1.9738710816430616,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.4105,
      "step": 6289
    },
    {
      "epoch": 1.9741849425242262,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.4007,
      "step": 6290
    },
    {
      "epoch": 1.9744988034053905,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.4438,
      "step": 6291
    },
    {
      "epoch": 1.974812664286555,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.3228,
      "step": 6292
    },
    {
      "epoch": 1.9751265251677195,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.4804,
      "step": 6293
    },
    {
      "epoch": 1.9754403860488838,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.4462,
      "step": 6294
    },
    {
      "epoch": 1.9757542469300482,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.2349,
      "step": 6295
    },
    {
      "epoch": 1.9760681078112126,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.2008,
      "step": 6296
    },
    {
      "epoch": 1.9763819686923771,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.3636,
      "step": 6297
    },
    {
      "epoch": 1.9766958295735415,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.3708,
      "step": 6298
    },
    {
      "epoch": 1.977009690454706,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.4474,
      "step": 6299
    },
    {
      "epoch": 1.9773235513358705,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3537,
      "step": 6300
    },
    {
      "epoch": 1.9776374122170348,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.3297,
      "step": 6301
    },
    {
      "epoch": 1.9779512730981992,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.5156,
      "step": 6302
    },
    {
      "epoch": 1.9782651339793635,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.1508,
      "step": 6303
    },
    {
      "epoch": 1.978578994860528,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.2305,
      "step": 6304
    },
    {
      "epoch": 1.9788928557416925,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.2829,
      "step": 6305
    },
    {
      "epoch": 1.979206716622857,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.318,
      "step": 6306
    },
    {
      "epoch": 1.9795205775040214,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.3873,
      "step": 6307
    },
    {
      "epoch": 1.9798344383851858,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.2953,
      "step": 6308
    },
    {
      "epoch": 1.9801482992663502,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.4672,
      "step": 6309
    },
    {
      "epoch": 1.9804621601475145,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.4184,
      "step": 6310
    },
    {
      "epoch": 1.980776021028679,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.4374,
      "step": 6311
    },
    {
      "epoch": 1.9810898819098435,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.4165,
      "step": 6312
    },
    {
      "epoch": 1.9814037427910078,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.7993,
      "step": 6313
    },
    {
      "epoch": 1.9817176036721724,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.467,
      "step": 6314
    },
    {
      "epoch": 1.9820314645533368,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.8822,
      "step": 6315
    },
    {
      "epoch": 1.9823453254345011,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.7527,
      "step": 6316
    },
    {
      "epoch": 1.9826591863156655,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.7072,
      "step": 6317
    },
    {
      "epoch": 1.9829730471968299,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.754,
      "step": 6318
    },
    {
      "epoch": 1.9832869080779945,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.5057,
      "step": 6319
    },
    {
      "epoch": 1.9836007689591588,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 2.2067,
      "step": 6320
    },
    {
      "epoch": 1.9839146298403234,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.9669,
      "step": 6321
    },
    {
      "epoch": 1.9842284907214878,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.9307,
      "step": 6322
    },
    {
      "epoch": 1.9845423516026521,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 2.0883,
      "step": 6323
    },
    {
      "epoch": 1.9848562124838165,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 2.1274,
      "step": 6324
    },
    {
      "epoch": 1.9851700733649809,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.9849,
      "step": 6325
    },
    {
      "epoch": 1.9854839342461454,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.7451,
      "step": 6326
    },
    {
      "epoch": 1.9857977951273098,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.8038,
      "step": 6327
    },
    {
      "epoch": 1.9861116560084744,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 2.4209,
      "step": 6328
    },
    {
      "epoch": 1.9864255168896388,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.948,
      "step": 6329
    },
    {
      "epoch": 1.9867393777708031,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 2.096,
      "step": 6330
    },
    {
      "epoch": 1.9870532386519675,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.7269,
      "step": 6331
    },
    {
      "epoch": 1.9873670995331318,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.4452,
      "step": 6332
    },
    {
      "epoch": 1.9876809604142962,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.6012,
      "step": 6333
    },
    {
      "epoch": 1.9879948212954608,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 2.0283,
      "step": 6334
    },
    {
      "epoch": 1.9883086821766254,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 2.1029,
      "step": 6335
    },
    {
      "epoch": 1.9886225430577897,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 2.6477,
      "step": 6336
    },
    {
      "epoch": 1.988936403938954,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.549,
      "step": 6337
    },
    {
      "epoch": 1.9892502648201185,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.464,
      "step": 6338
    },
    {
      "epoch": 1.9895641257012828,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4544,
      "step": 6339
    },
    {
      "epoch": 1.9898779865824472,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.4293,
      "step": 6340
    },
    {
      "epoch": 1.9901918474636118,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.3721,
      "step": 6341
    },
    {
      "epoch": 1.9905057083447761,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.4958,
      "step": 6342
    },
    {
      "epoch": 1.9908195692259407,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.2461,
      "step": 6343
    },
    {
      "epoch": 1.991133430107105,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.3111,
      "step": 6344
    },
    {
      "epoch": 1.9914472909882694,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.3667,
      "step": 6345
    },
    {
      "epoch": 1.9917611518694338,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.3126,
      "step": 6346
    },
    {
      "epoch": 1.9920750127505982,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.4414,
      "step": 6347
    },
    {
      "epoch": 1.9923888736317628,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.4253,
      "step": 6348
    },
    {
      "epoch": 1.9927027345129271,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.2938,
      "step": 6349
    },
    {
      "epoch": 1.9930165953940917,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 1.5293,
      "step": 6350
    },
    {
      "epoch": 1.993330456275256,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.4518,
      "step": 6351
    },
    {
      "epoch": 1.9936443171564204,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.3971,
      "step": 6352
    },
    {
      "epoch": 1.9939581780375848,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.5387,
      "step": 6353
    },
    {
      "epoch": 1.9942720389187492,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.8592,
      "step": 6354
    },
    {
      "epoch": 1.9945858997999137,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.5055,
      "step": 6355
    },
    {
      "epoch": 1.994899760681078,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.6136,
      "step": 6356
    },
    {
      "epoch": 1.9952136215622427,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.7147,
      "step": 6357
    },
    {
      "epoch": 1.995527482443407,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.4932,
      "step": 6358
    },
    {
      "epoch": 1.995527482443407,
      "eval_loss": 1.7805804014205933,
      "eval_runtime": 155.909,
      "eval_samples_per_second": 6.414,
      "eval_steps_per_second": 6.414,
      "step": 6358
    },
    {
      "epoch": 1.995527482443407,
      "mmlu_eval_accuracy": 0.409254918544326,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "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": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.55,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.3076923076923077,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.25,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.2727272727272727,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5465116279069767,
      "mmlu_eval_accuracy_moral_disputes": 0.3684210526315789,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.42424242424242425,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.3345751797039473,
      "step": 6358
    },
    {
      "epoch": 1.9958413433245714,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.6581,
      "step": 6359
    },
    {
      "epoch": 1.9961552042057358,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.9552,
      "step": 6360
    },
    {
      "epoch": 1.9964690650869001,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.8435,
      "step": 6361
    },
    {
      "epoch": 1.9967829259680645,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 2.5541,
      "step": 6362
    },
    {
      "epoch": 1.997096786849229,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 2.1237,
      "step": 6363
    },
    {
      "epoch": 1.9974106477303935,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 2.0774,
      "step": 6364
    },
    {
      "epoch": 1.997724508611558,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 2.0212,
      "step": 6365
    },
    {
      "epoch": 1.9980383694927224,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 2.1173,
      "step": 6366
    },
    {
      "epoch": 1.9983522303738868,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.6951,
      "step": 6367
    },
    {
      "epoch": 1.9986660912550511,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 2.1471,
      "step": 6368
    },
    {
      "epoch": 1.9989799521362155,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 2.0464,
      "step": 6369
    },
    {
      "epoch": 1.99929381301738,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.8361,
      "step": 6370
    },
    {
      "epoch": 1.9996076738985444,
      "grad_norm": 1.0,
      "learning_rate": 0.0002,
      "loss": 2.2817,
      "step": 6371
    },
    {
      "epoch": 1.999921534779709,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 2.1017,
      "step": 6372
    },
    {
      "epoch": 2.0002353956608734,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.5499,
      "step": 6373
    },
    {
      "epoch": 2.0005492565420377,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.3603,
      "step": 6374
    },
    {
      "epoch": 2.000863117423202,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1976,
      "step": 6375
    },
    {
      "epoch": 2.0011769783043665,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2112,
      "step": 6376
    },
    {
      "epoch": 2.001490839185531,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2047,
      "step": 6377
    },
    {
      "epoch": 2.0018047000666956,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.3085,
      "step": 6378
    },
    {
      "epoch": 2.00211856094786,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.2039,
      "step": 6379
    },
    {
      "epoch": 2.0024324218290244,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.2814,
      "step": 6380
    },
    {
      "epoch": 2.0027462827101887,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.3825,
      "step": 6381
    },
    {
      "epoch": 2.003060143591353,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.281,
      "step": 6382
    },
    {
      "epoch": 2.0033740044725175,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.2624,
      "step": 6383
    },
    {
      "epoch": 2.003687865353682,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.1571,
      "step": 6384
    },
    {
      "epoch": 2.0040017262348466,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.2529,
      "step": 6385
    },
    {
      "epoch": 2.004315587116011,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.1393,
      "step": 6386
    },
    {
      "epoch": 2.0046294479971754,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.1735,
      "step": 6387
    },
    {
      "epoch": 2.0049433088783397,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.0209,
      "step": 6388
    },
    {
      "epoch": 2.005257169759504,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.2537,
      "step": 6389
    },
    {
      "epoch": 2.0055710306406684,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.2957,
      "step": 6390
    },
    {
      "epoch": 2.005884891521833,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.1899,
      "step": 6391
    },
    {
      "epoch": 2.006198752402997,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.1961,
      "step": 6392
    },
    {
      "epoch": 2.006512613284162,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.2249,
      "step": 6393
    },
    {
      "epoch": 2.0068264741653263,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.1087,
      "step": 6394
    },
    {
      "epoch": 2.0071403350464907,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.3485,
      "step": 6395
    },
    {
      "epoch": 2.007454195927655,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.1663,
      "step": 6396
    },
    {
      "epoch": 2.0077680568088194,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.168,
      "step": 6397
    },
    {
      "epoch": 2.008081917689984,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.1941,
      "step": 6398
    },
    {
      "epoch": 2.008395778571148,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.281,
      "step": 6399
    },
    {
      "epoch": 2.008709639452313,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.2982,
      "step": 6400
    },
    {
      "epoch": 2.0090235003334773,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.223,
      "step": 6401
    },
    {
      "epoch": 2.0093373612146417,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.5336,
      "step": 6402
    },
    {
      "epoch": 2.009651222095806,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.5812,
      "step": 6403
    },
    {
      "epoch": 2.0099650829769704,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.1882,
      "step": 6404
    },
    {
      "epoch": 2.0102789438581348,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.5586,
      "step": 6405
    },
    {
      "epoch": 2.010592804739299,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.7124,
      "step": 6406
    },
    {
      "epoch": 2.010906665620464,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.5269,
      "step": 6407
    },
    {
      "epoch": 2.0112205265016283,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.8241,
      "step": 6408
    },
    {
      "epoch": 2.0115343873827927,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.5197,
      "step": 6409
    },
    {
      "epoch": 2.011848248263957,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.5815,
      "step": 6410
    },
    {
      "epoch": 2.0121621091451214,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.6764,
      "step": 6411
    },
    {
      "epoch": 2.0124759700262858,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.8467,
      "step": 6412
    },
    {
      "epoch": 2.01278983090745,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.572,
      "step": 6413
    },
    {
      "epoch": 2.0131036917886145,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.4858,
      "step": 6414
    },
    {
      "epoch": 2.0134175526697793,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.422,
      "step": 6415
    },
    {
      "epoch": 2.0137314135509436,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.4264,
      "step": 6416
    },
    {
      "epoch": 2.014045274432108,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.5191,
      "step": 6417
    },
    {
      "epoch": 2.0143591353132724,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.4893,
      "step": 6418
    },
    {
      "epoch": 2.0146729961944367,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 1.6836,
      "step": 6419
    },
    {
      "epoch": 2.014986857075601,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 1.4381,
      "step": 6420
    },
    {
      "epoch": 2.0153007179567655,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 1.3994,
      "step": 6421
    },
    {
      "epoch": 2.0156145788379303,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002,
      "loss": 1.7388,
      "step": 6422
    },
    {
      "epoch": 2.0159284397190946,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.5368,
      "step": 6423
    },
    {
      "epoch": 2.016242300600259,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.3737,
      "step": 6424
    },
    {
      "epoch": 2.0165561614814234,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.3951,
      "step": 6425
    },
    {
      "epoch": 2.0168700223625877,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.273,
      "step": 6426
    },
    {
      "epoch": 2.017183883243752,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2884,
      "step": 6427
    },
    {
      "epoch": 2.0174977441249164,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.2472,
      "step": 6428
    },
    {
      "epoch": 2.0178116050060813,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.3831,
      "step": 6429
    },
    {
      "epoch": 2.0181254658872456,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.2233,
      "step": 6430
    },
    {
      "epoch": 2.01843932676841,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.3126,
      "step": 6431
    },
    {
      "epoch": 2.0187531876495743,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.1973,
      "step": 6432
    },
    {
      "epoch": 2.0190670485307387,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.2813,
      "step": 6433
    },
    {
      "epoch": 2.019380909411903,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.1946,
      "step": 6434
    },
    {
      "epoch": 2.0196947702930674,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.269,
      "step": 6435
    },
    {
      "epoch": 2.0200086311742322,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.2773,
      "step": 6436
    },
    {
      "epoch": 2.0203224920553966,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.1583,
      "step": 6437
    },
    {
      "epoch": 2.020636352936561,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.097,
      "step": 6438
    },
    {
      "epoch": 2.0209502138177253,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.1401,
      "step": 6439
    },
    {
      "epoch": 2.0212640746988897,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.1959,
      "step": 6440
    },
    {
      "epoch": 2.021577935580054,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.1521,
      "step": 6441
    },
    {
      "epoch": 2.0218917964612184,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.1557,
      "step": 6442
    },
    {
      "epoch": 2.022205657342383,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.1005,
      "step": 6443
    },
    {
      "epoch": 2.0225195182235476,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.1871,
      "step": 6444
    },
    {
      "epoch": 2.022833379104712,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.0931,
      "step": 6445
    },
    {
      "epoch": 2.0231472399858763,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.1432,
      "step": 6446
    },
    {
      "epoch": 2.0234611008670407,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.1383,
      "step": 6447
    },
    {
      "epoch": 2.023774961748205,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.0328,
      "step": 6448
    },
    {
      "epoch": 2.0240888226293694,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.1496,
      "step": 6449
    },
    {
      "epoch": 2.0244026835105338,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.1561,
      "step": 6450
    },
    {
      "epoch": 2.0247165443916986,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.3405,
      "step": 6451
    },
    {
      "epoch": 2.025030405272863,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.1838,
      "step": 6452
    },
    {
      "epoch": 2.0253442661540273,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.5417,
      "step": 6453
    },
    {
      "epoch": 2.0256581270351917,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.3732,
      "step": 6454
    },
    {
      "epoch": 2.025971987916356,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 1.2126,
      "step": 6455
    },
    {
      "epoch": 2.0262858487975204,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.2923,
      "step": 6456
    },
    {
      "epoch": 2.0265997096786847,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.7343,
      "step": 6457
    },
    {
      "epoch": 2.0269135705598496,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.4832,
      "step": 6458
    },
    {
      "epoch": 2.027227431441014,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.3652,
      "step": 6459
    },
    {
      "epoch": 2.0275412923221783,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.3195,
      "step": 6460
    },
    {
      "epoch": 2.0278551532033426,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 1.5438,
      "step": 6461
    },
    {
      "epoch": 2.028169014084507,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.885,
      "step": 6462
    },
    {
      "epoch": 2.0284828749656714,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.3795,
      "step": 6463
    },
    {
      "epoch": 2.0287967358468357,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.4408,
      "step": 6464
    },
    {
      "epoch": 2.0291105967280005,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.591,
      "step": 6465
    },
    {
      "epoch": 2.029424457609165,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 2.3045,
      "step": 6466
    },
    {
      "epoch": 2.0297383184903293,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.4852,
      "step": 6467
    },
    {
      "epoch": 2.0300521793714936,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 1.5878,
      "step": 6468
    },
    {
      "epoch": 2.030366040252658,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002,
      "loss": 1.3267,
      "step": 6469
    },
    {
      "epoch": 2.0306799011338224,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.4204,
      "step": 6470
    },
    {
      "epoch": 2.0309937620149867,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.4928,
      "step": 6471
    },
    {
      "epoch": 2.031307622896151,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 2.0492,
      "step": 6472
    },
    {
      "epoch": 2.031621483777316,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.4144,
      "step": 6473
    },
    {
      "epoch": 2.0319353446584802,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.1082,
      "step": 6474
    },
    {
      "epoch": 2.0322492055396446,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.2631,
      "step": 6475
    },
    {
      "epoch": 2.032563066420809,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.247,
      "step": 6476
    },
    {
      "epoch": 2.0328769273019733,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.3581,
      "step": 6477
    },
    {
      "epoch": 2.0331907881831377,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.215,
      "step": 6478
    },
    {
      "epoch": 2.033504649064302,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.261,
      "step": 6479
    },
    {
      "epoch": 2.033818509945467,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.211,
      "step": 6480
    },
    {
      "epoch": 2.0341323708266312,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.1635,
      "step": 6481
    },
    {
      "epoch": 2.0344462317077956,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.2252,
      "step": 6482
    },
    {
      "epoch": 2.03476009258896,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.1542,
      "step": 6483
    },
    {
      "epoch": 2.0350739534701243,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.1,
      "step": 6484
    },
    {
      "epoch": 2.0353878143512887,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.224,
      "step": 6485
    },
    {
      "epoch": 2.035701675232453,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.205,
      "step": 6486
    },
    {
      "epoch": 2.036015536113618,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.2151,
      "step": 6487
    },
    {
      "epoch": 2.036329396994782,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.247,
      "step": 6488
    },
    {
      "epoch": 2.0366432578759466,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.3503,
      "step": 6489
    },
    {
      "epoch": 2.036957118757111,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.2557,
      "step": 6490
    },
    {
      "epoch": 2.0372709796382753,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.0906,
      "step": 6491
    },
    {
      "epoch": 2.0375848405194397,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.1621,
      "step": 6492
    },
    {
      "epoch": 2.037898701400604,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.1906,
      "step": 6493
    },
    {
      "epoch": 2.0382125622817684,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.1041,
      "step": 6494
    },
    {
      "epoch": 2.038526423162933,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.0958,
      "step": 6495
    },
    {
      "epoch": 2.0388402840440976,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 0.9641,
      "step": 6496
    },
    {
      "epoch": 2.039154144925262,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.2378,
      "step": 6497
    },
    {
      "epoch": 2.0394680058064263,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 0.9743,
      "step": 6498
    },
    {
      "epoch": 2.0397818666875907,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.1066,
      "step": 6499
    },
    {
      "epoch": 2.040095727568755,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.1521,
      "step": 6500
    },
    {
      "epoch": 2.0404095884499194,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.2821,
      "step": 6501
    },
    {
      "epoch": 2.040723449331084,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.0824,
      "step": 6502
    },
    {
      "epoch": 2.0410373102122485,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.3349,
      "step": 6503
    },
    {
      "epoch": 2.041351171093413,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.1959,
      "step": 6504
    },
    {
      "epoch": 2.0416650319745773,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.5071,
      "step": 6505
    },
    {
      "epoch": 2.0419788928557416,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0002,
      "loss": 1.4881,
      "step": 6506
    },
    {
      "epoch": 2.042292753736906,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.5681,
      "step": 6507
    },
    {
      "epoch": 2.0426066146180704,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.2219,
      "step": 6508
    },
    {
      "epoch": 2.042920475499235,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 1.6498,
      "step": 6509
    },
    {
      "epoch": 2.0432343363803995,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.3468,
      "step": 6510
    },
    {
      "epoch": 2.043548197261564,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.6375,
      "step": 6511
    },
    {
      "epoch": 2.0438620581427283,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.8956,
      "step": 6512
    },
    {
      "epoch": 2.0441759190238926,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 1.5777,
      "step": 6513
    },
    {
      "epoch": 2.044489779905057,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002,
      "loss": 1.9436,
      "step": 6514
    },
    {
      "epoch": 2.0448036407862213,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.5448,
      "step": 6515
    },
    {
      "epoch": 2.045117501667386,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.9716,
      "step": 6516
    },
    {
      "epoch": 2.0454313625485505,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.3166,
      "step": 6517
    },
    {
      "epoch": 2.045745223429715,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.3701,
      "step": 6518
    },
    {
      "epoch": 2.0460590843108792,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.8284,
      "step": 6519
    },
    {
      "epoch": 2.0463729451920436,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.585,
      "step": 6520
    },
    {
      "epoch": 2.046686806073208,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.7104,
      "step": 6521
    },
    {
      "epoch": 2.0470006669543723,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 1.7063,
      "step": 6522
    },
    {
      "epoch": 2.0473145278355367,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.4015,
      "step": 6523
    },
    {
      "epoch": 2.0476283887167015,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.2702,
      "step": 6524
    },
    {
      "epoch": 2.047942249597866,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.3201,
      "step": 6525
    },
    {
      "epoch": 2.0482561104790302,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.3815,
      "step": 6526
    },
    {
      "epoch": 2.0485699713601946,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.2473,
      "step": 6527
    },
    {
      "epoch": 2.048883832241359,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.2444,
      "step": 6528
    },
    {
      "epoch": 2.0491976931225233,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.1019,
      "step": 6529
    },
    {
      "epoch": 2.0495115540036877,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.4394,
      "step": 6530
    },
    {
      "epoch": 2.0498254148848525,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.2125,
      "step": 6531
    },
    {
      "epoch": 2.050139275766017,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.2068,
      "step": 6532
    },
    {
      "epoch": 2.050453136647181,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.2661,
      "step": 6533
    },
    {
      "epoch": 2.0507669975283456,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.2278,
      "step": 6534
    },
    {
      "epoch": 2.05108085840951,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.1051,
      "step": 6535
    },
    {
      "epoch": 2.0513947192906743,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.2101,
      "step": 6536
    },
    {
      "epoch": 2.0517085801718387,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.1954,
      "step": 6537
    },
    {
      "epoch": 2.0520224410530035,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.1359,
      "step": 6538
    },
    {
      "epoch": 2.052336301934168,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.3137,
      "step": 6539
    },
    {
      "epoch": 2.052650162815332,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.3257,
      "step": 6540
    },
    {
      "epoch": 2.0529640236964966,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.135,
      "step": 6541
    },
    {
      "epoch": 2.053277884577661,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.3097,
      "step": 6542
    },
    {
      "epoch": 2.0535917454588253,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.1883,
      "step": 6543
    },
    {
      "epoch": 2.0539056063399896,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.0568,
      "step": 6544
    },
    {
      "epoch": 2.054219467221154,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.1958,
      "step": 6545
    },
    {
      "epoch": 2.054219467221154,
      "eval_loss": 1.843532919883728,
      "eval_runtime": 123.3514,
      "eval_samples_per_second": 8.107,
      "eval_steps_per_second": 8.107,
      "step": 6545
    },
    {
      "epoch": 2.054219467221154,
      "mmlu_eval_accuracy": 0.4295219152661749,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.7272727272727273,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "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": 0.2926829268292683,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.6,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5232558139534884,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.28,
      "mmlu_eval_accuracy_nutrition": 0.36363636363636365,
      "mmlu_eval_accuracy_philosophy": 0.4117647058823529,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.3141378964146004,
      "step": 6545
    },
    {
      "epoch": 2.054533328102319,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.4153,
      "step": 6546
    },
    {
      "epoch": 2.054847188983483,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.1459,
      "step": 6547
    },
    {
      "epoch": 2.0551610498646475,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 1.0317,
      "step": 6548
    },
    {
      "epoch": 2.055474910745812,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.3588,
      "step": 6549
    },
    {
      "epoch": 2.0557887716269763,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.1621,
      "step": 6550
    },
    {
      "epoch": 2.0561026325081406,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.1384,
      "step": 6551
    },
    {
      "epoch": 2.056416493389305,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.1263,
      "step": 6552
    },
    {
      "epoch": 2.05673035427047,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.4628,
      "step": 6553
    },
    {
      "epoch": 2.057044215151634,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.399,
      "step": 6554
    },
    {
      "epoch": 2.0573580760327985,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.04,
      "step": 6555
    },
    {
      "epoch": 2.057671936913963,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.4969,
      "step": 6556
    },
    {
      "epoch": 2.0579857977951272,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.2983,
      "step": 6557
    },
    {
      "epoch": 2.0582996586762916,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.3996,
      "step": 6558
    },
    {
      "epoch": 2.058613519557456,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002,
      "loss": 1.6716,
      "step": 6559
    },
    {
      "epoch": 2.058927380438621,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.2771,
      "step": 6560
    },
    {
      "epoch": 2.059241241319785,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.5109,
      "step": 6561
    },
    {
      "epoch": 2.0595551022009495,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.5093,
      "step": 6562
    },
    {
      "epoch": 2.059868963082114,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.3985,
      "step": 6563
    },
    {
      "epoch": 2.0601828239632782,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.4276,
      "step": 6564
    },
    {
      "epoch": 2.0604966848444426,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.6134,
      "step": 6565
    },
    {
      "epoch": 2.060810545725607,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 1.8538,
      "step": 6566
    },
    {
      "epoch": 2.0611244066067718,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002,
      "loss": 1.8285,
      "step": 6567
    },
    {
      "epoch": 2.061438267487936,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.8899,
      "step": 6568
    },
    {
      "epoch": 2.0617521283691005,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 1.6118,
      "step": 6569
    },
    {
      "epoch": 2.062065989250265,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002,
      "loss": 1.4798,
      "step": 6570
    },
    {
      "epoch": 2.062379850131429,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 1.3975,
      "step": 6571
    },
    {
      "epoch": 2.0626937110125936,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 1.5737,
      "step": 6572
    },
    {
      "epoch": 2.063007571893758,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.482,
      "step": 6573
    },
    {
      "epoch": 2.0633214327749223,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.2845,
      "step": 6574
    },
    {
      "epoch": 2.063635293656087,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.296,
      "step": 6575
    },
    {
      "epoch": 2.0639491545372515,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.2488,
      "step": 6576
    },
    {
      "epoch": 2.064263015418416,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.2408,
      "step": 6577
    },
    {
      "epoch": 2.06457687629958,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.1978,
      "step": 6578
    },
    {
      "epoch": 2.0648907371807446,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.3885,
      "step": 6579
    },
    {
      "epoch": 2.065204598061909,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.2603,
      "step": 6580
    },
    {
      "epoch": 2.0655184589430733,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.2214,
      "step": 6581
    },
    {
      "epoch": 2.065832319824238,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.1807,
      "step": 6582
    },
    {
      "epoch": 2.0661461807054025,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.1363,
      "step": 6583
    },
    {
      "epoch": 2.066460041586567,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.3298,
      "step": 6584
    },
    {
      "epoch": 2.066773902467731,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.1943,
      "step": 6585
    },
    {
      "epoch": 2.0670877633488955,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.1347,
      "step": 6586
    },
    {
      "epoch": 2.06740162423006,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.1102,
      "step": 6587
    },
    {
      "epoch": 2.0677154851112243,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.0893,
      "step": 6588
    },
    {
      "epoch": 2.068029345992389,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.1005,
      "step": 6589
    },
    {
      "epoch": 2.0683432068735534,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.2608,
      "step": 6590
    },
    {
      "epoch": 2.068657067754718,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.1202,
      "step": 6591
    },
    {
      "epoch": 2.068970928635882,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.2543,
      "step": 6592
    },
    {
      "epoch": 2.0692847895170465,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 0.998,
      "step": 6593
    },
    {
      "epoch": 2.069598650398211,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.0883,
      "step": 6594
    },
    {
      "epoch": 2.0699125112793753,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.1443,
      "step": 6595
    },
    {
      "epoch": 2.0702263721605396,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.0458,
      "step": 6596
    },
    {
      "epoch": 2.0705402330417044,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.285,
      "step": 6597
    },
    {
      "epoch": 2.070854093922869,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.2554,
      "step": 6598
    },
    {
      "epoch": 2.071167954804033,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.3008,
      "step": 6599
    },
    {
      "epoch": 2.0714818156851975,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.1847,
      "step": 6600
    },
    {
      "epoch": 2.071795676566362,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.0627,
      "step": 6601
    },
    {
      "epoch": 2.0721095374475262,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.1344,
      "step": 6602
    },
    {
      "epoch": 2.0724233983286906,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 1.2701,
      "step": 6603
    },
    {
      "epoch": 2.0727372592098554,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.4701,
      "step": 6604
    },
    {
      "epoch": 2.0730511200910198,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.1509,
      "step": 6605
    },
    {
      "epoch": 2.073364980972184,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002,
      "loss": 1.4165,
      "step": 6606
    },
    {
      "epoch": 2.0736788418533485,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.3624,
      "step": 6607
    },
    {
      "epoch": 2.073992702734513,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.6507,
      "step": 6608
    },
    {
      "epoch": 2.0743065636156772,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002,
      "loss": 1.6727,
      "step": 6609
    },
    {
      "epoch": 2.0746204244968416,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002,
      "loss": 1.6514,
      "step": 6610
    },
    {
      "epoch": 2.0749342853780064,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.4112,
      "step": 6611
    },
    {
      "epoch": 2.0752481462591708,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.4614,
      "step": 6612
    },
    {
      "epoch": 2.075562007140335,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.3049,
      "step": 6613
    },
    {
      "epoch": 2.0758758680214995,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002,
      "loss": 1.5404,
      "step": 6614
    },
    {
      "epoch": 2.076189728902664,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 2.244,
      "step": 6615
    },
    {
      "epoch": 2.076503589783828,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.6382,
      "step": 6616
    },
    {
      "epoch": 2.0768174506649926,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.4715,
      "step": 6617
    },
    {
      "epoch": 2.0771313115461574,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.3005,
      "step": 6618
    },
    {
      "epoch": 2.0774451724273217,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.6693,
      "step": 6619
    },
    {
      "epoch": 2.077759033308486,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.3043,
      "step": 6620
    },
    {
      "epoch": 2.0780728941896505,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.5473,
      "step": 6621
    },
    {
      "epoch": 2.078386755070815,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002,
      "loss": 1.8934,
      "step": 6622
    },
    {
      "epoch": 2.078700615951979,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.5022,
      "step": 6623
    },
    {
      "epoch": 2.0790144768331436,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.1386,
      "step": 6624
    },
    {
      "epoch": 2.0793283377143084,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.226,
      "step": 6625
    },
    {
      "epoch": 2.0796421985954727,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.3001,
      "step": 6626
    },
    {
      "epoch": 2.079956059476637,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.1973,
      "step": 6627
    },
    {
      "epoch": 2.0802699203578014,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.3712,
      "step": 6628
    },
    {
      "epoch": 2.080583781238966,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.4416,
      "step": 6629
    },
    {
      "epoch": 2.08089764212013,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.2155,
      "step": 6630
    },
    {
      "epoch": 2.0812115030012945,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.3736,
      "step": 6631
    },
    {
      "epoch": 2.081525363882459,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.2758,
      "step": 6632
    },
    {
      "epoch": 2.0818392247636237,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.352,
      "step": 6633
    },
    {
      "epoch": 2.082153085644788,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.1818,
      "step": 6634
    },
    {
      "epoch": 2.0824669465259524,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.1351,
      "step": 6635
    },
    {
      "epoch": 2.082780807407117,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.1143,
      "step": 6636
    },
    {
      "epoch": 2.083094668288281,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.2945,
      "step": 6637
    },
    {
      "epoch": 2.0834085291694455,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.1505,
      "step": 6638
    },
    {
      "epoch": 2.08372239005061,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.1879,
      "step": 6639
    },
    {
      "epoch": 2.0840362509317747,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.1508,
      "step": 6640
    },
    {
      "epoch": 2.084350111812939,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.2983,
      "step": 6641
    },
    {
      "epoch": 2.0846639726941034,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.1299,
      "step": 6642
    },
    {
      "epoch": 2.084977833575268,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.2565,
      "step": 6643
    },
    {
      "epoch": 2.085291694456432,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.112,
      "step": 6644
    },
    {
      "epoch": 2.0856055553375965,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.3483,
      "step": 6645
    },
    {
      "epoch": 2.085919416218761,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.1211,
      "step": 6646
    },
    {
      "epoch": 2.0862332770999252,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.0277,
      "step": 6647
    },
    {
      "epoch": 2.08654713798109,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.2961,
      "step": 6648
    },
    {
      "epoch": 2.0868609988622544,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.2166,
      "step": 6649
    },
    {
      "epoch": 2.0871748597434188,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.1088,
      "step": 6650
    },
    {
      "epoch": 2.087488720624583,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.2482,
      "step": 6651
    },
    {
      "epoch": 2.0878025815057475,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.2477,
      "step": 6652
    },
    {
      "epoch": 2.088116442386912,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.3587,
      "step": 6653
    },
    {
      "epoch": 2.088430303268076,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.4881,
      "step": 6654
    },
    {
      "epoch": 2.088744164149241,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.436,
      "step": 6655
    },
    {
      "epoch": 2.0890580250304054,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.517,
      "step": 6656
    },
    {
      "epoch": 2.0893718859115697,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.6483,
      "step": 6657
    },
    {
      "epoch": 2.089685746792734,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.2482,
      "step": 6658
    },
    {
      "epoch": 2.0899996076738985,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.3802,
      "step": 6659
    },
    {
      "epoch": 2.090313468555063,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.4399,
      "step": 6660
    },
    {
      "epoch": 2.090627329436227,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002,
      "loss": 1.6742,
      "step": 6661
    },
    {
      "epoch": 2.090941190317392,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.3579,
      "step": 6662
    },
    {
      "epoch": 2.0912550511985564,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.8549,
      "step": 6663
    },
    {
      "epoch": 2.0915689120797207,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.7403,
      "step": 6664
    },
    {
      "epoch": 2.091882772960885,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002,
      "loss": 1.4915,
      "step": 6665
    },
    {
      "epoch": 2.0921966338420495,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.6108,
      "step": 6666
    },
    {
      "epoch": 2.092510494723214,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 1.7772,
      "step": 6667
    },
    {
      "epoch": 2.092824355604378,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.3944,
      "step": 6668
    },
    {
      "epoch": 2.093138216485543,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.5062,
      "step": 6669
    },
    {
      "epoch": 2.0934520773667074,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.2579,
      "step": 6670
    },
    {
      "epoch": 2.0937659382478717,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002,
      "loss": 1.3374,
      "step": 6671
    },
    {
      "epoch": 2.094079799129036,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002,
      "loss": 2.0661,
      "step": 6672
    },
    {
      "epoch": 2.0943936600102004,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.3999,
      "step": 6673
    },
    {
      "epoch": 2.094707520891365,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.232,
      "step": 6674
    },
    {
      "epoch": 2.095021381772529,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.3604,
      "step": 6675
    },
    {
      "epoch": 2.095335242653694,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.2852,
      "step": 6676
    },
    {
      "epoch": 2.0956491035348583,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.1618,
      "step": 6677
    },
    {
      "epoch": 2.0959629644160227,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.1367,
      "step": 6678
    },
    {
      "epoch": 2.096276825297187,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2307,
      "step": 6679
    },
    {
      "epoch": 2.0965906861783514,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.3044,
      "step": 6680
    },
    {
      "epoch": 2.096904547059516,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.1697,
      "step": 6681
    },
    {
      "epoch": 2.09721840794068,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.1658,
      "step": 6682
    },
    {
      "epoch": 2.0975322688218445,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.3785,
      "step": 6683
    },
    {
      "epoch": 2.0978461297030093,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.3131,
      "step": 6684
    },
    {
      "epoch": 2.0981599905841737,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.1546,
      "step": 6685
    },
    {
      "epoch": 2.098473851465338,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.1966,
      "step": 6686
    },
    {
      "epoch": 2.0987877123465024,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.0451,
      "step": 6687
    },
    {
      "epoch": 2.0991015732276668,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.2748,
      "step": 6688
    },
    {
      "epoch": 2.099415434108831,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.122,
      "step": 6689
    },
    {
      "epoch": 2.0997292949899955,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.0628,
      "step": 6690
    },
    {
      "epoch": 2.1000431558711603,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.091,
      "step": 6691
    },
    {
      "epoch": 2.1003570167523247,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.17,
      "step": 6692
    },
    {
      "epoch": 2.100670877633489,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2284,
      "step": 6693
    },
    {
      "epoch": 2.1009847385146534,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.2128,
      "step": 6694
    },
    {
      "epoch": 2.1012985993958178,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 0.9812,
      "step": 6695
    },
    {
      "epoch": 2.101612460276982,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.0904,
      "step": 6696
    },
    {
      "epoch": 2.1019263211581465,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.1386,
      "step": 6697
    },
    {
      "epoch": 2.1022401820393113,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.2994,
      "step": 6698
    },
    {
      "epoch": 2.1025540429204757,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.1322,
      "step": 6699
    },
    {
      "epoch": 2.10286790380164,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.1264,
      "step": 6700
    },
    {
      "epoch": 2.1031817646828044,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.6199,
      "step": 6701
    },
    {
      "epoch": 2.1034956255639687,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.6834,
      "step": 6702
    },
    {
      "epoch": 2.103809486445133,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.283,
      "step": 6703
    },
    {
      "epoch": 2.1041233473262975,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.4683,
      "step": 6704
    },
    {
      "epoch": 2.104437208207462,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.4372,
      "step": 6705
    },
    {
      "epoch": 2.1047510690886266,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.4158,
      "step": 6706
    },
    {
      "epoch": 2.105064929969791,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.4875,
      "step": 6707
    },
    {
      "epoch": 2.1053787908509554,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.571,
      "step": 6708
    },
    {
      "epoch": 2.1056926517321197,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.564,
      "step": 6709
    },
    {
      "epoch": 2.106006512613284,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 1.6821,
      "step": 6710
    },
    {
      "epoch": 2.1063203734944485,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.6252,
      "step": 6711
    },
    {
      "epoch": 2.106634234375613,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.2611,
      "step": 6712
    },
    {
      "epoch": 2.1069480952567776,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.285,
      "step": 6713
    },
    {
      "epoch": 2.107261956137942,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0002,
      "loss": 1.5855,
      "step": 6714
    },
    {
      "epoch": 2.1075758170191063,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.8855,
      "step": 6715
    },
    {
      "epoch": 2.1078896779002707,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.5941,
      "step": 6716
    },
    {
      "epoch": 2.108203538781435,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.6059,
      "step": 6717
    },
    {
      "epoch": 2.1085173996625994,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.7762,
      "step": 6718
    },
    {
      "epoch": 2.108831260543764,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.3035,
      "step": 6719
    },
    {
      "epoch": 2.1091451214249286,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 1.5713,
      "step": 6720
    },
    {
      "epoch": 2.109458982306093,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002,
      "loss": 1.6413,
      "step": 6721
    },
    {
      "epoch": 2.1097728431872573,
      "grad_norm": 0.84375,
      "learning_rate": 0.0002,
      "loss": 2.1769,
      "step": 6722
    },
    {
      "epoch": 2.1100867040684217,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.7023,
      "step": 6723
    },
    {
      "epoch": 2.110400564949586,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.2982,
      "step": 6724
    },
    {
      "epoch": 2.1107144258307504,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.1938,
      "step": 6725
    },
    {
      "epoch": 2.111028286711915,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.4027,
      "step": 6726
    },
    {
      "epoch": 2.1113421475930796,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.1081,
      "step": 6727
    },
    {
      "epoch": 2.111656008474244,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.1019,
      "step": 6728
    },
    {
      "epoch": 2.1119698693554083,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.3549,
      "step": 6729
    },
    {
      "epoch": 2.1122837302365727,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.2879,
      "step": 6730
    },
    {
      "epoch": 2.112597591117737,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.2641,
      "step": 6731
    },
    {
      "epoch": 2.1129114519989014,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.1367,
      "step": 6732
    },
    {
      "epoch": 2.1129114519989014,
      "eval_loss": 1.8845981359481812,
      "eval_runtime": 123.4066,
      "eval_samples_per_second": 8.103,
      "eval_steps_per_second": 8.103,
      "step": 6732
    },
    {
      "epoch": 2.1129114519989014,
      "mmlu_eval_accuracy": 0.4160017211600709,
      "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": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.0,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "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": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.65,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.5,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.25,
      "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": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
      "mmlu_eval_accuracy_miscellaneous": 0.5232558139534884,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.3058823529411765,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.3333333333333333,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.1772874047813036,
      "step": 6732
    },
    {
      "epoch": 2.1132253128800658,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.15,
      "step": 6733
    },
    {
      "epoch": 2.11353917376123,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.1879,
      "step": 6734
    },
    {
      "epoch": 2.113853034642395,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.2414,
      "step": 6735
    },
    {
      "epoch": 2.1141668955235593,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.1459,
      "step": 6736
    },
    {
      "epoch": 2.1144807564047237,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.0652,
      "step": 6737
    },
    {
      "epoch": 2.114794617285888,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.3751,
      "step": 6738
    },
    {
      "epoch": 2.1151084781670524,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.1655,
      "step": 6739
    },
    {
      "epoch": 2.1154223390482167,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.1875,
      "step": 6740
    },
    {
      "epoch": 2.115736199929381,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.1378,
      "step": 6741
    },
    {
      "epoch": 2.116050060810546,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.0022,
      "step": 6742
    },
    {
      "epoch": 2.1163639216917103,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.1581,
      "step": 6743
    },
    {
      "epoch": 2.1166777825728746,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.117,
      "step": 6744
    },
    {
      "epoch": 2.116991643454039,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.2402,
      "step": 6745
    },
    {
      "epoch": 2.1173055043352034,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.2288,
      "step": 6746
    },
    {
      "epoch": 2.1176193652163677,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.056,
      "step": 6747
    },
    {
      "epoch": 2.117933226097532,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.3126,
      "step": 6748
    },
    {
      "epoch": 2.118247086978697,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.049,
      "step": 6749
    },
    {
      "epoch": 2.1185609478598613,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.1198,
      "step": 6750
    },
    {
      "epoch": 2.1188748087410256,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.2,
      "step": 6751
    },
    {
      "epoch": 2.11918866962219,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.5001,
      "step": 6752
    },
    {
      "epoch": 2.1195025305033544,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.1871,
      "step": 6753
    },
    {
      "epoch": 2.1198163913845187,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.4034,
      "step": 6754
    },
    {
      "epoch": 2.120130252265683,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 1.3546,
      "step": 6755
    },
    {
      "epoch": 2.1204441131468474,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.7463,
      "step": 6756
    },
    {
      "epoch": 2.1207579740280122,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.5346,
      "step": 6757
    },
    {
      "epoch": 2.1210718349091766,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.2616,
      "step": 6758
    },
    {
      "epoch": 2.121385695790341,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.5472,
      "step": 6759
    },
    {
      "epoch": 2.1216995566715053,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.5406,
      "step": 6760
    },
    {
      "epoch": 2.1220134175526697,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.3958,
      "step": 6761
    },
    {
      "epoch": 2.122327278433834,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.6495,
      "step": 6762
    },
    {
      "epoch": 2.1226411393149984,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002,
      "loss": 1.7449,
      "step": 6763
    },
    {
      "epoch": 2.1229550001961632,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 1.8357,
      "step": 6764
    },
    {
      "epoch": 2.1232688610773276,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.804,
      "step": 6765
    },
    {
      "epoch": 2.123582721958492,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.5576,
      "step": 6766
    },
    {
      "epoch": 2.1238965828396563,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.4532,
      "step": 6767
    },
    {
      "epoch": 2.1242104437208207,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.497,
      "step": 6768
    },
    {
      "epoch": 2.124524304601985,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.3321,
      "step": 6769
    },
    {
      "epoch": 2.1248381654831494,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.3592,
      "step": 6770
    },
    {
      "epoch": 2.125152026364314,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0002,
      "loss": 1.8384,
      "step": 6771
    },
    {
      "epoch": 2.1254658872454786,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 2.0317,
      "step": 6772
    },
    {
      "epoch": 2.125779748126643,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.875,
      "step": 6773
    },
    {
      "epoch": 2.1260936090078073,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.2051,
      "step": 6774
    },
    {
      "epoch": 2.1264074698889717,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3364,
      "step": 6775
    },
    {
      "epoch": 2.126721330770136,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.292,
      "step": 6776
    },
    {
      "epoch": 2.1270351916513004,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.1784,
      "step": 6777
    },
    {
      "epoch": 2.127349052532465,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.1701,
      "step": 6778
    },
    {
      "epoch": 2.1276629134136296,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2025,
      "step": 6779
    },
    {
      "epoch": 2.127976774294794,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.2073,
      "step": 6780
    },
    {
      "epoch": 2.1282906351759583,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.1526,
      "step": 6781
    },
    {
      "epoch": 2.1286044960571227,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2911,
      "step": 6782
    },
    {
      "epoch": 2.128918356938287,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.2722,
      "step": 6783
    },
    {
      "epoch": 2.1292322178194514,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.0901,
      "step": 6784
    },
    {
      "epoch": 2.129546078700616,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.1132,
      "step": 6785
    },
    {
      "epoch": 2.1298599395817805,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.1682,
      "step": 6786
    },
    {
      "epoch": 2.130173800462945,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.2096,
      "step": 6787
    },
    {
      "epoch": 2.1304876613441093,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.3941,
      "step": 6788
    },
    {
      "epoch": 2.1308015222252736,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.0099,
      "step": 6789
    },
    {
      "epoch": 2.131115383106438,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.2074,
      "step": 6790
    },
    {
      "epoch": 2.1314292439876024,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.114,
      "step": 6791
    },
    {
      "epoch": 2.1317431048687667,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.1693,
      "step": 6792
    },
    {
      "epoch": 2.1320569657499315,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.1457,
      "step": 6793
    },
    {
      "epoch": 2.132370826631096,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.1432,
      "step": 6794
    },
    {
      "epoch": 2.1326846875122603,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.1823,
      "step": 6795
    },
    {
      "epoch": 2.1329985483934246,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.2051,
      "step": 6796
    },
    {
      "epoch": 2.133312409274589,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.1095,
      "step": 6797
    },
    {
      "epoch": 2.1336262701557533,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.0579,
      "step": 6798
    },
    {
      "epoch": 2.1339401310369177,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.4059,
      "step": 6799
    },
    {
      "epoch": 2.1342539919180825,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.3724,
      "step": 6800
    },
    {
      "epoch": 2.134567852799247,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.3845,
      "step": 6801
    },
    {
      "epoch": 2.1348817136804112,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 1.3795,
      "step": 6802
    },
    {
      "epoch": 2.1351955745615756,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.3792,
      "step": 6803
    },
    {
      "epoch": 2.13550943544274,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.271,
      "step": 6804
    },
    {
      "epoch": 2.1358232963239043,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.3843,
      "step": 6805
    },
    {
      "epoch": 2.1361371572050687,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.2448,
      "step": 6806
    },
    {
      "epoch": 2.136451018086233,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.5088,
      "step": 6807
    },
    {
      "epoch": 2.136764878967398,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.6733,
      "step": 6808
    },
    {
      "epoch": 2.1370787398485622,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.533,
      "step": 6809
    },
    {
      "epoch": 2.1373926007297266,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.6491,
      "step": 6810
    },
    {
      "epoch": 2.137706461610891,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.5106,
      "step": 6811
    },
    {
      "epoch": 2.1380203224920553,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.5473,
      "step": 6812
    },
    {
      "epoch": 2.1383341833732197,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.2842,
      "step": 6813
    },
    {
      "epoch": 2.138648044254384,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.8684,
      "step": 6814
    },
    {
      "epoch": 2.138961905135549,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.5224,
      "step": 6815
    },
    {
      "epoch": 2.139275766016713,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.7703,
      "step": 6816
    },
    {
      "epoch": 2.1395896268978776,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002,
      "loss": 1.9016,
      "step": 6817
    },
    {
      "epoch": 2.139903487779042,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 1.7186,
      "step": 6818
    },
    {
      "epoch": 2.1402173486602063,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002,
      "loss": 1.9751,
      "step": 6819
    },
    {
      "epoch": 2.1405312095413707,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.6175,
      "step": 6820
    },
    {
      "epoch": 2.140845070422535,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.618,
      "step": 6821
    },
    {
      "epoch": 2.1411589313037,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002,
      "loss": 2.2178,
      "step": 6822
    },
    {
      "epoch": 2.141472792184864,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.6464,
      "step": 6823
    },
    {
      "epoch": 2.1417866530660286,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.2822,
      "step": 6824
    },
    {
      "epoch": 2.142100513947193,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.1868,
      "step": 6825
    },
    {
      "epoch": 2.1424143748283573,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.1706,
      "step": 6826
    },
    {
      "epoch": 2.1427282357095216,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.2724,
      "step": 6827
    },
    {
      "epoch": 2.143042096590686,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.2537,
      "step": 6828
    },
    {
      "epoch": 2.143355957471851,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.5385,
      "step": 6829
    },
    {
      "epoch": 2.143669818353015,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.3214,
      "step": 6830
    },
    {
      "epoch": 2.1439836792341795,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.296,
      "step": 6831
    },
    {
      "epoch": 2.144297540115344,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.278,
      "step": 6832
    },
    {
      "epoch": 2.1446114009965083,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.2616,
      "step": 6833
    },
    {
      "epoch": 2.1449252618776726,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.225,
      "step": 6834
    },
    {
      "epoch": 2.145239122758837,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.1737,
      "step": 6835
    },
    {
      "epoch": 2.145552983640002,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2207,
      "step": 6836
    },
    {
      "epoch": 2.145866844521166,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.1458,
      "step": 6837
    },
    {
      "epoch": 2.1461807054023305,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.2683,
      "step": 6838
    },
    {
      "epoch": 2.146494566283495,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.1668,
      "step": 6839
    },
    {
      "epoch": 2.1468084271646592,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.2434,
      "step": 6840
    },
    {
      "epoch": 2.1471222880458236,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.1644,
      "step": 6841
    },
    {
      "epoch": 2.147436148926988,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0002,
      "loss": 1.3738,
      "step": 6842
    },
    {
      "epoch": 2.1477500098081523,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.2088,
      "step": 6843
    },
    {
      "epoch": 2.148063870689317,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.3195,
      "step": 6844
    },
    {
      "epoch": 2.1483777315704815,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.2277,
      "step": 6845
    },
    {
      "epoch": 2.148691592451646,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.1324,
      "step": 6846
    },
    {
      "epoch": 2.1490054533328102,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.149,
      "step": 6847
    },
    {
      "epoch": 2.1493193142139746,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.3268,
      "step": 6848
    },
    {
      "epoch": 2.149633175095139,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.1116,
      "step": 6849
    },
    {
      "epoch": 2.1499470359763033,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.118,
      "step": 6850
    },
    {
      "epoch": 2.150260896857468,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.6066,
      "step": 6851
    },
    {
      "epoch": 2.1505747577386325,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.2709,
      "step": 6852
    },
    {
      "epoch": 2.150888618619797,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.3302,
      "step": 6853
    },
    {
      "epoch": 2.151202479500961,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.3063,
      "step": 6854
    },
    {
      "epoch": 2.1515163403821256,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002,
      "loss": 1.6667,
      "step": 6855
    },
    {
      "epoch": 2.15183020126329,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.4132,
      "step": 6856
    },
    {
      "epoch": 2.1521440621444543,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.4771,
      "step": 6857
    },
    {
      "epoch": 2.1524579230256187,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.428,
      "step": 6858
    },
    {
      "epoch": 2.1527717839067835,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.6305,
      "step": 6859
    },
    {
      "epoch": 2.153085644787948,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.5886,
      "step": 6860
    },
    {
      "epoch": 2.153399505669112,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.5487,
      "step": 6861
    },
    {
      "epoch": 2.1537133665502766,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002,
      "loss": 2.0663,
      "step": 6862
    },
    {
      "epoch": 2.154027227431441,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.6055,
      "step": 6863
    },
    {
      "epoch": 2.1543410883126053,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 1.6957,
      "step": 6864
    },
    {
      "epoch": 2.1546549491937697,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.2985,
      "step": 6865
    },
    {
      "epoch": 2.1549688100749345,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002,
      "loss": 2.0634,
      "step": 6866
    },
    {
      "epoch": 2.155282670956099,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0002,
      "loss": 1.799,
      "step": 6867
    },
    {
      "epoch": 2.155596531837263,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.7505,
      "step": 6868
    },
    {
      "epoch": 2.1559103927184275,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0002,
      "loss": 1.3157,
      "step": 6869
    },
    {
      "epoch": 2.156224253599592,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.5125,
      "step": 6870
    },
    {
      "epoch": 2.1565381144807563,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.4366,
      "step": 6871
    },
    {
      "epoch": 2.1568519753619206,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 2.0376,
      "step": 6872
    },
    {
      "epoch": 2.1571658362430854,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.9673,
      "step": 6873
    },
    {
      "epoch": 2.15747969712425,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.1919,
      "step": 6874
    },
    {
      "epoch": 2.157793558005414,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.3039,
      "step": 6875
    },
    {
      "epoch": 2.1581074188865785,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.2509,
      "step": 6876
    },
    {
      "epoch": 2.158421279767743,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.2111,
      "step": 6877
    },
    {
      "epoch": 2.1587351406489073,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.3505,
      "step": 6878
    },
    {
      "epoch": 2.1590490015300716,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.1955,
      "step": 6879
    },
    {
      "epoch": 2.1593628624112364,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.3208,
      "step": 6880
    },
    {
      "epoch": 2.159676723292401,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.177,
      "step": 6881
    },
    {
      "epoch": 2.159990584173565,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.1524,
      "step": 6882
    },
    {
      "epoch": 2.1603044450547295,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.1777,
      "step": 6883
    },
    {
      "epoch": 2.160618305935894,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.2514,
      "step": 6884
    },
    {
      "epoch": 2.1609321668170582,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.2055,
      "step": 6885
    },
    {
      "epoch": 2.1612460276982226,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.1553,
      "step": 6886
    },
    {
      "epoch": 2.1615598885793874,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.2612,
      "step": 6887
    },
    {
      "epoch": 2.1618737494605518,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.1441,
      "step": 6888
    },
    {
      "epoch": 2.162187610341716,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.1753,
      "step": 6889
    },
    {
      "epoch": 2.1625014712228805,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.1841,
      "step": 6890
    },
    {
      "epoch": 2.162815332104045,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.0857,
      "step": 6891
    },
    {
      "epoch": 2.1631291929852092,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.3838,
      "step": 6892
    },
    {
      "epoch": 2.1634430538663736,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.4182,
      "step": 6893
    },
    {
      "epoch": 2.163756914747538,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.3201,
      "step": 6894
    },
    {
      "epoch": 2.1640707756287028,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 0.9924,
      "step": 6895
    },
    {
      "epoch": 2.164384636509867,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.1636,
      "step": 6896
    },
    {
      "epoch": 2.1646984973910315,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.1477,
      "step": 6897
    },
    {
      "epoch": 2.165012358272196,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.147,
      "step": 6898
    },
    {
      "epoch": 2.16532621915336,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.2233,
      "step": 6899
    },
    {
      "epoch": 2.1656400800345246,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.4101,
      "step": 6900
    },
    {
      "epoch": 2.165953940915689,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.0818,
      "step": 6901
    },
    {
      "epoch": 2.1662678017968537,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.5216,
      "step": 6902
    },
    {
      "epoch": 2.166581662678018,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.611,
      "step": 6903
    },
    {
      "epoch": 2.1668955235591825,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.5473,
      "step": 6904
    },
    {
      "epoch": 2.167209384440347,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.4276,
      "step": 6905
    },
    {
      "epoch": 2.167523245321511,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.6521,
      "step": 6906
    },
    {
      "epoch": 2.1678371062026756,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.5468,
      "step": 6907
    },
    {
      "epoch": 2.16815096708384,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 1.5741,
      "step": 6908
    },
    {
      "epoch": 2.1684648279650043,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.5127,
      "step": 6909
    },
    {
      "epoch": 2.168778688846169,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002,
      "loss": 1.6309,
      "step": 6910
    },
    {
      "epoch": 2.1690925497273335,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.3051,
      "step": 6911
    },
    {
      "epoch": 2.169406410608498,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.7332,
      "step": 6912
    },
    {
      "epoch": 2.169720271489662,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002,
      "loss": 1.7025,
      "step": 6913
    },
    {
      "epoch": 2.1700341323708265,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.4458,
      "step": 6914
    },
    {
      "epoch": 2.170347993251991,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.2982,
      "step": 6915
    },
    {
      "epoch": 2.1706618541331553,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.9314,
      "step": 6916
    },
    {
      "epoch": 2.17097571501432,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 1.8198,
      "step": 6917
    },
    {
      "epoch": 2.1712895758954844,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 1.6126,
      "step": 6918
    },
    {
      "epoch": 2.171603436776649,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.4458,
      "step": 6919
    },
    {
      "epoch": 2.171603436776649,
      "eval_loss": 1.9108567237854004,
      "eval_runtime": 123.0755,
      "eval_samples_per_second": 8.125,
      "eval_steps_per_second": 8.125,
      "step": 6919
    },
    {
      "epoch": 2.171603436776649,
      "mmlu_eval_accuracy": 0.41087949761996856,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.0,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
      "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": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4418604651162791,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.5833333333333334,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.38461538461538464,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.25,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.27,
      "mmlu_eval_accuracy_nutrition": 0.3939393939393939,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.2857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.27647058823529413,
      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.2962962962962963,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.3333333333333333,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.2637697486590125,
      "step": 6919
    },
    {
      "epoch": 2.171917297657813,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.2948,
      "step": 6920
    },
    {
      "epoch": 2.1722311585389775,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.8296,
      "step": 6921
    },
    {
      "epoch": 2.172545019420142,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002,
      "loss": 1.8349,
      "step": 6922
    },
    {
      "epoch": 2.1728588803013063,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.6083,
      "step": 6923
    },
    {
      "epoch": 2.173172741182471,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.4099,
      "step": 6924
    },
    {
      "epoch": 2.1734866020636354,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.159,
      "step": 6925
    },
    {
      "epoch": 2.1738004629448,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.399,
      "step": 6926
    },
    {
      "epoch": 2.174114323825964,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.3964,
      "step": 6927
    },
    {
      "epoch": 2.1744281847071285,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.1958,
      "step": 6928
    },
    {
      "epoch": 2.174742045588293,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.3266,
      "step": 6929
    },
    {
      "epoch": 2.1750559064694572,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.3394,
      "step": 6930
    },
    {
      "epoch": 2.175369767350622,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.2084,
      "step": 6931
    },
    {
      "epoch": 2.1756836282317864,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.4368,
      "step": 6932
    },
    {
      "epoch": 2.1759974891129508,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.1479,
      "step": 6933
    },
    {
      "epoch": 2.176311349994115,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.2597,
      "step": 6934
    },
    {
      "epoch": 2.1766252108752795,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.0972,
      "step": 6935
    },
    {
      "epoch": 2.176939071756444,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.3139,
      "step": 6936
    },
    {
      "epoch": 2.177252932637608,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.2088,
      "step": 6937
    },
    {
      "epoch": 2.177566793518773,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.2534,
      "step": 6938
    },
    {
      "epoch": 2.1778806543999374,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.2668,
      "step": 6939
    },
    {
      "epoch": 2.1781945152811018,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.3018,
      "step": 6940
    },
    {
      "epoch": 2.178508376162266,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.2134,
      "step": 6941
    },
    {
      "epoch": 2.1788222370434305,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.1556,
      "step": 6942
    },
    {
      "epoch": 2.179136097924595,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.1412,
      "step": 6943
    },
    {
      "epoch": 2.179449958805759,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.0205,
      "step": 6944
    },
    {
      "epoch": 2.1797638196869236,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.1013,
      "step": 6945
    },
    {
      "epoch": 2.1800776805680884,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.1996,
      "step": 6946
    },
    {
      "epoch": 2.1803915414492527,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.1859,
      "step": 6947
    },
    {
      "epoch": 2.180705402330417,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.1421,
      "step": 6948
    },
    {
      "epoch": 2.1810192632115815,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.2526,
      "step": 6949
    },
    {
      "epoch": 2.181333124092746,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.3302,
      "step": 6950
    },
    {
      "epoch": 2.18164698497391,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.5186,
      "step": 6951
    },
    {
      "epoch": 2.1819608458550745,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.2947,
      "step": 6952
    },
    {
      "epoch": 2.1822747067362394,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.526,
      "step": 6953
    },
    {
      "epoch": 2.1825885676174037,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 1.2987,
      "step": 6954
    },
    {
      "epoch": 2.182902428498568,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.2986,
      "step": 6955
    },
    {
      "epoch": 2.1832162893797324,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002,
      "loss": 1.4097,
      "step": 6956
    },
    {
      "epoch": 2.183530150260897,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.6956,
      "step": 6957
    },
    {
      "epoch": 2.183844011142061,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002,
      "loss": 1.4406,
      "step": 6958
    },
    {
      "epoch": 2.1841578720232255,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.5417,
      "step": 6959
    },
    {
      "epoch": 2.18447173290439,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.5466,
      "step": 6960
    },
    {
      "epoch": 2.1847855937855547,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.4767,
      "step": 6961
    },
    {
      "epoch": 2.185099454666719,
      "grad_norm": 1.125,
      "learning_rate": 0.0002,
      "loss": 1.9594,
      "step": 6962
    },
    {
      "epoch": 2.1854133155478834,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002,
      "loss": 1.803,
      "step": 6963
    },
    {
      "epoch": 2.185727176429048,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 2.0183,
      "step": 6964
    },
    {
      "epoch": 2.186041037310212,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.8014,
      "step": 6965
    },
    {
      "epoch": 2.1863548981913765,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.3876,
      "step": 6966
    },
    {
      "epoch": 2.186668759072541,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 2.0449,
      "step": 6967
    },
    {
      "epoch": 2.1869826199537057,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 1.6801,
      "step": 6968
    },
    {
      "epoch": 2.18729648083487,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.2796,
      "step": 6969
    },
    {
      "epoch": 2.1876103417160344,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.5317,
      "step": 6970
    },
    {
      "epoch": 2.1879242025971988,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.6394,
      "step": 6971
    },
    {
      "epoch": 2.188238063478363,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 1.4911,
      "step": 6972
    },
    {
      "epoch": 2.1885519243595275,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.6311,
      "step": 6973
    },
    {
      "epoch": 2.188865785240692,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.2183,
      "step": 6974
    },
    {
      "epoch": 2.1891796461218567,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.3587,
      "step": 6975
    },
    {
      "epoch": 2.189493507003021,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.4424,
      "step": 6976
    },
    {
      "epoch": 2.1898073678841854,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.1984,
      "step": 6977
    },
    {
      "epoch": 2.1901212287653498,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.3143,
      "step": 6978
    },
    {
      "epoch": 2.190435089646514,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.3413,
      "step": 6979
    },
    {
      "epoch": 2.1907489505276785,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2222,
      "step": 6980
    },
    {
      "epoch": 2.191062811408843,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.1879,
      "step": 6981
    },
    {
      "epoch": 2.1913766722900077,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.1602,
      "step": 6982
    },
    {
      "epoch": 2.191690533171172,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.3725,
      "step": 6983
    },
    {
      "epoch": 2.1920043940523364,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.2544,
      "step": 6984
    },
    {
      "epoch": 2.1923182549335007,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.2809,
      "step": 6985
    },
    {
      "epoch": 2.192632115814665,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.0753,
      "step": 6986
    },
    {
      "epoch": 2.1929459766958295,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.1883,
      "step": 6987
    },
    {
      "epoch": 2.193259837576994,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.1687,
      "step": 6988
    },
    {
      "epoch": 2.1935736984581586,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.2067,
      "step": 6989
    },
    {
      "epoch": 2.193887559339323,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.0681,
      "step": 6990
    },
    {
      "epoch": 2.1942014202204874,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.0512,
      "step": 6991
    },
    {
      "epoch": 2.1945152811016517,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.2436,
      "step": 6992
    },
    {
      "epoch": 2.194829141982816,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.0949,
      "step": 6993
    },
    {
      "epoch": 2.1951430028639805,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.1986,
      "step": 6994
    },
    {
      "epoch": 2.195456863745145,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.2812,
      "step": 6995
    },
    {
      "epoch": 2.1957707246263096,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.2332,
      "step": 6996
    },
    {
      "epoch": 2.196084585507474,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.0782,
      "step": 6997
    },
    {
      "epoch": 2.1963984463886383,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.2484,
      "step": 6998
    },
    {
      "epoch": 2.1967123072698027,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 1.347,
      "step": 6999
    },
    {
      "epoch": 2.197026168150967,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.2324,
      "step": 7000
    },
    {
      "epoch": 2.1973400290321314,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.2807,
      "step": 7001
    },
    {
      "epoch": 2.197653889913296,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.6561,
      "step": 7002
    },
    {
      "epoch": 2.19796775079446,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 1.2975,
      "step": 7003
    },
    {
      "epoch": 2.198281611675625,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 1.3722,
      "step": 7004
    },
    {
      "epoch": 2.1985954725567893,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 1.5477,
      "step": 7005
    },
    {
      "epoch": 2.1989093334379537,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 1.5554,
      "step": 7006
    },
    {
      "epoch": 2.199223194319118,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.5507,
      "step": 7007
    },
    {
      "epoch": 2.1995370552002824,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.4281,
      "step": 7008
    },
    {
      "epoch": 2.199850916081447,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.5234,
      "step": 7009
    },
    {
      "epoch": 2.200164776962611,
      "grad_norm": 1.125,
      "learning_rate": 0.0002,
      "loss": 1.9167,
      "step": 7010
    },
    {
      "epoch": 2.200478637843776,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002,
      "loss": 1.6815,
      "step": 7011
    },
    {
      "epoch": 2.2007924987249403,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.6578,
      "step": 7012
    },
    {
      "epoch": 2.2011063596061047,
      "grad_norm": 0.90625,
      "learning_rate": 0.0002,
      "loss": 1.7545,
      "step": 7013
    },
    {
      "epoch": 2.201420220487269,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.8026,
      "step": 7014
    },
    {
      "epoch": 2.2017340813684334,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.4013,
      "step": 7015
    },
    {
      "epoch": 2.2020479422495978,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002,
      "loss": 1.7612,
      "step": 7016
    },
    {
      "epoch": 2.202361803130762,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 1.5383,
      "step": 7017
    },
    {
      "epoch": 2.2026756640119265,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.3195,
      "step": 7018
    },
    {
      "epoch": 2.2029895248930913,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.4813,
      "step": 7019
    },
    {
      "epoch": 2.2033033857742557,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.6449,
      "step": 7020
    },
    {
      "epoch": 2.20361724665542,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 1.4753,
      "step": 7021
    },
    {
      "epoch": 2.2039311075365844,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002,
      "loss": 1.9823,
      "step": 7022
    },
    {
      "epoch": 2.2042449684177488,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.7635,
      "step": 7023
    },
    {
      "epoch": 2.204558829298913,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.3181,
      "step": 7024
    },
    {
      "epoch": 2.2048726901800775,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.1876,
      "step": 7025
    },
    {
      "epoch": 2.2051865510612423,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.2938,
      "step": 7026
    },
    {
      "epoch": 2.2055004119424066,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.265,
      "step": 7027
    },
    {
      "epoch": 2.205814272823571,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.21,
      "step": 7028
    },
    {
      "epoch": 2.2061281337047354,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 1.3629,
      "step": 7029
    },
    {
      "epoch": 2.2064419945858997,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.2804,
      "step": 7030
    },
    {
      "epoch": 2.206755855467064,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.3064,
      "step": 7031
    },
    {
      "epoch": 2.2070697163482285,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.2229,
      "step": 7032
    },
    {
      "epoch": 2.2073835772293933,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.1505,
      "step": 7033
    },
    {
      "epoch": 2.2076974381105576,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.2146,
      "step": 7034
    },
    {
      "epoch": 2.208011298991722,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.1732,
      "step": 7035
    },
    {
      "epoch": 2.2083251598728864,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.1353,
      "step": 7036
    },
    {
      "epoch": 2.2086390207540507,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.1966,
      "step": 7037
    },
    {
      "epoch": 2.208952881635215,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.2293,
      "step": 7038
    },
    {
      "epoch": 2.2092667425163794,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.1095,
      "step": 7039
    },
    {
      "epoch": 2.2095806033975443,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.0926,
      "step": 7040
    },
    {
      "epoch": 2.2098944642787086,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.0617,
      "step": 7041
    },
    {
      "epoch": 2.210208325159873,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.1509,
      "step": 7042
    },
    {
      "epoch": 2.2105221860410373,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.1196,
      "step": 7043
    },
    {
      "epoch": 2.2108360469222017,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.222,
      "step": 7044
    },
    {
      "epoch": 2.211149907803366,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.0556,
      "step": 7045
    },
    {
      "epoch": 2.2114637686845304,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.1312,
      "step": 7046
    },
    {
      "epoch": 2.2117776295656952,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.1114,
      "step": 7047
    },
    {
      "epoch": 2.2120914904468596,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.0889,
      "step": 7048
    },
    {
      "epoch": 2.212405351328024,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.066,
      "step": 7049
    },
    {
      "epoch": 2.2127192122091883,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 1.3159,
      "step": 7050
    },
    {
      "epoch": 2.2130330730903527,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.2005,
      "step": 7051
    },
    {
      "epoch": 2.213346933971517,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.4483,
      "step": 7052
    },
    {
      "epoch": 2.2136607948526814,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.5394,
      "step": 7053
    },
    {
      "epoch": 2.2139746557338458,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.2109,
      "step": 7054
    },
    {
      "epoch": 2.2142885166150106,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0002,
      "loss": 1.902,
      "step": 7055
    },
    {
      "epoch": 2.214602377496175,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.7741,
      "step": 7056
    },
    {
      "epoch": 2.2149162383773393,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 1.4993,
      "step": 7057
    },
    {
      "epoch": 2.2152300992585037,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.4955,
      "step": 7058
    },
    {
      "epoch": 2.215543960139668,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.4148,
      "step": 7059
    },
    {
      "epoch": 2.2158578210208324,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0002,
      "loss": 1.8686,
      "step": 7060
    },
    {
      "epoch": 2.2161716819019968,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0002,
      "loss": 1.7474,
      "step": 7061
    },
    {
      "epoch": 2.2164855427831616,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.6311,
      "step": 7062
    },
    {
      "epoch": 2.216799403664326,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.5525,
      "step": 7063
    },
    {
      "epoch": 2.2171132645454903,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.8235,
      "step": 7064
    },
    {
      "epoch": 2.2174271254266547,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.6988,
      "step": 7065
    },
    {
      "epoch": 2.217740986307819,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 2.1018,
      "step": 7066
    },
    {
      "epoch": 2.2180548471889834,
      "grad_norm": 1.25,
      "learning_rate": 0.0002,
      "loss": 1.8875,
      "step": 7067
    },
    {
      "epoch": 2.2183687080701477,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002,
      "loss": 2.0372,
      "step": 7068
    },
    {
      "epoch": 2.218682568951312,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.2988,
      "step": 7069
    },
    {
      "epoch": 2.218996429832477,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.523,
      "step": 7070
    },
    {
      "epoch": 2.2193102907136413,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.7166,
      "step": 7071
    },
    {
      "epoch": 2.2196241515948056,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.7636,
      "step": 7072
    },
    {
      "epoch": 2.21993801247597,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.4759,
      "step": 7073
    },
    {
      "epoch": 2.2202518733571344,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.2731,
      "step": 7074
    },
    {
      "epoch": 2.2205657342382987,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.1947,
      "step": 7075
    },
    {
      "epoch": 2.220879595119463,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.2262,
      "step": 7076
    },
    {
      "epoch": 2.221193456000628,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.4372,
      "step": 7077
    },
    {
      "epoch": 2.2215073168817923,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.2085,
      "step": 7078
    },
    {
      "epoch": 2.2218211777629566,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.2133,
      "step": 7079
    },
    {
      "epoch": 2.222135038644121,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.256,
      "step": 7080
    },
    {
      "epoch": 2.2224488995252853,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.1858,
      "step": 7081
    },
    {
      "epoch": 2.2227627604064497,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.2447,
      "step": 7082
    },
    {
      "epoch": 2.223076621287614,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.1893,
      "step": 7083
    },
    {
      "epoch": 2.223390482168779,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.3268,
      "step": 7084
    },
    {
      "epoch": 2.2237043430499432,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.1757,
      "step": 7085
    },
    {
      "epoch": 2.2240182039311076,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.1648,
      "step": 7086
    },
    {
      "epoch": 2.224332064812272,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.3042,
      "step": 7087
    },
    {
      "epoch": 2.2246459256934363,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.261,
      "step": 7088
    },
    {
      "epoch": 2.2249597865746007,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.2222,
      "step": 7089
    },
    {
      "epoch": 2.225273647455765,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.1377,
      "step": 7090
    },
    {
      "epoch": 2.22558750833693,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.1968,
      "step": 7091
    },
    {
      "epoch": 2.2259013692180942,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.0236,
      "step": 7092
    },
    {
      "epoch": 2.2262152300992586,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.0986,
      "step": 7093
    },
    {
      "epoch": 2.226529090980423,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.205,
      "step": 7094
    },
    {
      "epoch": 2.2268429518615873,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.2683,
      "step": 7095
    },
    {
      "epoch": 2.2271568127427517,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.2679,
      "step": 7096
    },
    {
      "epoch": 2.227470673623916,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.156,
      "step": 7097
    },
    {
      "epoch": 2.227784534505081,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.1063,
      "step": 7098
    },
    {
      "epoch": 2.228098395386245,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.1573,
      "step": 7099
    },
    {
      "epoch": 2.2284122562674096,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.1298,
      "step": 7100
    },
    {
      "epoch": 2.228726117148574,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.3207,
      "step": 7101
    },
    {
      "epoch": 2.2290399780297383,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.22,
      "step": 7102
    },
    {
      "epoch": 2.2293538389109027,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.2568,
      "step": 7103
    },
    {
      "epoch": 2.229667699792067,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.303,
      "step": 7104
    },
    {
      "epoch": 2.2299815606732314,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.5303,
      "step": 7105
    },
    {
      "epoch": 2.230295421554396,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.5333,
      "step": 7106
    },
    {
      "epoch": 2.230295421554396,
      "eval_loss": 1.8770476579666138,
      "eval_runtime": 123.3685,
      "eval_samples_per_second": 8.106,
      "eval_steps_per_second": 8.106,
      "step": 7106
    },
    {
      "epoch": 2.230295421554396,
      "mmlu_eval_accuracy": 0.4073725220978079,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.09090909090909091,
      "mmlu_eval_accuracy_conceptual_physics": 0.5,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.21951219512195122,
      "mmlu_eval_accuracy_formal_logic": 0.42857142857142855,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.3333333333333333,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.5333333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.4230769230769231,
      "mmlu_eval_accuracy_human_aging": 0.5652173913043478,
      "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": 0.4444444444444444,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.76,
      "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
      "mmlu_eval_accuracy_miscellaneous": 0.5232558139534884,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.31,
      "mmlu_eval_accuracy_nutrition": 0.36363636363636365,
      "mmlu_eval_accuracy_philosophy": 0.38235294117647056,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
      "mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.2777777777777778,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.329874158294897,
      "step": 7106
    },
    {
      "epoch": 2.2306092824355606,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 1.5482,
      "step": 7107
    },
    {
      "epoch": 2.230923143316725,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002,
      "loss": 1.9437,
      "step": 7108
    },
    {
      "epoch": 2.2312370041978893,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 1.5668,
      "step": 7109
    },
    {
      "epoch": 2.2315508650790536,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 1.4053,
      "step": 7110
    },
    {
      "epoch": 2.231864725960218,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.5723,
      "step": 7111
    },
    {
      "epoch": 2.2321785868413824,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.7933,
      "step": 7112
    },
    {
      "epoch": 2.232492447722547,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.4577,
      "step": 7113
    },
    {
      "epoch": 2.2328063086037115,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.2109,
      "step": 7114
    },
    {
      "epoch": 2.233120169484876,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 1.6556,
      "step": 7115
    },
    {
      "epoch": 2.2334340303660403,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.8344,
      "step": 7116
    },
    {
      "epoch": 2.2337478912472046,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.5766,
      "step": 7117
    },
    {
      "epoch": 2.234061752128369,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.3747,
      "step": 7118
    },
    {
      "epoch": 2.2343756130095334,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.3569,
      "step": 7119
    },
    {
      "epoch": 2.2346894738906977,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 1.3268,
      "step": 7120
    },
    {
      "epoch": 2.2350033347718625,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 1.7115,
      "step": 7121
    },
    {
      "epoch": 2.235317195653027,
      "grad_norm": 0.859375,
      "learning_rate": 0.0002,
      "loss": 2.0211,
      "step": 7122
    },
    {
      "epoch": 2.2356310565341913,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.5885,
      "step": 7123
    },
    {
      "epoch": 2.2359449174153556,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.2717,
      "step": 7124
    },
    {
      "epoch": 2.23625877829652,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.3483,
      "step": 7125
    },
    {
      "epoch": 2.2365726391776843,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.385,
      "step": 7126
    },
    {
      "epoch": 2.2368865000588487,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4571,
      "step": 7127
    },
    {
      "epoch": 2.2372003609400135,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.3417,
      "step": 7128
    },
    {
      "epoch": 2.237514221821178,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.249,
      "step": 7129
    },
    {
      "epoch": 2.2378280827023422,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.2537,
      "step": 7130
    },
    {
      "epoch": 2.2381419435835066,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.2247,
      "step": 7131
    },
    {
      "epoch": 2.238455804464671,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.449,
      "step": 7132
    },
    {
      "epoch": 2.2387696653458353,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.3207,
      "step": 7133
    },
    {
      "epoch": 2.2390835262269997,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.2411,
      "step": 7134
    },
    {
      "epoch": 2.2393973871081645,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.1709,
      "step": 7135
    },
    {
      "epoch": 2.239711247989329,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.1944,
      "step": 7136
    },
    {
      "epoch": 2.240025108870493,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.42,
      "step": 7137
    },
    {
      "epoch": 2.2403389697516576,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.1596,
      "step": 7138
    },
    {
      "epoch": 2.240652830632822,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.3256,
      "step": 7139
    },
    {
      "epoch": 2.2409666915139863,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.066,
      "step": 7140
    },
    {
      "epoch": 2.2412805523951507,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.314,
      "step": 7141
    },
    {
      "epoch": 2.2415944132763155,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.0471,
      "step": 7142
    },
    {
      "epoch": 2.24190827415748,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.1894,
      "step": 7143
    },
    {
      "epoch": 2.242222135038644,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 1.4259,
      "step": 7144
    },
    {
      "epoch": 2.2425359959198086,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.1759,
      "step": 7145
    },
    {
      "epoch": 2.242849856800973,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.0261,
      "step": 7146
    },
    {
      "epoch": 2.2431637176821373,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.4349,
      "step": 7147
    },
    {
      "epoch": 2.2434775785633017,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.2358,
      "step": 7148
    },
    {
      "epoch": 2.2437914394444665,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.3888,
      "step": 7149
    },
    {
      "epoch": 2.244105300325631,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.1028,
      "step": 7150
    },
    {
      "epoch": 2.244419161206795,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.4284,
      "step": 7151
    },
    {
      "epoch": 2.2447330220879596,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.2476,
      "step": 7152
    },
    {
      "epoch": 2.245046882969124,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.7325,
      "step": 7153
    },
    {
      "epoch": 2.2453607438502883,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.2499,
      "step": 7154
    },
    {
      "epoch": 2.2456746047314526,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.5747,
      "step": 7155
    },
    {
      "epoch": 2.245988465612617,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.5954,
      "step": 7156
    },
    {
      "epoch": 2.246302326493782,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002,
      "loss": 1.643,
      "step": 7157
    },
    {
      "epoch": 2.246616187374946,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.609,
      "step": 7158
    },
    {
      "epoch": 2.2469300482561105,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.9222,
      "step": 7159
    },
    {
      "epoch": 2.247243909137275,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.3925,
      "step": 7160
    },
    {
      "epoch": 2.2475577700184393,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 1.4812,
      "step": 7161
    },
    {
      "epoch": 2.2478716308996036,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 1.9347,
      "step": 7162
    },
    {
      "epoch": 2.248185491780768,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.7704,
      "step": 7163
    },
    {
      "epoch": 2.248499352661933,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0002,
      "loss": 1.734,
      "step": 7164
    },
    {
      "epoch": 2.248813213543097,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.2706,
      "step": 7165
    },
    {
      "epoch": 2.2491270744242615,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.4148,
      "step": 7166
    },
    {
      "epoch": 2.249440935305426,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.772,
      "step": 7167
    },
    {
      "epoch": 2.2497547961865902,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.549,
      "step": 7168
    },
    {
      "epoch": 2.2500686570677546,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.2603,
      "step": 7169
    },
    {
      "epoch": 2.250382517948919,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.5354,
      "step": 7170
    },
    {
      "epoch": 2.2506963788300833,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002,
      "loss": 2.1267,
      "step": 7171
    },
    {
      "epoch": 2.251010239711248,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002,
      "loss": 2.3051,
      "step": 7172
    },
    {
      "epoch": 2.2513241005924125,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.5763,
      "step": 7173
    },
    {
      "epoch": 2.251637961473577,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.2335,
      "step": 7174
    },
    {
      "epoch": 2.2519518223547412,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.2123,
      "step": 7175
    },
    {
      "epoch": 2.2522656832359056,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.5065,
      "step": 7176
    },
    {
      "epoch": 2.25257954411707,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.1668,
      "step": 7177
    },
    {
      "epoch": 2.2528934049982343,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.4468,
      "step": 7178
    },
    {
      "epoch": 2.253207265879399,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2751,
      "step": 7179
    },
    {
      "epoch": 2.2535211267605635,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.1354,
      "step": 7180
    },
    {
      "epoch": 2.253834987641728,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.2021,
      "step": 7181
    },
    {
      "epoch": 2.254148848522892,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.0891,
      "step": 7182
    },
    {
      "epoch": 2.2544627094040566,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.2342,
      "step": 7183
    },
    {
      "epoch": 2.254776570285221,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.2501,
      "step": 7184
    },
    {
      "epoch": 2.2550904311663853,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.326,
      "step": 7185
    },
    {
      "epoch": 2.25540429204755,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.2328,
      "step": 7186
    },
    {
      "epoch": 2.2557181529287145,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2954,
      "step": 7187
    },
    {
      "epoch": 2.256032013809879,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.161,
      "step": 7188
    },
    {
      "epoch": 2.256345874691043,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.1598,
      "step": 7189
    },
    {
      "epoch": 2.2566597355722076,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.1464,
      "step": 7190
    },
    {
      "epoch": 2.256973596453372,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.1065,
      "step": 7191
    },
    {
      "epoch": 2.2572874573345363,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.2306,
      "step": 7192
    },
    {
      "epoch": 2.257601318215701,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.1382,
      "step": 7193
    },
    {
      "epoch": 2.2579151790968655,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.1865,
      "step": 7194
    },
    {
      "epoch": 2.25822903997803,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.2209,
      "step": 7195
    },
    {
      "epoch": 2.258542900859194,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.1252,
      "step": 7196
    },
    {
      "epoch": 2.2588567617403585,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.3115,
      "step": 7197
    },
    {
      "epoch": 2.259170622621523,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.0223,
      "step": 7198
    },
    {
      "epoch": 2.2594844835026873,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.2491,
      "step": 7199
    },
    {
      "epoch": 2.259798344383852,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.3421,
      "step": 7200
    },
    {
      "epoch": 2.2601122052650164,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.2766,
      "step": 7201
    },
    {
      "epoch": 2.260426066146181,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.0719,
      "step": 7202
    },
    {
      "epoch": 2.260739927027345,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.1353,
      "step": 7203
    },
    {
      "epoch": 2.2610537879085095,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.3326,
      "step": 7204
    },
    {
      "epoch": 2.261367648789674,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.6516,
      "step": 7205
    },
    {
      "epoch": 2.2616815096708383,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.3945,
      "step": 7206
    },
    {
      "epoch": 2.261995370552003,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.9937,
      "step": 7207
    },
    {
      "epoch": 2.2623092314331674,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.6256,
      "step": 7208
    },
    {
      "epoch": 2.262623092314332,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002,
      "loss": 1.6463,
      "step": 7209
    },
    {
      "epoch": 2.262936953195496,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.5671,
      "step": 7210
    },
    {
      "epoch": 2.2632508140766605,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.4976,
      "step": 7211
    },
    {
      "epoch": 2.263564674957825,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.4574,
      "step": 7212
    },
    {
      "epoch": 2.2638785358389892,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 2.0653,
      "step": 7213
    },
    {
      "epoch": 2.2641923967201536,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 1.6966,
      "step": 7214
    },
    {
      "epoch": 2.2645062576013184,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.6403,
      "step": 7215
    },
    {
      "epoch": 2.2648201184824828,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.7164,
      "step": 7216
    },
    {
      "epoch": 2.265133979363647,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 1.7745,
      "step": 7217
    },
    {
      "epoch": 2.2654478402448115,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.2363,
      "step": 7218
    },
    {
      "epoch": 2.265761701125976,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.526,
      "step": 7219
    },
    {
      "epoch": 2.26607556200714,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 1.691,
      "step": 7220
    },
    {
      "epoch": 2.2663894228883046,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.7359,
      "step": 7221
    },
    {
      "epoch": 2.266703283769469,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002,
      "loss": 2.234,
      "step": 7222
    },
    {
      "epoch": 2.2670171446506338,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.6254,
      "step": 7223
    },
    {
      "epoch": 2.267331005531798,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.2291,
      "step": 7224
    },
    {
      "epoch": 2.2676448664129625,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.3581,
      "step": 7225
    },
    {
      "epoch": 2.267958727294127,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.3394,
      "step": 7226
    },
    {
      "epoch": 2.268272588175291,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.3901,
      "step": 7227
    },
    {
      "epoch": 2.2685864490564556,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.286,
      "step": 7228
    },
    {
      "epoch": 2.26890030993762,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.2478,
      "step": 7229
    },
    {
      "epoch": 2.2692141708187847,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.298,
      "step": 7230
    },
    {
      "epoch": 2.269528031699949,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.1992,
      "step": 7231
    },
    {
      "epoch": 2.2698418925811135,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.1478,
      "step": 7232
    },
    {
      "epoch": 2.270155753462278,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.1495,
      "step": 7233
    },
    {
      "epoch": 2.270469614343442,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.1202,
      "step": 7234
    },
    {
      "epoch": 2.2707834752246066,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.1152,
      "step": 7235
    },
    {
      "epoch": 2.271097336105771,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2451,
      "step": 7236
    },
    {
      "epoch": 2.2714111969869357,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.1696,
      "step": 7237
    },
    {
      "epoch": 2.2717250578681,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.1456,
      "step": 7238
    },
    {
      "epoch": 2.2720389187492644,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.2046,
      "step": 7239
    },
    {
      "epoch": 2.272352779630429,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.0277,
      "step": 7240
    },
    {
      "epoch": 2.272666640511593,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.1189,
      "step": 7241
    },
    {
      "epoch": 2.2729805013927575,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.1998,
      "step": 7242
    },
    {
      "epoch": 2.273294362273922,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.2429,
      "step": 7243
    },
    {
      "epoch": 2.2736082231550867,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.1494,
      "step": 7244
    },
    {
      "epoch": 2.273922084036251,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.1338,
      "step": 7245
    },
    {
      "epoch": 2.2742359449174154,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.2464,
      "step": 7246
    },
    {
      "epoch": 2.27454980579858,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.1089,
      "step": 7247
    },
    {
      "epoch": 2.274863666679744,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.2631,
      "step": 7248
    },
    {
      "epoch": 2.2751775275609085,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.3957,
      "step": 7249
    },
    {
      "epoch": 2.275491388442073,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 1.3219,
      "step": 7250
    },
    {
      "epoch": 2.2758052493232377,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 1.3588,
      "step": 7251
    },
    {
      "epoch": 2.276119110204402,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0002,
      "loss": 1.6856,
      "step": 7252
    },
    {
      "epoch": 2.2764329710855664,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.1168,
      "step": 7253
    },
    {
      "epoch": 2.2767468319667308,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.5774,
      "step": 7254
    },
    {
      "epoch": 2.277060692847895,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.6688,
      "step": 7255
    },
    {
      "epoch": 2.2773745537290595,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.4973,
      "step": 7256
    },
    {
      "epoch": 2.277688414610224,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.688,
      "step": 7257
    },
    {
      "epoch": 2.2780022754913887,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.2582,
      "step": 7258
    },
    {
      "epoch": 2.278316136372553,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.5272,
      "step": 7259
    },
    {
      "epoch": 2.2786299972537174,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.5659,
      "step": 7260
    },
    {
      "epoch": 2.2789438581348818,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.6533,
      "step": 7261
    },
    {
      "epoch": 2.279257719016046,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.6592,
      "step": 7262
    },
    {
      "epoch": 2.2795715798972105,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.5342,
      "step": 7263
    },
    {
      "epoch": 2.279885440778375,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002,
      "loss": 1.795,
      "step": 7264
    },
    {
      "epoch": 2.2801993016595397,
      "grad_norm": 0.84375,
      "learning_rate": 0.0002,
      "loss": 1.6707,
      "step": 7265
    },
    {
      "epoch": 2.280513162540704,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002,
      "loss": 2.1182,
      "step": 7266
    },
    {
      "epoch": 2.2808270234218684,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002,
      "loss": 1.7529,
      "step": 7267
    },
    {
      "epoch": 2.2811408843030327,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0002,
      "loss": 1.5335,
      "step": 7268
    },
    {
      "epoch": 2.281454745184197,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002,
      "loss": 1.4642,
      "step": 7269
    },
    {
      "epoch": 2.2817686060653615,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.5671,
      "step": 7270
    },
    {
      "epoch": 2.282082466946526,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.6643,
      "step": 7271
    },
    {
      "epoch": 2.28239632782769,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.7303,
      "step": 7272
    },
    {
      "epoch": 2.2827101887088546,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.6393,
      "step": 7273
    },
    {
      "epoch": 2.2830240495900194,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.2254,
      "step": 7274
    },
    {
      "epoch": 2.2833379104711837,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.2888,
      "step": 7275
    },
    {
      "epoch": 2.283651771352348,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.3103,
      "step": 7276
    },
    {
      "epoch": 2.2839656322335125,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.2934,
      "step": 7277
    },
    {
      "epoch": 2.284279493114677,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.1373,
      "step": 7278
    },
    {
      "epoch": 2.284593353995841,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.2261,
      "step": 7279
    },
    {
      "epoch": 2.2849072148770055,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.2468,
      "step": 7280
    },
    {
      "epoch": 2.2852210757581704,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.1644,
      "step": 7281
    },
    {
      "epoch": 2.2855349366393347,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.3607,
      "step": 7282
    },
    {
      "epoch": 2.285848797520499,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.1685,
      "step": 7283
    },
    {
      "epoch": 2.2861626584016634,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.2223,
      "step": 7284
    },
    {
      "epoch": 2.286476519282828,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2889,
      "step": 7285
    },
    {
      "epoch": 2.286790380163992,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.276,
      "step": 7286
    },
    {
      "epoch": 2.2871042410451565,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.1342,
      "step": 7287
    },
    {
      "epoch": 2.2874181019263213,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.2093,
      "step": 7288
    },
    {
      "epoch": 2.2877319628074857,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.2935,
      "step": 7289
    },
    {
      "epoch": 2.28804582368865,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.1321,
      "step": 7290
    },
    {
      "epoch": 2.2883596845698144,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.0876,
      "step": 7291
    },
    {
      "epoch": 2.288673545450979,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.1969,
      "step": 7292
    },
    {
      "epoch": 2.288987406332143,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.0424,
      "step": 7293
    },
    {
      "epoch": 2.288987406332143,
      "eval_loss": 1.8512505292892456,
      "eval_runtime": 123.2149,
      "eval_samples_per_second": 8.116,
      "eval_steps_per_second": 8.116,
      "step": 7293
    },
    {
      "epoch": 2.288987406332143,
      "mmlu_eval_accuracy": 0.4048621994273181,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.0,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.18181818181818182,
      "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": 0.21951219512195122,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "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": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.55,
      "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.5,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.16666666666666666,
      "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": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5348837209302325,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.36363636363636365,
      "mmlu_eval_accuracy_philosophy": 0.4117647058823529,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.3888888888888889,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.305676959644806,
      "step": 7293
    },
    {
      "epoch": 2.2893012672133075,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.2267,
      "step": 7294
    },
    {
      "epoch": 2.2896151280944723,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.2125,
      "step": 7295
    },
    {
      "epoch": 2.2899289889756367,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.1225,
      "step": 7296
    },
    {
      "epoch": 2.290242849856801,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 1.2745,
      "step": 7297
    },
    {
      "epoch": 2.2905567107379654,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.113,
      "step": 7298
    },
    {
      "epoch": 2.2908705716191298,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.0614,
      "step": 7299
    },
    {
      "epoch": 2.291184432500294,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.0341,
      "step": 7300
    },
    {
      "epoch": 2.2914982933814585,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.3243,
      "step": 7301
    },
    {
      "epoch": 2.2918121542626233,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.1798,
      "step": 7302
    },
    {
      "epoch": 2.2921260151437877,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.1524,
      "step": 7303
    },
    {
      "epoch": 2.292439876024952,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.4504,
      "step": 7304
    },
    {
      "epoch": 2.2927537369061164,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.362,
      "step": 7305
    },
    {
      "epoch": 2.2930675977872808,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.2939,
      "step": 7306
    },
    {
      "epoch": 2.293381458668445,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002,
      "loss": 1.5712,
      "step": 7307
    },
    {
      "epoch": 2.2936953195496095,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.7939,
      "step": 7308
    },
    {
      "epoch": 2.2940091804307743,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.5016,
      "step": 7309
    },
    {
      "epoch": 2.2943230413119386,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002,
      "loss": 1.8059,
      "step": 7310
    },
    {
      "epoch": 2.294636902193103,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.3105,
      "step": 7311
    },
    {
      "epoch": 2.2949507630742674,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.532,
      "step": 7312
    },
    {
      "epoch": 2.2952646239554317,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.5416,
      "step": 7313
    },
    {
      "epoch": 2.295578484836596,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.8934,
      "step": 7314
    },
    {
      "epoch": 2.2958923457177605,
      "grad_norm": 1.828125,
      "learning_rate": 0.0002,
      "loss": 1.5368,
      "step": 7315
    },
    {
      "epoch": 2.2962062065989253,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.4952,
      "step": 7316
    },
    {
      "epoch": 2.2965200674800896,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.3947,
      "step": 7317
    },
    {
      "epoch": 2.296833928361254,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.5034,
      "step": 7318
    },
    {
      "epoch": 2.2971477892424184,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002,
      "loss": 1.6191,
      "step": 7319
    },
    {
      "epoch": 2.2974616501235827,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.3039,
      "step": 7320
    },
    {
      "epoch": 2.297775511004747,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0002,
      "loss": 1.8442,
      "step": 7321
    },
    {
      "epoch": 2.2980893718859114,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.7986,
      "step": 7322
    },
    {
      "epoch": 2.298403232767076,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.4466,
      "step": 7323
    },
    {
      "epoch": 2.29871709364824,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.3619,
      "step": 7324
    },
    {
      "epoch": 2.299030954529405,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.1874,
      "step": 7325
    },
    {
      "epoch": 2.2993448154105693,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.3146,
      "step": 7326
    },
    {
      "epoch": 2.2996586762917337,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.3598,
      "step": 7327
    },
    {
      "epoch": 2.299972537172898,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.3186,
      "step": 7328
    },
    {
      "epoch": 2.3002863980540624,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.3518,
      "step": 7329
    },
    {
      "epoch": 2.300600258935227,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.376,
      "step": 7330
    },
    {
      "epoch": 2.300914119816391,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.1383,
      "step": 7331
    },
    {
      "epoch": 2.301227980697556,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.2397,
      "step": 7332
    },
    {
      "epoch": 2.3015418415787203,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.1586,
      "step": 7333
    },
    {
      "epoch": 2.3018557024598847,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.3042,
      "step": 7334
    },
    {
      "epoch": 2.302169563341049,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.3113,
      "step": 7335
    },
    {
      "epoch": 2.3024834242222134,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.1487,
      "step": 7336
    },
    {
      "epoch": 2.3027972851033778,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.2513,
      "step": 7337
    },
    {
      "epoch": 2.303111145984542,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2273,
      "step": 7338
    },
    {
      "epoch": 2.303425006865707,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.2175,
      "step": 7339
    },
    {
      "epoch": 2.3037388677468713,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.0618,
      "step": 7340
    },
    {
      "epoch": 2.3040527286280357,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.1736,
      "step": 7341
    },
    {
      "epoch": 2.3043665895092,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.2536,
      "step": 7342
    },
    {
      "epoch": 2.3046804503903644,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.1165,
      "step": 7343
    },
    {
      "epoch": 2.3049943112715288,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.1144,
      "step": 7344
    },
    {
      "epoch": 2.305308172152693,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.2865,
      "step": 7345
    },
    {
      "epoch": 2.305622033033858,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.1935,
      "step": 7346
    },
    {
      "epoch": 2.3059358939150223,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.1487,
      "step": 7347
    },
    {
      "epoch": 2.3062497547961867,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.141,
      "step": 7348
    },
    {
      "epoch": 2.306563615677351,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.2915,
      "step": 7349
    },
    {
      "epoch": 2.3068774765585154,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.1457,
      "step": 7350
    },
    {
      "epoch": 2.3071913374396797,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.0873,
      "step": 7351
    },
    {
      "epoch": 2.307505198320844,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 0.9769,
      "step": 7352
    },
    {
      "epoch": 2.307819059202009,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.2883,
      "step": 7353
    },
    {
      "epoch": 2.3081329200831733,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.3833,
      "step": 7354
    },
    {
      "epoch": 2.3084467809643376,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 1.2168,
      "step": 7355
    },
    {
      "epoch": 2.308760641845502,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.3564,
      "step": 7356
    },
    {
      "epoch": 2.3090745027266664,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.5277,
      "step": 7357
    },
    {
      "epoch": 2.3093883636078307,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.4543,
      "step": 7358
    },
    {
      "epoch": 2.309702224488995,
      "grad_norm": 0.859375,
      "learning_rate": 0.0002,
      "loss": 1.9339,
      "step": 7359
    },
    {
      "epoch": 2.31001608537016,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002,
      "loss": 1.7903,
      "step": 7360
    },
    {
      "epoch": 2.3103299462513243,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.324,
      "step": 7361
    },
    {
      "epoch": 2.3106438071324886,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.5345,
      "step": 7362
    },
    {
      "epoch": 2.310957668013653,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.8501,
      "step": 7363
    },
    {
      "epoch": 2.3112715288948174,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0002,
      "loss": 1.4822,
      "step": 7364
    },
    {
      "epoch": 2.3115853897759817,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.565,
      "step": 7365
    },
    {
      "epoch": 2.311899250657146,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.6208,
      "step": 7366
    },
    {
      "epoch": 2.312213111538311,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.7077,
      "step": 7367
    },
    {
      "epoch": 2.3125269724194752,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.7671,
      "step": 7368
    },
    {
      "epoch": 2.3128408333006396,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 1.5872,
      "step": 7369
    },
    {
      "epoch": 2.313154694181804,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002,
      "loss": 1.4489,
      "step": 7370
    },
    {
      "epoch": 2.3134685550629683,
      "grad_norm": 0.90625,
      "learning_rate": 0.0002,
      "loss": 1.6627,
      "step": 7371
    },
    {
      "epoch": 2.3137824159441327,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002,
      "loss": 2.599,
      "step": 7372
    },
    {
      "epoch": 2.314096276825297,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.4697,
      "step": 7373
    },
    {
      "epoch": 2.3144101377064614,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.2035,
      "step": 7374
    },
    {
      "epoch": 2.314723998587626,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.3164,
      "step": 7375
    },
    {
      "epoch": 2.3150378594687906,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.2702,
      "step": 7376
    },
    {
      "epoch": 2.315351720349955,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.1878,
      "step": 7377
    },
    {
      "epoch": 2.3156655812311193,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.2336,
      "step": 7378
    },
    {
      "epoch": 2.3159794421122837,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.2553,
      "step": 7379
    },
    {
      "epoch": 2.316293302993448,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.1999,
      "step": 7380
    },
    {
      "epoch": 2.3166071638746124,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.1778,
      "step": 7381
    },
    {
      "epoch": 2.3169210247557768,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.0516,
      "step": 7382
    },
    {
      "epoch": 2.3172348856369416,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.1528,
      "step": 7383
    },
    {
      "epoch": 2.317548746518106,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.3735,
      "step": 7384
    },
    {
      "epoch": 2.3178626073992703,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.2022,
      "step": 7385
    },
    {
      "epoch": 2.3181764682804347,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.3049,
      "step": 7386
    },
    {
      "epoch": 2.318490329161599,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.1096,
      "step": 7387
    },
    {
      "epoch": 2.3188041900427634,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.1619,
      "step": 7388
    },
    {
      "epoch": 2.3191180509239278,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.3097,
      "step": 7389
    },
    {
      "epoch": 2.3194319118050926,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.253,
      "step": 7390
    },
    {
      "epoch": 2.319745772686257,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.1523,
      "step": 7391
    },
    {
      "epoch": 2.3200596335674213,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.284,
      "step": 7392
    },
    {
      "epoch": 2.3203734944485856,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.2375,
      "step": 7393
    },
    {
      "epoch": 2.32068735532975,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.095,
      "step": 7394
    },
    {
      "epoch": 2.3210012162109144,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 0.9855,
      "step": 7395
    },
    {
      "epoch": 2.3213150770920787,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.144,
      "step": 7396
    },
    {
      "epoch": 2.3216289379732435,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.1119,
      "step": 7397
    },
    {
      "epoch": 2.321942798854408,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.1326,
      "step": 7398
    },
    {
      "epoch": 2.3222566597355723,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.0936,
      "step": 7399
    },
    {
      "epoch": 2.3225705206167366,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.4238,
      "step": 7400
    },
    {
      "epoch": 2.322884381497901,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.5229,
      "step": 7401
    },
    {
      "epoch": 2.3231982423790654,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.5519,
      "step": 7402
    },
    {
      "epoch": 2.3235121032602297,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.3097,
      "step": 7403
    },
    {
      "epoch": 2.3238259641413945,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.4459,
      "step": 7404
    },
    {
      "epoch": 2.324139825022559,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.7622,
      "step": 7405
    },
    {
      "epoch": 2.3244536859037233,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.5607,
      "step": 7406
    },
    {
      "epoch": 2.3247675467848876,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.6303,
      "step": 7407
    },
    {
      "epoch": 2.325081407666052,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.6072,
      "step": 7408
    },
    {
      "epoch": 2.3253952685472163,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.5817,
      "step": 7409
    },
    {
      "epoch": 2.3257091294283807,
      "grad_norm": 0.75,
      "learning_rate": 0.0002,
      "loss": 1.5388,
      "step": 7410
    },
    {
      "epoch": 2.3260229903095455,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.4529,
      "step": 7411
    },
    {
      "epoch": 2.32633685119071,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.5822,
      "step": 7412
    },
    {
      "epoch": 2.3266507120718742,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 2.0121,
      "step": 7413
    },
    {
      "epoch": 2.3269645729530386,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.5172,
      "step": 7414
    },
    {
      "epoch": 2.327278433834203,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002,
      "loss": 1.7323,
      "step": 7415
    },
    {
      "epoch": 2.3275922947153673,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002,
      "loss": 2.0126,
      "step": 7416
    },
    {
      "epoch": 2.3279061555965317,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 1.4165,
      "step": 7417
    },
    {
      "epoch": 2.3282200164776965,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 1.2381,
      "step": 7418
    },
    {
      "epoch": 2.328533877358861,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.2845,
      "step": 7419
    },
    {
      "epoch": 2.3288477382400252,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.3803,
      "step": 7420
    },
    {
      "epoch": 2.3291615991211896,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.509,
      "step": 7421
    },
    {
      "epoch": 2.329475460002354,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 1.7149,
      "step": 7422
    },
    {
      "epoch": 2.3297893208835183,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.6408,
      "step": 7423
    },
    {
      "epoch": 2.3301031817646827,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.3261,
      "step": 7424
    },
    {
      "epoch": 2.330417042645847,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.2783,
      "step": 7425
    },
    {
      "epoch": 2.330730903527012,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.3528,
      "step": 7426
    },
    {
      "epoch": 2.331044764408176,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.4038,
      "step": 7427
    },
    {
      "epoch": 2.3313586252893406,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.2779,
      "step": 7428
    },
    {
      "epoch": 2.331672486170505,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.5166,
      "step": 7429
    },
    {
      "epoch": 2.3319863470516693,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.1432,
      "step": 7430
    },
    {
      "epoch": 2.3323002079328337,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.2069,
      "step": 7431
    },
    {
      "epoch": 2.332614068813998,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.3675,
      "step": 7432
    },
    {
      "epoch": 2.3329279296951624,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.2625,
      "step": 7433
    },
    {
      "epoch": 2.333241790576327,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.1656,
      "step": 7434
    },
    {
      "epoch": 2.3335556514574916,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.0226,
      "step": 7435
    },
    {
      "epoch": 2.333869512338656,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.1365,
      "step": 7436
    },
    {
      "epoch": 2.3341833732198203,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.1946,
      "step": 7437
    },
    {
      "epoch": 2.3344972341009846,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.247,
      "step": 7438
    },
    {
      "epoch": 2.334811094982149,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.1363,
      "step": 7439
    },
    {
      "epoch": 2.3351249558633134,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.1791,
      "step": 7440
    },
    {
      "epoch": 2.335438816744478,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2446,
      "step": 7441
    },
    {
      "epoch": 2.3357526776256425,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.2902,
      "step": 7442
    },
    {
      "epoch": 2.336066538506807,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.1283,
      "step": 7443
    },
    {
      "epoch": 2.3363803993879713,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.1388,
      "step": 7444
    },
    {
      "epoch": 2.3366942602691356,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.4106,
      "step": 7445
    },
    {
      "epoch": 2.3370081211503,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.0115,
      "step": 7446
    },
    {
      "epoch": 2.3373219820314644,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.1647,
      "step": 7447
    },
    {
      "epoch": 2.337635842912629,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.2104,
      "step": 7448
    },
    {
      "epoch": 2.3379497037937935,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.393,
      "step": 7449
    },
    {
      "epoch": 2.338263564674958,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 1.2629,
      "step": 7450
    },
    {
      "epoch": 2.3385774255561222,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.2744,
      "step": 7451
    },
    {
      "epoch": 2.3388912864372866,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.3664,
      "step": 7452
    },
    {
      "epoch": 2.339205147318451,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.2567,
      "step": 7453
    },
    {
      "epoch": 2.3395190081996153,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.4276,
      "step": 7454
    },
    {
      "epoch": 2.33983286908078,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.4587,
      "step": 7455
    },
    {
      "epoch": 2.3401467299619445,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.423,
      "step": 7456
    },
    {
      "epoch": 2.340460590843109,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.3277,
      "step": 7457
    },
    {
      "epoch": 2.3407744517242732,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.2941,
      "step": 7458
    },
    {
      "epoch": 2.3410883126054376,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.5982,
      "step": 7459
    },
    {
      "epoch": 2.341402173486602,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002,
      "loss": 1.3236,
      "step": 7460
    },
    {
      "epoch": 2.3417160343677663,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.4465,
      "step": 7461
    },
    {
      "epoch": 2.342029895248931,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.789,
      "step": 7462
    },
    {
      "epoch": 2.3423437561300955,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.8546,
      "step": 7463
    },
    {
      "epoch": 2.34265761701126,
      "grad_norm": 0.75,
      "learning_rate": 0.0002,
      "loss": 1.9523,
      "step": 7464
    },
    {
      "epoch": 2.342971477892424,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 1.8784,
      "step": 7465
    },
    {
      "epoch": 2.3432853387735886,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.8553,
      "step": 7466
    },
    {
      "epoch": 2.343599199654753,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.4822,
      "step": 7467
    },
    {
      "epoch": 2.3439130605359173,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.4605,
      "step": 7468
    },
    {
      "epoch": 2.344226921417082,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 1.6513,
      "step": 7469
    },
    {
      "epoch": 2.3445407822982465,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.7872,
      "step": 7470
    },
    {
      "epoch": 2.344854643179411,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002,
      "loss": 1.4599,
      "step": 7471
    },
    {
      "epoch": 2.345168504060575,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 2.2205,
      "step": 7472
    },
    {
      "epoch": 2.3454823649417396,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.5578,
      "step": 7473
    },
    {
      "epoch": 2.345796225822904,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.1971,
      "step": 7474
    },
    {
      "epoch": 2.3461100867040683,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.2586,
      "step": 7475
    },
    {
      "epoch": 2.346423947585233,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.2479,
      "step": 7476
    },
    {
      "epoch": 2.3467378084663975,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.332,
      "step": 7477
    },
    {
      "epoch": 2.347051669347562,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.1617,
      "step": 7478
    },
    {
      "epoch": 2.347365530228726,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.2061,
      "step": 7479
    },
    {
      "epoch": 2.3476793911098905,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.2111,
      "step": 7480
    },
    {
      "epoch": 2.3476793911098905,
      "eval_loss": 1.8935425281524658,
      "eval_runtime": 123.4423,
      "eval_samples_per_second": 8.101,
      "eval_steps_per_second": 8.101,
      "step": 7480
    },
    {
      "epoch": 2.3476793911098905,
      "mmlu_eval_accuracy": 0.3774387542205415,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.36363636363636365,
      "mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
      "mmlu_eval_accuracy_college_biology": 0.1875,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
      "mmlu_eval_accuracy_elementary_mathematics": 0.21951219512195122,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.2,
      "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": 0.3333333333333333,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.38095238095238093,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.5333333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_world_history": 0.34615384615384615,
      "mmlu_eval_accuracy_human_aging": 0.5217391304347826,
      "mmlu_eval_accuracy_human_sexuality": 0.16666666666666666,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.68,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5,
      "mmlu_eval_accuracy_moral_disputes": 0.5789473684210527,
      "mmlu_eval_accuracy_moral_scenarios": 0.32,
      "mmlu_eval_accuracy_nutrition": 0.5151515151515151,
      "mmlu_eval_accuracy_philosophy": 0.35294117647058826,
      "mmlu_eval_accuracy_prehistory": 0.2571428571428571,
      "mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_psychology": 0.3333333333333333,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.2222222222222222,
      "mmlu_eval_accuracy_sociology": 0.4090909090909091,
      "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
      "mmlu_eval_accuracy_virology": 0.2777777777777778,
      "mmlu_eval_accuracy_world_religions": 0.47368421052631576,
      "mmlu_loss": 1.1910356640776802,
      "step": 7480
    },
    {
      "epoch": 2.347993251991055,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.1232,
      "step": 7481
    },
    {
      "epoch": 2.3483071128722193,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.213,
      "step": 7482
    },
    {
      "epoch": 2.3486209737533836,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.1983,
      "step": 7483
    },
    {
      "epoch": 2.348934834634548,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.2784,
      "step": 7484
    },
    {
      "epoch": 2.349248695515713,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.1297,
      "step": 7485
    },
    {
      "epoch": 2.349562556396877,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.1954,
      "step": 7486
    },
    {
      "epoch": 2.3498764172780415,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2602,
      "step": 7487
    },
    {
      "epoch": 2.350190278159206,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.2229,
      "step": 7488
    },
    {
      "epoch": 2.3505041390403703,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.2742,
      "step": 7489
    },
    {
      "epoch": 2.3508179999215346,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.1714,
      "step": 7490
    },
    {
      "epoch": 2.351131860802699,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.079,
      "step": 7491
    },
    {
      "epoch": 2.351445721683864,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.1073,
      "step": 7492
    },
    {
      "epoch": 2.351759582565028,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.121,
      "step": 7493
    },
    {
      "epoch": 2.3520734434461925,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.3863,
      "step": 7494
    },
    {
      "epoch": 2.352387304327357,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.0825,
      "step": 7495
    },
    {
      "epoch": 2.3527011652085212,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.2917,
      "step": 7496
    },
    {
      "epoch": 2.3530150260896856,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.1381,
      "step": 7497
    },
    {
      "epoch": 2.35332888697085,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.4117,
      "step": 7498
    },
    {
      "epoch": 2.3536427478520148,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.2255,
      "step": 7499
    },
    {
      "epoch": 2.353956608733179,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.351,
      "step": 7500
    },
    {
      "epoch": 2.3542704696143435,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.1369,
      "step": 7501
    },
    {
      "epoch": 2.354584330495508,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.3572,
      "step": 7502
    },
    {
      "epoch": 2.3548981913766722,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.4492,
      "step": 7503
    },
    {
      "epoch": 2.3552120522578366,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.4165,
      "step": 7504
    },
    {
      "epoch": 2.355525913139001,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.3938,
      "step": 7505
    },
    {
      "epoch": 2.3558397740201658,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.5443,
      "step": 7506
    },
    {
      "epoch": 2.35615363490133,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.513,
      "step": 7507
    },
    {
      "epoch": 2.3564674957824945,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.4569,
      "step": 7508
    },
    {
      "epoch": 2.356781356663659,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002,
      "loss": 2.0693,
      "step": 7509
    },
    {
      "epoch": 2.357095217544823,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.4543,
      "step": 7510
    },
    {
      "epoch": 2.3574090784259876,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.6564,
      "step": 7511
    },
    {
      "epoch": 2.357722939307152,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 1.4297,
      "step": 7512
    },
    {
      "epoch": 2.3580368001883167,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.2009,
      "step": 7513
    },
    {
      "epoch": 2.358350661069481,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.4381,
      "step": 7514
    },
    {
      "epoch": 2.3586645219506455,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.6546,
      "step": 7515
    },
    {
      "epoch": 2.35897838283181,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 1.5741,
      "step": 7516
    },
    {
      "epoch": 2.359292243712974,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.8156,
      "step": 7517
    },
    {
      "epoch": 2.3596061045941386,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002,
      "loss": 1.6448,
      "step": 7518
    },
    {
      "epoch": 2.359919965475303,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.2672,
      "step": 7519
    },
    {
      "epoch": 2.3602338263564677,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.348,
      "step": 7520
    },
    {
      "epoch": 2.360547687237632,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.6876,
      "step": 7521
    },
    {
      "epoch": 2.3608615481187964,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002,
      "loss": 2.4533,
      "step": 7522
    },
    {
      "epoch": 2.361175408999961,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.5787,
      "step": 7523
    },
    {
      "epoch": 2.361489269881125,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.4307,
      "step": 7524
    },
    {
      "epoch": 2.3618031307622895,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.2291,
      "step": 7525
    },
    {
      "epoch": 2.362116991643454,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.2793,
      "step": 7526
    },
    {
      "epoch": 2.3624308525246187,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.2511,
      "step": 7527
    },
    {
      "epoch": 2.362744713405783,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.2715,
      "step": 7528
    },
    {
      "epoch": 2.3630585742869474,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.2203,
      "step": 7529
    },
    {
      "epoch": 2.363372435168112,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.3121,
      "step": 7530
    },
    {
      "epoch": 2.363686296049276,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2616,
      "step": 7531
    },
    {
      "epoch": 2.3640001569304405,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.1409,
      "step": 7532
    },
    {
      "epoch": 2.364314017811605,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.127,
      "step": 7533
    },
    {
      "epoch": 2.3646278786927692,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2468,
      "step": 7534
    },
    {
      "epoch": 2.3649417395739336,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.2356,
      "step": 7535
    },
    {
      "epoch": 2.3652556004550984,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2608,
      "step": 7536
    },
    {
      "epoch": 2.365569461336263,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.2708,
      "step": 7537
    },
    {
      "epoch": 2.365883322217427,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 0.9963,
      "step": 7538
    },
    {
      "epoch": 2.3661971830985915,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.2078,
      "step": 7539
    },
    {
      "epoch": 2.366511043979756,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.1852,
      "step": 7540
    },
    {
      "epoch": 2.3668249048609202,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.2221,
      "step": 7541
    },
    {
      "epoch": 2.3671387657420846,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.169,
      "step": 7542
    },
    {
      "epoch": 2.3674526266232494,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.0417,
      "step": 7543
    },
    {
      "epoch": 2.3677664875044138,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.2473,
      "step": 7544
    },
    {
      "epoch": 2.368080348385578,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.2241,
      "step": 7545
    },
    {
      "epoch": 2.3683942092667425,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.0858,
      "step": 7546
    },
    {
      "epoch": 2.368708070147907,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.1838,
      "step": 7547
    },
    {
      "epoch": 2.369021931029071,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.2032,
      "step": 7548
    },
    {
      "epoch": 2.3693357919102356,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.2844,
      "step": 7549
    },
    {
      "epoch": 2.3696496527914004,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 1.4577,
      "step": 7550
    },
    {
      "epoch": 2.3699635136725647,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.3346,
      "step": 7551
    },
    {
      "epoch": 2.370277374553729,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.4026,
      "step": 7552
    },
    {
      "epoch": 2.3705912354348935,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.2713,
      "step": 7553
    },
    {
      "epoch": 2.370905096316058,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.3758,
      "step": 7554
    },
    {
      "epoch": 2.371218957197222,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.4112,
      "step": 7555
    },
    {
      "epoch": 2.3715328180783866,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.5466,
      "step": 7556
    },
    {
      "epoch": 2.3718466789595514,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.4977,
      "step": 7557
    },
    {
      "epoch": 2.3721605398407157,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.4294,
      "step": 7558
    },
    {
      "epoch": 2.37247440072188,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.338,
      "step": 7559
    },
    {
      "epoch": 2.3727882616030445,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 1.8784,
      "step": 7560
    },
    {
      "epoch": 2.373102122484209,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.5111,
      "step": 7561
    },
    {
      "epoch": 2.373415983365373,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.8523,
      "step": 7562
    },
    {
      "epoch": 2.3737298442465375,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.5381,
      "step": 7563
    },
    {
      "epoch": 2.3740437051277024,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.4634,
      "step": 7564
    },
    {
      "epoch": 2.3743575660088667,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.4628,
      "step": 7565
    },
    {
      "epoch": 2.374671426890031,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0002,
      "loss": 1.8217,
      "step": 7566
    },
    {
      "epoch": 2.3749852877711954,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.9564,
      "step": 7567
    },
    {
      "epoch": 2.37529914865236,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 1.5601,
      "step": 7568
    },
    {
      "epoch": 2.375613009533524,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.2211,
      "step": 7569
    },
    {
      "epoch": 2.3759268704146885,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 1.2683,
      "step": 7570
    },
    {
      "epoch": 2.3762407312958533,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 1.6888,
      "step": 7571
    },
    {
      "epoch": 2.3765545921770177,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 2.1861,
      "step": 7572
    },
    {
      "epoch": 2.376868453058182,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.6138,
      "step": 7573
    },
    {
      "epoch": 2.3771823139393464,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.489,
      "step": 7574
    },
    {
      "epoch": 2.377496174820511,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.2304,
      "step": 7575
    },
    {
      "epoch": 2.377810035701675,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.4376,
      "step": 7576
    },
    {
      "epoch": 2.3781238965828395,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.3264,
      "step": 7577
    },
    {
      "epoch": 2.3784377574640043,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.3431,
      "step": 7578
    },
    {
      "epoch": 2.3787516183451687,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.3488,
      "step": 7579
    },
    {
      "epoch": 2.379065479226333,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.358,
      "step": 7580
    },
    {
      "epoch": 2.3793793401074974,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.282,
      "step": 7581
    },
    {
      "epoch": 2.3796932009886618,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.1545,
      "step": 7582
    },
    {
      "epoch": 2.380007061869826,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.2513,
      "step": 7583
    },
    {
      "epoch": 2.3803209227509905,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.2063,
      "step": 7584
    },
    {
      "epoch": 2.380634783632155,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.1785,
      "step": 7585
    },
    {
      "epoch": 2.3809486445133192,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.159,
      "step": 7586
    },
    {
      "epoch": 2.381262505394484,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.3261,
      "step": 7587
    },
    {
      "epoch": 2.3815763662756484,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.1649,
      "step": 7588
    },
    {
      "epoch": 2.3818902271568128,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.2118,
      "step": 7589
    },
    {
      "epoch": 2.382204088037977,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.2561,
      "step": 7590
    },
    {
      "epoch": 2.3825179489191415,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.1133,
      "step": 7591
    },
    {
      "epoch": 2.382831809800306,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.3717,
      "step": 7592
    },
    {
      "epoch": 2.38314567068147,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.0774,
      "step": 7593
    },
    {
      "epoch": 2.383459531562635,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.2753,
      "step": 7594
    },
    {
      "epoch": 2.3837733924437994,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.3099,
      "step": 7595
    },
    {
      "epoch": 2.3840872533249637,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.2474,
      "step": 7596
    },
    {
      "epoch": 2.384401114206128,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.1762,
      "step": 7597
    },
    {
      "epoch": 2.3847149750872925,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.0864,
      "step": 7598
    },
    {
      "epoch": 2.385028835968457,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.2909,
      "step": 7599
    },
    {
      "epoch": 2.385342696849621,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.1975,
      "step": 7600
    },
    {
      "epoch": 2.385656557730786,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.1689,
      "step": 7601
    },
    {
      "epoch": 2.3859704186119504,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.4181,
      "step": 7602
    },
    {
      "epoch": 2.3862842794931147,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.4699,
      "step": 7603
    },
    {
      "epoch": 2.386598140374279,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.4413,
      "step": 7604
    },
    {
      "epoch": 2.3869120012554434,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.3046,
      "step": 7605
    },
    {
      "epoch": 2.387225862136608,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.4347,
      "step": 7606
    },
    {
      "epoch": 2.387539723017772,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.5458,
      "step": 7607
    },
    {
      "epoch": 2.387853583898937,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002,
      "loss": 2.0128,
      "step": 7608
    },
    {
      "epoch": 2.3881674447801013,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.4937,
      "step": 7609
    },
    {
      "epoch": 2.3884813056612657,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.6072,
      "step": 7610
    },
    {
      "epoch": 2.38879516654243,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 1.7504,
      "step": 7611
    },
    {
      "epoch": 2.3891090274235944,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.6268,
      "step": 7612
    },
    {
      "epoch": 2.389422888304759,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.8109,
      "step": 7613
    },
    {
      "epoch": 2.389736749185923,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.383,
      "step": 7614
    },
    {
      "epoch": 2.390050610067088,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.8421,
      "step": 7615
    },
    {
      "epoch": 2.3903644709482523,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 1.9172,
      "step": 7616
    },
    {
      "epoch": 2.3906783318294167,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.1838,
      "step": 7617
    },
    {
      "epoch": 2.390992192710581,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.6015,
      "step": 7618
    },
    {
      "epoch": 2.3913060535917454,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 1.7481,
      "step": 7619
    },
    {
      "epoch": 2.39161991447291,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 1.6861,
      "step": 7620
    },
    {
      "epoch": 2.391933775354074,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.4042,
      "step": 7621
    },
    {
      "epoch": 2.392247636235239,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.8288,
      "step": 7622
    },
    {
      "epoch": 2.3925614971164033,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.7302,
      "step": 7623
    },
    {
      "epoch": 2.3928753579975677,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.2854,
      "step": 7624
    },
    {
      "epoch": 2.393189218878732,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.1566,
      "step": 7625
    },
    {
      "epoch": 2.3935030797598964,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.3792,
      "step": 7626
    },
    {
      "epoch": 2.3938169406410608,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.19,
      "step": 7627
    },
    {
      "epoch": 2.394130801522225,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.4167,
      "step": 7628
    },
    {
      "epoch": 2.39444466240339,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.3733,
      "step": 7629
    },
    {
      "epoch": 2.3947585232845543,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.1987,
      "step": 7630
    },
    {
      "epoch": 2.3950723841657187,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.3062,
      "step": 7631
    },
    {
      "epoch": 2.395386245046883,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.2396,
      "step": 7632
    },
    {
      "epoch": 2.3957001059280474,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.2648,
      "step": 7633
    },
    {
      "epoch": 2.3960139668092117,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.2623,
      "step": 7634
    },
    {
      "epoch": 2.396327827690376,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.3733,
      "step": 7635
    },
    {
      "epoch": 2.3966416885715405,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.2008,
      "step": 7636
    },
    {
      "epoch": 2.3969555494527053,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.346,
      "step": 7637
    },
    {
      "epoch": 2.3972694103338696,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.1256,
      "step": 7638
    },
    {
      "epoch": 2.397583271215034,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.0135,
      "step": 7639
    },
    {
      "epoch": 2.3978971320961984,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3414,
      "step": 7640
    },
    {
      "epoch": 2.3982109929773627,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.1485,
      "step": 7641
    },
    {
      "epoch": 2.398524853858527,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.1127,
      "step": 7642
    },
    {
      "epoch": 2.3988387147396915,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.1797,
      "step": 7643
    },
    {
      "epoch": 2.399152575620856,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.1985,
      "step": 7644
    },
    {
      "epoch": 2.3994664365020206,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.046,
      "step": 7645
    },
    {
      "epoch": 2.399780297383185,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.336,
      "step": 7646
    },
    {
      "epoch": 2.4000941582643494,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.2497,
      "step": 7647
    },
    {
      "epoch": 2.4004080191455137,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.3432,
      "step": 7648
    },
    {
      "epoch": 2.400721880026678,
      "grad_norm": 0.359375,
      "learning_rate": 0.0002,
      "loss": 1.17,
      "step": 7649
    },
    {
      "epoch": 2.4010357409078424,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.1773,
      "step": 7650
    },
    {
      "epoch": 2.401349601789007,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.1109,
      "step": 7651
    },
    {
      "epoch": 2.4016634626701716,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.3537,
      "step": 7652
    },
    {
      "epoch": 2.401977323551336,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 1.1591,
      "step": 7653
    },
    {
      "epoch": 2.4022911844325003,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.4013,
      "step": 7654
    },
    {
      "epoch": 2.4026050453136647,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.6239,
      "step": 7655
    },
    {
      "epoch": 2.402918906194829,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.8728,
      "step": 7656
    },
    {
      "epoch": 2.4032327670759934,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.4437,
      "step": 7657
    },
    {
      "epoch": 2.403546627957158,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002,
      "loss": 2.0085,
      "step": 7658
    },
    {
      "epoch": 2.4038604888383226,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 1.9269,
      "step": 7659
    },
    {
      "epoch": 2.404174349719487,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.3327,
      "step": 7660
    },
    {
      "epoch": 2.4044882106006513,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.8981,
      "step": 7661
    },
    {
      "epoch": 2.4048020714818157,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.319,
      "step": 7662
    },
    {
      "epoch": 2.40511593236298,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.4913,
      "step": 7663
    },
    {
      "epoch": 2.4054297932441444,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 1.9475,
      "step": 7664
    },
    {
      "epoch": 2.4057436541253088,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.7267,
      "step": 7665
    },
    {
      "epoch": 2.4060575150064736,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.3683,
      "step": 7666
    },
    {
      "epoch": 2.406371375887638,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002,
      "loss": 1.7277,
      "step": 7667
    },
    {
      "epoch": 2.406371375887638,
      "eval_loss": 1.8807933330535889,
      "eval_runtime": 123.1847,
      "eval_samples_per_second": 8.118,
      "eval_steps_per_second": 8.118,
      "step": 7667
    },
    {
      "epoch": 2.406371375887638,
      "mmlu_eval_accuracy": 0.41309661099491723,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.3125,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.22727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.1951219512195122,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "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": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.65,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.4230769230769231,
      "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
      "mmlu_eval_accuracy_human_sexuality": 0.25,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.4444444444444444,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.8,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5232558139534884,
      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.42424242424242425,
      "mmlu_eval_accuracy_philosophy": 0.4117647058823529,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.3333333333333333,
      "mmlu_eval_accuracy_world_religions": 0.5263157894736842,
      "mmlu_loss": 1.1642392259516956,
      "step": 7667
    },
    {
      "epoch": 2.4066852367688023,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002,
      "loss": 1.3319,
      "step": 7668
    },
    {
      "epoch": 2.4069990976499667,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.5124,
      "step": 7669
    },
    {
      "epoch": 2.407312958531131,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.7205,
      "step": 7670
    },
    {
      "epoch": 2.4076268194122954,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002,
      "loss": 1.8737,
      "step": 7671
    },
    {
      "epoch": 2.4079406802934598,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002,
      "loss": 2.1204,
      "step": 7672
    },
    {
      "epoch": 2.4082545411746246,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.6131,
      "step": 7673
    },
    {
      "epoch": 2.408568402055789,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1128,
      "step": 7674
    },
    {
      "epoch": 2.4088822629369533,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.3572,
      "step": 7675
    },
    {
      "epoch": 2.4091961238181177,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.3329,
      "step": 7676
    },
    {
      "epoch": 2.409509984699282,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.4067,
      "step": 7677
    },
    {
      "epoch": 2.4098238455804464,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.2161,
      "step": 7678
    },
    {
      "epoch": 2.4101377064616107,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.2445,
      "step": 7679
    },
    {
      "epoch": 2.4104515673427755,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.3158,
      "step": 7680
    },
    {
      "epoch": 2.41076542822394,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.3432,
      "step": 7681
    },
    {
      "epoch": 2.4110792891051043,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.1417,
      "step": 7682
    },
    {
      "epoch": 2.4113931499862686,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.3482,
      "step": 7683
    },
    {
      "epoch": 2.411707010867433,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.1931,
      "step": 7684
    },
    {
      "epoch": 2.4120208717485974,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.325,
      "step": 7685
    },
    {
      "epoch": 2.4123347326297617,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.2304,
      "step": 7686
    },
    {
      "epoch": 2.4126485935109265,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.0299,
      "step": 7687
    },
    {
      "epoch": 2.412962454392091,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.1227,
      "step": 7688
    },
    {
      "epoch": 2.4132763152732553,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.0972,
      "step": 7689
    },
    {
      "epoch": 2.4135901761544196,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.1396,
      "step": 7690
    },
    {
      "epoch": 2.413904037035584,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.2557,
      "step": 7691
    },
    {
      "epoch": 2.4142178979167483,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.0896,
      "step": 7692
    },
    {
      "epoch": 2.4145317587979127,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.1399,
      "step": 7693
    },
    {
      "epoch": 2.414845619679077,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.0606,
      "step": 7694
    },
    {
      "epoch": 2.4151594805602414,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.3909,
      "step": 7695
    },
    {
      "epoch": 2.4154733414414062,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.2732,
      "step": 7696
    },
    {
      "epoch": 2.4157872023225706,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.2149,
      "step": 7697
    },
    {
      "epoch": 2.416101063203735,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.2243,
      "step": 7698
    },
    {
      "epoch": 2.4164149240848993,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.3006,
      "step": 7699
    },
    {
      "epoch": 2.4167287849660637,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.1579,
      "step": 7700
    },
    {
      "epoch": 2.417042645847228,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.2826,
      "step": 7701
    },
    {
      "epoch": 2.4173565067283924,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.4205,
      "step": 7702
    },
    {
      "epoch": 2.4176703676095572,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.3593,
      "step": 7703
    },
    {
      "epoch": 2.4179842284907216,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.3245,
      "step": 7704
    },
    {
      "epoch": 2.418298089371886,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.3828,
      "step": 7705
    },
    {
      "epoch": 2.4186119502530503,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.7869,
      "step": 7706
    },
    {
      "epoch": 2.4189258111342147,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.4622,
      "step": 7707
    },
    {
      "epoch": 2.419239672015379,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002,
      "loss": 1.5695,
      "step": 7708
    },
    {
      "epoch": 2.4195535328965434,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.482,
      "step": 7709
    },
    {
      "epoch": 2.419867393777708,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.6119,
      "step": 7710
    },
    {
      "epoch": 2.4201812546588726,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.5564,
      "step": 7711
    },
    {
      "epoch": 2.420495115540037,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 2.0089,
      "step": 7712
    },
    {
      "epoch": 2.4208089764212013,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.567,
      "step": 7713
    },
    {
      "epoch": 2.4211228373023657,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 2.0211,
      "step": 7714
    },
    {
      "epoch": 2.42143669818353,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002,
      "loss": 1.5029,
      "step": 7715
    },
    {
      "epoch": 2.4217505590646944,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 2.0539,
      "step": 7716
    },
    {
      "epoch": 2.422064419945859,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.7136,
      "step": 7717
    },
    {
      "epoch": 2.4223782808270236,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002,
      "loss": 1.4389,
      "step": 7718
    },
    {
      "epoch": 2.422692141708188,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.5873,
      "step": 7719
    },
    {
      "epoch": 2.4230060025893523,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.3931,
      "step": 7720
    },
    {
      "epoch": 2.4233198634705166,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.6665,
      "step": 7721
    },
    {
      "epoch": 2.423633724351681,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 2.0482,
      "step": 7722
    },
    {
      "epoch": 2.4239475852328454,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.5806,
      "step": 7723
    },
    {
      "epoch": 2.42426144611401,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.2527,
      "step": 7724
    },
    {
      "epoch": 2.4245753069951745,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.2892,
      "step": 7725
    },
    {
      "epoch": 2.424889167876339,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002,
      "loss": 1.307,
      "step": 7726
    },
    {
      "epoch": 2.4252030287575033,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.3538,
      "step": 7727
    },
    {
      "epoch": 2.4255168896386676,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.2927,
      "step": 7728
    },
    {
      "epoch": 2.425830750519832,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.2561,
      "step": 7729
    },
    {
      "epoch": 2.4261446114009964,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.3518,
      "step": 7730
    },
    {
      "epoch": 2.426458472282161,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.2576,
      "step": 7731
    },
    {
      "epoch": 2.4267723331633255,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.191,
      "step": 7732
    },
    {
      "epoch": 2.42708619404449,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.2097,
      "step": 7733
    },
    {
      "epoch": 2.4274000549256542,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.36,
      "step": 7734
    },
    {
      "epoch": 2.4277139158068186,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.3429,
      "step": 7735
    },
    {
      "epoch": 2.428027776687983,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.1519,
      "step": 7736
    },
    {
      "epoch": 2.4283416375691473,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.14,
      "step": 7737
    },
    {
      "epoch": 2.428655498450312,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.2666,
      "step": 7738
    },
    {
      "epoch": 2.4289693593314765,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.2836,
      "step": 7739
    },
    {
      "epoch": 2.429283220212641,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.0682,
      "step": 7740
    },
    {
      "epoch": 2.4295970810938052,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.0658,
      "step": 7741
    },
    {
      "epoch": 2.4299109419749696,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.0664,
      "step": 7742
    },
    {
      "epoch": 2.430224802856134,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.202,
      "step": 7743
    },
    {
      "epoch": 2.4305386637372983,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.1887,
      "step": 7744
    },
    {
      "epoch": 2.4308525246184627,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.1607,
      "step": 7745
    },
    {
      "epoch": 2.431166385499627,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.1466,
      "step": 7746
    },
    {
      "epoch": 2.431480246380792,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.2324,
      "step": 7747
    },
    {
      "epoch": 2.431794107261956,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.1999,
      "step": 7748
    },
    {
      "epoch": 2.4321079681431206,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.1594,
      "step": 7749
    },
    {
      "epoch": 2.432421829024285,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.2988,
      "step": 7750
    },
    {
      "epoch": 2.4327356899054493,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.6062,
      "step": 7751
    },
    {
      "epoch": 2.4330495507866137,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.1449,
      "step": 7752
    },
    {
      "epoch": 2.433363411667778,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.6523,
      "step": 7753
    },
    {
      "epoch": 2.433677272548943,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.2602,
      "step": 7754
    },
    {
      "epoch": 2.433991133430107,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.7511,
      "step": 7755
    },
    {
      "epoch": 2.4343049943112716,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.275,
      "step": 7756
    },
    {
      "epoch": 2.434618855192436,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.7591,
      "step": 7757
    },
    {
      "epoch": 2.4349327160736003,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.526,
      "step": 7758
    },
    {
      "epoch": 2.4352465769547647,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.5648,
      "step": 7759
    },
    {
      "epoch": 2.435560437835929,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.6415,
      "step": 7760
    },
    {
      "epoch": 2.435874298717094,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.7678,
      "step": 7761
    },
    {
      "epoch": 2.436188159598258,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.5692,
      "step": 7762
    },
    {
      "epoch": 2.4365020204794225,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.4177,
      "step": 7763
    },
    {
      "epoch": 2.436815881360587,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 1.6958,
      "step": 7764
    },
    {
      "epoch": 2.4371297422417513,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.8524,
      "step": 7765
    },
    {
      "epoch": 2.4374436031229156,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.3297,
      "step": 7766
    },
    {
      "epoch": 2.43775746400408,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002,
      "loss": 1.6228,
      "step": 7767
    },
    {
      "epoch": 2.438071324885245,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002,
      "loss": 1.4021,
      "step": 7768
    },
    {
      "epoch": 2.438385185766409,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.2827,
      "step": 7769
    },
    {
      "epoch": 2.4386990466475735,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002,
      "loss": 1.8537,
      "step": 7770
    },
    {
      "epoch": 2.439012907528738,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.3937,
      "step": 7771
    },
    {
      "epoch": 2.4393267684099023,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002,
      "loss": 2.5302,
      "step": 7772
    },
    {
      "epoch": 2.4396406292910666,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.5574,
      "step": 7773
    },
    {
      "epoch": 2.439954490172231,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.2218,
      "step": 7774
    },
    {
      "epoch": 2.440268351053396,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.1567,
      "step": 7775
    },
    {
      "epoch": 2.44058221193456,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.2398,
      "step": 7776
    },
    {
      "epoch": 2.4408960728157245,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.2381,
      "step": 7777
    },
    {
      "epoch": 2.441209933696889,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.2362,
      "step": 7778
    },
    {
      "epoch": 2.4415237945780532,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.2873,
      "step": 7779
    },
    {
      "epoch": 2.4418376554592176,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.3658,
      "step": 7780
    },
    {
      "epoch": 2.442151516340382,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.2976,
      "step": 7781
    },
    {
      "epoch": 2.4424653772215468,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.1607,
      "step": 7782
    },
    {
      "epoch": 2.442779238102711,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.3697,
      "step": 7783
    },
    {
      "epoch": 2.4430930989838755,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.3822,
      "step": 7784
    },
    {
      "epoch": 2.44340695986504,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.0846,
      "step": 7785
    },
    {
      "epoch": 2.4437208207462042,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.2304,
      "step": 7786
    },
    {
      "epoch": 2.4440346816273686,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.289,
      "step": 7787
    },
    {
      "epoch": 2.444348542508533,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.2578,
      "step": 7788
    },
    {
      "epoch": 2.4446624033896978,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.2021,
      "step": 7789
    },
    {
      "epoch": 2.444976264270862,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.1785,
      "step": 7790
    },
    {
      "epoch": 2.4452901251520265,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.2989,
      "step": 7791
    },
    {
      "epoch": 2.445603986033191,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.1449,
      "step": 7792
    },
    {
      "epoch": 2.445917846914355,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.4418,
      "step": 7793
    },
    {
      "epoch": 2.4462317077955196,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.153,
      "step": 7794
    },
    {
      "epoch": 2.446545568676684,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.2507,
      "step": 7795
    },
    {
      "epoch": 2.4468594295578483,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.1606,
      "step": 7796
    },
    {
      "epoch": 2.4471732904390127,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.2137,
      "step": 7797
    },
    {
      "epoch": 2.4474871513201775,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.2885,
      "step": 7798
    },
    {
      "epoch": 2.447801012201342,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.5443,
      "step": 7799
    },
    {
      "epoch": 2.448114873082506,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.2144,
      "step": 7800
    },
    {
      "epoch": 2.4484287339636706,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.3909,
      "step": 7801
    },
    {
      "epoch": 2.448742594844835,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 1.37,
      "step": 7802
    },
    {
      "epoch": 2.4490564557259993,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.2949,
      "step": 7803
    },
    {
      "epoch": 2.4493703166071636,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.7602,
      "step": 7804
    },
    {
      "epoch": 2.4496841774883285,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.4979,
      "step": 7805
    },
    {
      "epoch": 2.449998038369493,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.8162,
      "step": 7806
    },
    {
      "epoch": 2.450311899250657,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.7862,
      "step": 7807
    },
    {
      "epoch": 2.4506257601318215,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.7346,
      "step": 7808
    },
    {
      "epoch": 2.450939621012986,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002,
      "loss": 1.6698,
      "step": 7809
    },
    {
      "epoch": 2.4512534818941503,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.6021,
      "step": 7810
    },
    {
      "epoch": 2.4515673427753146,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.5084,
      "step": 7811
    },
    {
      "epoch": 2.4518812036564794,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.7023,
      "step": 7812
    },
    {
      "epoch": 2.452195064537644,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.7789,
      "step": 7813
    },
    {
      "epoch": 2.452508925418808,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002,
      "loss": 1.6994,
      "step": 7814
    },
    {
      "epoch": 2.4528227862999725,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.6418,
      "step": 7815
    },
    {
      "epoch": 2.453136647181137,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.5241,
      "step": 7816
    },
    {
      "epoch": 2.4534505080623012,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002,
      "loss": 1.5741,
      "step": 7817
    },
    {
      "epoch": 2.4537643689434656,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.3351,
      "step": 7818
    },
    {
      "epoch": 2.4540782298246304,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.2748,
      "step": 7819
    },
    {
      "epoch": 2.454392090705795,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.4102,
      "step": 7820
    },
    {
      "epoch": 2.454705951586959,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002,
      "loss": 1.4747,
      "step": 7821
    },
    {
      "epoch": 2.4550198124681235,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002,
      "loss": 1.8248,
      "step": 7822
    },
    {
      "epoch": 2.455333673349288,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.6179,
      "step": 7823
    },
    {
      "epoch": 2.4556475342304522,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.2985,
      "step": 7824
    },
    {
      "epoch": 2.4559613951116166,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.3181,
      "step": 7825
    },
    {
      "epoch": 2.4562752559927814,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.4005,
      "step": 7826
    },
    {
      "epoch": 2.4565891168739458,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.236,
      "step": 7827
    },
    {
      "epoch": 2.45690297775511,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.3331,
      "step": 7828
    },
    {
      "epoch": 2.4572168386362745,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.2984,
      "step": 7829
    },
    {
      "epoch": 2.457530699517439,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.2279,
      "step": 7830
    },
    {
      "epoch": 2.457844560398603,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.1096,
      "step": 7831
    },
    {
      "epoch": 2.4581584212797676,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.1899,
      "step": 7832
    },
    {
      "epoch": 2.4584722821609324,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.2368,
      "step": 7833
    },
    {
      "epoch": 2.4587861430420968,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2286,
      "step": 7834
    },
    {
      "epoch": 2.459100003923261,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.228,
      "step": 7835
    },
    {
      "epoch": 2.4594138648044255,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2755,
      "step": 7836
    },
    {
      "epoch": 2.45972772568559,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.2708,
      "step": 7837
    },
    {
      "epoch": 2.460041586566754,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.1981,
      "step": 7838
    },
    {
      "epoch": 2.4603554474479186,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.2338,
      "step": 7839
    },
    {
      "epoch": 2.4606693083290834,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.2709,
      "step": 7840
    },
    {
      "epoch": 2.4609831692102477,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.1224,
      "step": 7841
    },
    {
      "epoch": 2.461297030091412,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.2273,
      "step": 7842
    },
    {
      "epoch": 2.4616108909725765,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.1939,
      "step": 7843
    },
    {
      "epoch": 2.461924751853741,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.2023,
      "step": 7844
    },
    {
      "epoch": 2.462238612734905,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.2888,
      "step": 7845
    },
    {
      "epoch": 2.4625524736160695,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.1506,
      "step": 7846
    },
    {
      "epoch": 2.462866334497234,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.0527,
      "step": 7847
    },
    {
      "epoch": 2.4631801953783987,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.0532,
      "step": 7848
    },
    {
      "epoch": 2.463494056259563,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.1884,
      "step": 7849
    },
    {
      "epoch": 2.4638079171407274,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.3366,
      "step": 7850
    },
    {
      "epoch": 2.464121778021892,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.4498,
      "step": 7851
    },
    {
      "epoch": 2.464435638903056,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.3936,
      "step": 7852
    },
    {
      "epoch": 2.4647494997842205,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.3656,
      "step": 7853
    },
    {
      "epoch": 2.465063360665385,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.3649,
      "step": 7854
    },
    {
      "epoch": 2.465063360665385,
      "eval_loss": 1.8672763109207153,
      "eval_runtime": 123.3878,
      "eval_samples_per_second": 8.105,
      "eval_steps_per_second": 8.105,
      "step": 7854
    },
    {
      "epoch": 2.465063360665385,
      "mmlu_eval_accuracy": 0.4157366111016451,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.3125,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
      "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": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.6,
      "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": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.45454545454545453,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.6,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.4230769230769231,
      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
      "mmlu_eval_accuracy_human_sexuality": 0.16666666666666666,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.4444444444444444,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.84,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5232558139534884,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.42424242424242425,
      "mmlu_eval_accuracy_philosophy": 0.5,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.3333333333333333,
      "mmlu_eval_accuracy_world_religions": 0.5789473684210527,
      "mmlu_loss": 1.2390411817058093,
      "step": 7854
    },
    {
      "epoch": 2.4653772215465493,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.4188,
      "step": 7855
    },
    {
      "epoch": 2.465691082427714,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.7146,
      "step": 7856
    },
    {
      "epoch": 2.4660049433088784,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.5056,
      "step": 7857
    },
    {
      "epoch": 2.466318804190043,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.5195,
      "step": 7858
    },
    {
      "epoch": 2.466632665071207,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0002,
      "loss": 1.5999,
      "step": 7859
    },
    {
      "epoch": 2.4669465259523715,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 1.7737,
      "step": 7860
    },
    {
      "epoch": 2.467260386833536,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.6586,
      "step": 7861
    },
    {
      "epoch": 2.4675742477147002,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.7174,
      "step": 7862
    },
    {
      "epoch": 2.467888108595865,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.7259,
      "step": 7863
    },
    {
      "epoch": 2.4682019694770294,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.2782,
      "step": 7864
    },
    {
      "epoch": 2.4685158303581938,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 2.2778,
      "step": 7865
    },
    {
      "epoch": 2.468829691239358,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.5447,
      "step": 7866
    },
    {
      "epoch": 2.4691435521205225,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.4328,
      "step": 7867
    },
    {
      "epoch": 2.469457413001687,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.4528,
      "step": 7868
    },
    {
      "epoch": 2.4697712738828512,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.7626,
      "step": 7869
    },
    {
      "epoch": 2.470085134764016,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 1.6889,
      "step": 7870
    },
    {
      "epoch": 2.4703989956451804,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002,
      "loss": 2.2578,
      "step": 7871
    },
    {
      "epoch": 2.4707128565263448,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002,
      "loss": 2.0327,
      "step": 7872
    },
    {
      "epoch": 2.471026717407509,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.4924,
      "step": 7873
    },
    {
      "epoch": 2.4713405782886735,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.4465,
      "step": 7874
    },
    {
      "epoch": 2.471654439169838,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.0916,
      "step": 7875
    },
    {
      "epoch": 2.471968300051002,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.3171,
      "step": 7876
    },
    {
      "epoch": 2.472282160932167,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.3244,
      "step": 7877
    },
    {
      "epoch": 2.4725960218133314,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2169,
      "step": 7878
    },
    {
      "epoch": 2.4729098826944957,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.2338,
      "step": 7879
    },
    {
      "epoch": 2.47322374357566,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.212,
      "step": 7880
    },
    {
      "epoch": 2.4735376044568245,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.3354,
      "step": 7881
    },
    {
      "epoch": 2.473851465337989,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2763,
      "step": 7882
    },
    {
      "epoch": 2.474165326219153,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.3363,
      "step": 7883
    },
    {
      "epoch": 2.474479187100318,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.183,
      "step": 7884
    },
    {
      "epoch": 2.4747930479814824,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.2574,
      "step": 7885
    },
    {
      "epoch": 2.4751069088626467,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.0886,
      "step": 7886
    },
    {
      "epoch": 2.475420769743811,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.0826,
      "step": 7887
    },
    {
      "epoch": 2.4757346306249755,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.1898,
      "step": 7888
    },
    {
      "epoch": 2.47604849150614,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.16,
      "step": 7889
    },
    {
      "epoch": 2.476362352387304,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.2202,
      "step": 7890
    },
    {
      "epoch": 2.476676213268469,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.1367,
      "step": 7891
    },
    {
      "epoch": 2.4769900741496333,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.2364,
      "step": 7892
    },
    {
      "epoch": 2.4773039350307977,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.4015,
      "step": 7893
    },
    {
      "epoch": 2.477617795911962,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 1.124,
      "step": 7894
    },
    {
      "epoch": 2.4779316567931264,
      "grad_norm": 0.75,
      "learning_rate": 0.0002,
      "loss": 1.2469,
      "step": 7895
    },
    {
      "epoch": 2.478245517674291,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.113,
      "step": 7896
    },
    {
      "epoch": 2.478559378555455,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.3857,
      "step": 7897
    },
    {
      "epoch": 2.4788732394366195,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.1959,
      "step": 7898
    },
    {
      "epoch": 2.4791871003177843,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.3862,
      "step": 7899
    },
    {
      "epoch": 2.4795009611989487,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 0.9915,
      "step": 7900
    },
    {
      "epoch": 2.479814822080113,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.3092,
      "step": 7901
    },
    {
      "epoch": 2.4801286829612774,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.1203,
      "step": 7902
    },
    {
      "epoch": 2.480442543842442,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 1.2783,
      "step": 7903
    },
    {
      "epoch": 2.480756404723606,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.2813,
      "step": 7904
    },
    {
      "epoch": 2.4810702656047705,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002,
      "loss": 1.6298,
      "step": 7905
    },
    {
      "epoch": 2.481384126485935,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.4701,
      "step": 7906
    },
    {
      "epoch": 2.4816979873670997,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.3129,
      "step": 7907
    },
    {
      "epoch": 2.482011848248264,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.619,
      "step": 7908
    },
    {
      "epoch": 2.4823257091294284,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 1.8144,
      "step": 7909
    },
    {
      "epoch": 2.4826395700105928,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.5604,
      "step": 7910
    },
    {
      "epoch": 2.482953430891757,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.9081,
      "step": 7911
    },
    {
      "epoch": 2.4832672917729215,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.8851,
      "step": 7912
    },
    {
      "epoch": 2.483581152654086,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002,
      "loss": 1.7626,
      "step": 7913
    },
    {
      "epoch": 2.4838950135352507,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.7049,
      "step": 7914
    },
    {
      "epoch": 2.484208874416415,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002,
      "loss": 2.1731,
      "step": 7915
    },
    {
      "epoch": 2.4845227352975794,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.3674,
      "step": 7916
    },
    {
      "epoch": 2.4848365961787438,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002,
      "loss": 1.708,
      "step": 7917
    },
    {
      "epoch": 2.485150457059908,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.2027,
      "step": 7918
    },
    {
      "epoch": 2.4854643179410725,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.5399,
      "step": 7919
    },
    {
      "epoch": 2.485778178822237,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002,
      "loss": 1.9562,
      "step": 7920
    },
    {
      "epoch": 2.4860920397034016,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.6843,
      "step": 7921
    },
    {
      "epoch": 2.486405900584566,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002,
      "loss": 2.5556,
      "step": 7922
    },
    {
      "epoch": 2.4867197614657304,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.6263,
      "step": 7923
    },
    {
      "epoch": 2.4870336223468947,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.1233,
      "step": 7924
    },
    {
      "epoch": 2.487347483228059,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.2968,
      "step": 7925
    },
    {
      "epoch": 2.4876613441092235,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.2912,
      "step": 7926
    },
    {
      "epoch": 2.487975204990388,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.2559,
      "step": 7927
    },
    {
      "epoch": 2.4882890658715526,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.3162,
      "step": 7928
    },
    {
      "epoch": 2.488602926752717,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.3309,
      "step": 7929
    },
    {
      "epoch": 2.4889167876338814,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3008,
      "step": 7930
    },
    {
      "epoch": 2.4892306485150457,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.2109,
      "step": 7931
    },
    {
      "epoch": 2.48954450939621,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.1883,
      "step": 7932
    },
    {
      "epoch": 2.4898583702773744,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.2936,
      "step": 7933
    },
    {
      "epoch": 2.490172231158539,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.142,
      "step": 7934
    },
    {
      "epoch": 2.4904860920397036,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.2195,
      "step": 7935
    },
    {
      "epoch": 2.490799952920868,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.3405,
      "step": 7936
    },
    {
      "epoch": 2.4911138138020323,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.1507,
      "step": 7937
    },
    {
      "epoch": 2.4914276746831967,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.3229,
      "step": 7938
    },
    {
      "epoch": 2.491741535564361,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.188,
      "step": 7939
    },
    {
      "epoch": 2.4920553964455254,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.0513,
      "step": 7940
    },
    {
      "epoch": 2.49236925732669,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.0351,
      "step": 7941
    },
    {
      "epoch": 2.4926831182078546,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.1251,
      "step": 7942
    },
    {
      "epoch": 2.492996979089019,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.161,
      "step": 7943
    },
    {
      "epoch": 2.4933108399701833,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 0.9955,
      "step": 7944
    },
    {
      "epoch": 2.4936247008513477,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.2169,
      "step": 7945
    },
    {
      "epoch": 2.493938561732512,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.2239,
      "step": 7946
    },
    {
      "epoch": 2.4942524226136764,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.2049,
      "step": 7947
    },
    {
      "epoch": 2.4945662834948408,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.3993,
      "step": 7948
    },
    {
      "epoch": 2.4948801443760056,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.3348,
      "step": 7949
    },
    {
      "epoch": 2.49519400525717,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.5381,
      "step": 7950
    },
    {
      "epoch": 2.4955078661383343,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.3664,
      "step": 7951
    },
    {
      "epoch": 2.4958217270194987,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.3104,
      "step": 7952
    },
    {
      "epoch": 2.496135587900663,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 1.3256,
      "step": 7953
    },
    {
      "epoch": 2.4964494487818274,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.3649,
      "step": 7954
    },
    {
      "epoch": 2.4967633096629918,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.7772,
      "step": 7955
    },
    {
      "epoch": 2.497077170544156,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.3112,
      "step": 7956
    },
    {
      "epoch": 2.4973910314253205,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.3217,
      "step": 7957
    },
    {
      "epoch": 2.4977048923064853,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.3897,
      "step": 7958
    },
    {
      "epoch": 2.4980187531876497,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.6571,
      "step": 7959
    },
    {
      "epoch": 2.498332614068814,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.557,
      "step": 7960
    },
    {
      "epoch": 2.4986464749499784,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.8139,
      "step": 7961
    },
    {
      "epoch": 2.4989603358311427,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.8725,
      "step": 7962
    },
    {
      "epoch": 2.499274196712307,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.738,
      "step": 7963
    },
    {
      "epoch": 2.4995880575934715,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002,
      "loss": 2.0403,
      "step": 7964
    },
    {
      "epoch": 2.4999019184746363,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 1.8007,
      "step": 7965
    },
    {
      "epoch": 2.5002157793558006,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002,
      "loss": 1.7287,
      "step": 7966
    },
    {
      "epoch": 2.500529640236965,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0002,
      "loss": 1.4855,
      "step": 7967
    },
    {
      "epoch": 2.5008435011181294,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 1.5514,
      "step": 7968
    },
    {
      "epoch": 2.5011573619992937,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002,
      "loss": 1.3764,
      "step": 7969
    },
    {
      "epoch": 2.501471222880458,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.3529,
      "step": 7970
    },
    {
      "epoch": 2.5017850837616225,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002,
      "loss": 1.372,
      "step": 7971
    },
    {
      "epoch": 2.5020989446427873,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.7531,
      "step": 7972
    },
    {
      "epoch": 2.5024128055239516,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.4994,
      "step": 7973
    },
    {
      "epoch": 2.502726666405116,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.3102,
      "step": 7974
    },
    {
      "epoch": 2.5030405272862803,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.3701,
      "step": 7975
    },
    {
      "epoch": 2.5033543881674447,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.2651,
      "step": 7976
    },
    {
      "epoch": 2.503668249048609,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.1956,
      "step": 7977
    },
    {
      "epoch": 2.5039821099297734,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.3799,
      "step": 7978
    },
    {
      "epoch": 2.5042959708109382,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.1531,
      "step": 7979
    },
    {
      "epoch": 2.5046098316921026,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.1886,
      "step": 7980
    },
    {
      "epoch": 2.504923692573267,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.3328,
      "step": 7981
    },
    {
      "epoch": 2.5052375534544313,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.305,
      "step": 7982
    },
    {
      "epoch": 2.5055514143355957,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.251,
      "step": 7983
    },
    {
      "epoch": 2.50586527521676,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2927,
      "step": 7984
    },
    {
      "epoch": 2.5061791360979244,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.2262,
      "step": 7985
    },
    {
      "epoch": 2.5064929969790892,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2741,
      "step": 7986
    },
    {
      "epoch": 2.5068068578602536,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.1327,
      "step": 7987
    },
    {
      "epoch": 2.507120718741418,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.3341,
      "step": 7988
    },
    {
      "epoch": 2.5074345796225823,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.2381,
      "step": 7989
    },
    {
      "epoch": 2.5077484405037467,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.2776,
      "step": 7990
    },
    {
      "epoch": 2.508062301384911,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.1254,
      "step": 7991
    },
    {
      "epoch": 2.5083761622660754,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.1396,
      "step": 7992
    },
    {
      "epoch": 2.50869002314724,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.1002,
      "step": 7993
    },
    {
      "epoch": 2.5090038840284046,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.1628,
      "step": 7994
    },
    {
      "epoch": 2.509317744909569,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.3025,
      "step": 7995
    },
    {
      "epoch": 2.5096316057907333,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.107,
      "step": 7996
    },
    {
      "epoch": 2.5099454666718977,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.1702,
      "step": 7997
    },
    {
      "epoch": 2.510259327553062,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.212,
      "step": 7998
    },
    {
      "epoch": 2.5105731884342264,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.5089,
      "step": 7999
    },
    {
      "epoch": 2.510887049315391,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.3374,
      "step": 8000
    },
    {
      "epoch": 2.511200910196555,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.3548,
      "step": 8001
    },
    {
      "epoch": 2.51151477107772,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.3372,
      "step": 8002
    },
    {
      "epoch": 2.5118286319588843,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.3993,
      "step": 8003
    },
    {
      "epoch": 2.5121424928400486,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.4462,
      "step": 8004
    },
    {
      "epoch": 2.512456353721213,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.4214,
      "step": 8005
    },
    {
      "epoch": 2.5127702146023774,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.4534,
      "step": 8006
    },
    {
      "epoch": 2.513084075483542,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.4045,
      "step": 8007
    },
    {
      "epoch": 2.513397936364706,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.4717,
      "step": 8008
    },
    {
      "epoch": 2.513711797245871,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.6533,
      "step": 8009
    },
    {
      "epoch": 2.5140256581270353,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.5562,
      "step": 8010
    },
    {
      "epoch": 2.5143395190081996,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002,
      "loss": 1.9506,
      "step": 8011
    },
    {
      "epoch": 2.514653379889364,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.6297,
      "step": 8012
    },
    {
      "epoch": 2.5149672407705284,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 2.058,
      "step": 8013
    },
    {
      "epoch": 2.515281101651693,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.6586,
      "step": 8014
    },
    {
      "epoch": 2.515594962532857,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.2986,
      "step": 8015
    },
    {
      "epoch": 2.515908823414022,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002,
      "loss": 1.9196,
      "step": 8016
    },
    {
      "epoch": 2.5162226842951863,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.5324,
      "step": 8017
    },
    {
      "epoch": 2.5165365451763506,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002,
      "loss": 1.6906,
      "step": 8018
    },
    {
      "epoch": 2.516850406057515,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 1.4763,
      "step": 8019
    },
    {
      "epoch": 2.5171642669386793,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0002,
      "loss": 1.6944,
      "step": 8020
    },
    {
      "epoch": 2.5174781278198437,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.6834,
      "step": 8021
    },
    {
      "epoch": 2.517791988701008,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0002,
      "loss": 2.3512,
      "step": 8022
    },
    {
      "epoch": 2.518105849582173,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.689,
      "step": 8023
    },
    {
      "epoch": 2.5184197104633372,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.2366,
      "step": 8024
    },
    {
      "epoch": 2.5187335713445016,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.2755,
      "step": 8025
    },
    {
      "epoch": 2.519047432225666,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.201,
      "step": 8026
    },
    {
      "epoch": 2.5193612931068303,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.3067,
      "step": 8027
    },
    {
      "epoch": 2.5196751539879947,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.2367,
      "step": 8028
    },
    {
      "epoch": 2.519989014869159,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.3592,
      "step": 8029
    },
    {
      "epoch": 2.520302875750324,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.3156,
      "step": 8030
    },
    {
      "epoch": 2.520616736631488,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.3446,
      "step": 8031
    },
    {
      "epoch": 2.5209305975126526,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.1283,
      "step": 8032
    },
    {
      "epoch": 2.521244458393817,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.3188,
      "step": 8033
    },
    {
      "epoch": 2.5215583192749813,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.1979,
      "step": 8034
    },
    {
      "epoch": 2.5218721801561457,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.2261,
      "step": 8035
    },
    {
      "epoch": 2.52218604103731,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.1235,
      "step": 8036
    },
    {
      "epoch": 2.522499901918475,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.142,
      "step": 8037
    },
    {
      "epoch": 2.522813762799639,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.201,
      "step": 8038
    },
    {
      "epoch": 2.5231276236808036,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.136,
      "step": 8039
    },
    {
      "epoch": 2.523441484561968,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.2224,
      "step": 8040
    },
    {
      "epoch": 2.5237553454431323,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.2604,
      "step": 8041
    },
    {
      "epoch": 2.5237553454431323,
      "eval_loss": 1.8607443571090698,
      "eval_runtime": 123.55,
      "eval_samples_per_second": 8.094,
      "eval_steps_per_second": 8.094,
      "step": 8041
    },
    {
      "epoch": 2.5237553454431323,
      "mmlu_eval_accuracy": 0.4183529141967123,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.4,
      "mmlu_eval_accuracy_high_school_biology": 0.34375,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.42857142857142855,
      "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": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.6333333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.08695652173913043,
      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_world_history": 0.4230769230769231,
      "mmlu_eval_accuracy_human_aging": 0.5652173913043478,
      "mmlu_eval_accuracy_human_sexuality": 0.16666666666666666,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
      "mmlu_eval_accuracy_miscellaneous": 0.5232558139534884,
      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.3939393939393939,
      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.3333333333333333,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.2777777777777778,
      "mmlu_eval_accuracy_world_religions": 0.5789473684210527,
      "mmlu_loss": 1.2723540784213216,
      "step": 8041
    },
    {
      "epoch": 2.5240692063242967,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.2616,
      "step": 8042
    },
    {
      "epoch": 2.524383067205461,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.0516,
      "step": 8043
    },
    {
      "epoch": 2.524696928086626,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.2961,
      "step": 8044
    },
    {
      "epoch": 2.52501078896779,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.2861,
      "step": 8045
    },
    {
      "epoch": 2.5253246498489546,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.0504,
      "step": 8046
    },
    {
      "epoch": 2.525638510730119,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.3087,
      "step": 8047
    },
    {
      "epoch": 2.5259523716112833,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.1018,
      "step": 8048
    },
    {
      "epoch": 2.5262662324924476,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.4948,
      "step": 8049
    },
    {
      "epoch": 2.526580093373612,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.0693,
      "step": 8050
    },
    {
      "epoch": 2.526893954254777,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002,
      "loss": 1.274,
      "step": 8051
    },
    {
      "epoch": 2.5272078151359407,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.4565,
      "step": 8052
    },
    {
      "epoch": 2.5275216760171055,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.5235,
      "step": 8053
    },
    {
      "epoch": 2.52783553689827,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.5113,
      "step": 8054
    },
    {
      "epoch": 2.5281493977794343,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.4896,
      "step": 8055
    },
    {
      "epoch": 2.5284632586605986,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.741,
      "step": 8056
    },
    {
      "epoch": 2.528777119541763,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.6264,
      "step": 8057
    },
    {
      "epoch": 2.529090980422928,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.538,
      "step": 8058
    },
    {
      "epoch": 2.5294048413040917,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.7576,
      "step": 8059
    },
    {
      "epoch": 2.5297187021852565,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.4019,
      "step": 8060
    },
    {
      "epoch": 2.530032563066421,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.5436,
      "step": 8061
    },
    {
      "epoch": 2.5303464239475852,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 1.9869,
      "step": 8062
    },
    {
      "epoch": 2.5306602848287496,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.6505,
      "step": 8063
    },
    {
      "epoch": 2.530974145709914,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.7756,
      "step": 8064
    },
    {
      "epoch": 2.5312880065910788,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.7569,
      "step": 8065
    },
    {
      "epoch": 2.5316018674722427,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.69,
      "step": 8066
    },
    {
      "epoch": 2.5319157283534075,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.3747,
      "step": 8067
    },
    {
      "epoch": 2.532229589234572,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.4028,
      "step": 8068
    },
    {
      "epoch": 2.5325434501157362,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.5025,
      "step": 8069
    },
    {
      "epoch": 2.5328573109969006,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002,
      "loss": 1.563,
      "step": 8070
    },
    {
      "epoch": 2.533171171878065,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 1.8703,
      "step": 8071
    },
    {
      "epoch": 2.5334850327592293,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0002,
      "loss": 2.1131,
      "step": 8072
    },
    {
      "epoch": 2.5337988936403937,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.5661,
      "step": 8073
    },
    {
      "epoch": 2.5341127545215585,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.3115,
      "step": 8074
    },
    {
      "epoch": 2.534426615402723,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.2665,
      "step": 8075
    },
    {
      "epoch": 2.534740476283887,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.4033,
      "step": 8076
    },
    {
      "epoch": 2.5350543371650516,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.2886,
      "step": 8077
    },
    {
      "epoch": 2.535368198046216,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.1756,
      "step": 8078
    },
    {
      "epoch": 2.5356820589273803,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.2742,
      "step": 8079
    },
    {
      "epoch": 2.5359959198085447,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.1791,
      "step": 8080
    },
    {
      "epoch": 2.5363097806897095,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.2411,
      "step": 8081
    },
    {
      "epoch": 2.536623641570874,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.3197,
      "step": 8082
    },
    {
      "epoch": 2.536937502452038,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2508,
      "step": 8083
    },
    {
      "epoch": 2.5372513633332026,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.2557,
      "step": 8084
    },
    {
      "epoch": 2.537565224214367,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.2187,
      "step": 8085
    },
    {
      "epoch": 2.5378790850955313,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.2556,
      "step": 8086
    },
    {
      "epoch": 2.5381929459766956,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.3464,
      "step": 8087
    },
    {
      "epoch": 2.5385068068578605,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.156,
      "step": 8088
    },
    {
      "epoch": 2.538820667739025,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.19,
      "step": 8089
    },
    {
      "epoch": 2.539134528620189,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.2837,
      "step": 8090
    },
    {
      "epoch": 2.5394483895013535,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.187,
      "step": 8091
    },
    {
      "epoch": 2.539762250382518,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.2087,
      "step": 8092
    },
    {
      "epoch": 2.5400761112636823,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.1529,
      "step": 8093
    },
    {
      "epoch": 2.5403899721448466,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.2803,
      "step": 8094
    },
    {
      "epoch": 2.5407038330260114,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.2249,
      "step": 8095
    },
    {
      "epoch": 2.541017693907176,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.2193,
      "step": 8096
    },
    {
      "epoch": 2.54133155478834,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.2398,
      "step": 8097
    },
    {
      "epoch": 2.5416454156695045,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.3342,
      "step": 8098
    },
    {
      "epoch": 2.541959276550669,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 1.3602,
      "step": 8099
    },
    {
      "epoch": 2.5422731374318333,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.2073,
      "step": 8100
    },
    {
      "epoch": 2.5425869983129976,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 1.1154,
      "step": 8101
    },
    {
      "epoch": 2.5429008591941624,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.3494,
      "step": 8102
    },
    {
      "epoch": 2.5432147200753263,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.5837,
      "step": 8103
    },
    {
      "epoch": 2.543528580956491,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.5074,
      "step": 8104
    },
    {
      "epoch": 2.5438424418376555,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.4006,
      "step": 8105
    },
    {
      "epoch": 2.54415630271882,
      "grad_norm": 0.75,
      "learning_rate": 0.0002,
      "loss": 1.8382,
      "step": 8106
    },
    {
      "epoch": 2.5444701635999842,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.9058,
      "step": 8107
    },
    {
      "epoch": 2.5447840244811486,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.4466,
      "step": 8108
    },
    {
      "epoch": 2.5450978853623134,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.5384,
      "step": 8109
    },
    {
      "epoch": 2.5454117462434773,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002,
      "loss": 1.7193,
      "step": 8110
    },
    {
      "epoch": 2.545725607124642,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 1.823,
      "step": 8111
    },
    {
      "epoch": 2.5460394680058065,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.5359,
      "step": 8112
    },
    {
      "epoch": 2.546353328886971,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002,
      "loss": 1.5397,
      "step": 8113
    },
    {
      "epoch": 2.546667189768135,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.6799,
      "step": 8114
    },
    {
      "epoch": 2.5469810506492996,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 1.7347,
      "step": 8115
    },
    {
      "epoch": 2.5472949115304644,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 1.5677,
      "step": 8116
    },
    {
      "epoch": 2.5476087724116283,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.8451,
      "step": 8117
    },
    {
      "epoch": 2.547922633292793,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.8102,
      "step": 8118
    },
    {
      "epoch": 2.5482364941739575,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.5335,
      "step": 8119
    },
    {
      "epoch": 2.548550355055122,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.5471,
      "step": 8120
    },
    {
      "epoch": 2.548864215936286,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 1.6972,
      "step": 8121
    },
    {
      "epoch": 2.5491780768174506,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002,
      "loss": 2.1062,
      "step": 8122
    },
    {
      "epoch": 2.549491937698615,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.6585,
      "step": 8123
    },
    {
      "epoch": 2.5498057985797793,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.3591,
      "step": 8124
    },
    {
      "epoch": 2.550119659460944,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.4368,
      "step": 8125
    },
    {
      "epoch": 2.5504335203421085,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.3583,
      "step": 8126
    },
    {
      "epoch": 2.550747381223273,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.4122,
      "step": 8127
    },
    {
      "epoch": 2.551061242104437,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.3568,
      "step": 8128
    },
    {
      "epoch": 2.5513751029856016,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2018,
      "step": 8129
    },
    {
      "epoch": 2.551688963866766,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.2514,
      "step": 8130
    },
    {
      "epoch": 2.5520028247479303,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.1808,
      "step": 8131
    },
    {
      "epoch": 2.552316685629095,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.2321,
      "step": 8132
    },
    {
      "epoch": 2.5526305465102594,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.1657,
      "step": 8133
    },
    {
      "epoch": 2.552944407391424,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.3385,
      "step": 8134
    },
    {
      "epoch": 2.553258268272588,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.2294,
      "step": 8135
    },
    {
      "epoch": 2.5535721291537525,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.2754,
      "step": 8136
    },
    {
      "epoch": 2.553885990034917,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2386,
      "step": 8137
    },
    {
      "epoch": 2.5541998509160813,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.0577,
      "step": 8138
    },
    {
      "epoch": 2.554513711797246,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.2231,
      "step": 8139
    },
    {
      "epoch": 2.5548275726784104,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.1979,
      "step": 8140
    },
    {
      "epoch": 2.555141433559575,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.1414,
      "step": 8141
    },
    {
      "epoch": 2.555455294440739,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.1689,
      "step": 8142
    },
    {
      "epoch": 2.5557691553219035,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.1552,
      "step": 8143
    },
    {
      "epoch": 2.556083016203068,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.1334,
      "step": 8144
    },
    {
      "epoch": 2.5563968770842322,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.16,
      "step": 8145
    },
    {
      "epoch": 2.556710737965397,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.3493,
      "step": 8146
    },
    {
      "epoch": 2.5570245988465614,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.1908,
      "step": 8147
    },
    {
      "epoch": 2.5573384597277258,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.2483,
      "step": 8148
    },
    {
      "epoch": 2.55765232060889,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.5053,
      "step": 8149
    },
    {
      "epoch": 2.5579661814900545,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 0.9777,
      "step": 8150
    },
    {
      "epoch": 2.558280042371219,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.3726,
      "step": 8151
    },
    {
      "epoch": 2.5585939032523832,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 1.4535,
      "step": 8152
    },
    {
      "epoch": 2.558907764133548,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.4632,
      "step": 8153
    },
    {
      "epoch": 2.559221625014712,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.5399,
      "step": 8154
    },
    {
      "epoch": 2.5595354858958768,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 1.3817,
      "step": 8155
    },
    {
      "epoch": 2.559849346777041,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.6101,
      "step": 8156
    },
    {
      "epoch": 2.5601632076582055,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002,
      "loss": 1.5083,
      "step": 8157
    },
    {
      "epoch": 2.56047706853937,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.4321,
      "step": 8158
    },
    {
      "epoch": 2.560790929420534,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.9173,
      "step": 8159
    },
    {
      "epoch": 2.561104790301699,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002,
      "loss": 1.8915,
      "step": 8160
    },
    {
      "epoch": 2.561418651182863,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 1.6727,
      "step": 8161
    },
    {
      "epoch": 2.5617325120640277,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0002,
      "loss": 1.7423,
      "step": 8162
    },
    {
      "epoch": 2.562046372945192,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.8554,
      "step": 8163
    },
    {
      "epoch": 2.5623602338263565,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.7873,
      "step": 8164
    },
    {
      "epoch": 2.562674094707521,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.4724,
      "step": 8165
    },
    {
      "epoch": 2.562987955588685,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.6006,
      "step": 8166
    },
    {
      "epoch": 2.56330181646985,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.5194,
      "step": 8167
    },
    {
      "epoch": 2.563615677351014,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 1.6557,
      "step": 8168
    },
    {
      "epoch": 2.5639295382321787,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.5988,
      "step": 8169
    },
    {
      "epoch": 2.564243399113343,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.6732,
      "step": 8170
    },
    {
      "epoch": 2.5645572599945075,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002,
      "loss": 1.971,
      "step": 8171
    },
    {
      "epoch": 2.564871120875672,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.534,
      "step": 8172
    },
    {
      "epoch": 2.565184981756836,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.612,
      "step": 8173
    },
    {
      "epoch": 2.5654988426380005,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3118,
      "step": 8174
    },
    {
      "epoch": 2.565812703519165,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.186,
      "step": 8175
    },
    {
      "epoch": 2.5661265644003297,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.2368,
      "step": 8176
    },
    {
      "epoch": 2.566440425281494,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.2674,
      "step": 8177
    },
    {
      "epoch": 2.5667542861626584,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.1439,
      "step": 8178
    },
    {
      "epoch": 2.567068147043823,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.274,
      "step": 8179
    },
    {
      "epoch": 2.567382007924987,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.1665,
      "step": 8180
    },
    {
      "epoch": 2.5676958688061515,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.3386,
      "step": 8181
    },
    {
      "epoch": 2.568009729687316,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.1246,
      "step": 8182
    },
    {
      "epoch": 2.5683235905684807,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.2216,
      "step": 8183
    },
    {
      "epoch": 2.568637451449645,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.3027,
      "step": 8184
    },
    {
      "epoch": 2.5689513123308094,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.2367,
      "step": 8185
    },
    {
      "epoch": 2.569265173211974,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.3236,
      "step": 8186
    },
    {
      "epoch": 2.569579034093138,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.2682,
      "step": 8187
    },
    {
      "epoch": 2.5698928949743025,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.2099,
      "step": 8188
    },
    {
      "epoch": 2.570206755855467,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.1102,
      "step": 8189
    },
    {
      "epoch": 2.5705206167366317,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.3295,
      "step": 8190
    },
    {
      "epoch": 2.570834477617796,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.2174,
      "step": 8191
    },
    {
      "epoch": 2.5711483384989604,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.3828,
      "step": 8192
    },
    {
      "epoch": 2.5714621993801248,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.1708,
      "step": 8193
    },
    {
      "epoch": 2.571776060261289,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.0845,
      "step": 8194
    },
    {
      "epoch": 2.5720899211424535,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.2666,
      "step": 8195
    },
    {
      "epoch": 2.572403782023618,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.0794,
      "step": 8196
    },
    {
      "epoch": 2.5727176429047827,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.0743,
      "step": 8197
    },
    {
      "epoch": 2.573031503785947,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.1639,
      "step": 8198
    },
    {
      "epoch": 2.5733453646671114,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.3559,
      "step": 8199
    },
    {
      "epoch": 2.5736592255482758,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.1107,
      "step": 8200
    },
    {
      "epoch": 2.57397308642944,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.4383,
      "step": 8201
    },
    {
      "epoch": 2.5742869473106045,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.4989,
      "step": 8202
    },
    {
      "epoch": 2.574600808191769,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.4513,
      "step": 8203
    },
    {
      "epoch": 2.5749146690729336,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.3357,
      "step": 8204
    },
    {
      "epoch": 2.575228529954098,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.5024,
      "step": 8205
    },
    {
      "epoch": 2.5755423908352624,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.3976,
      "step": 8206
    },
    {
      "epoch": 2.5758562517164267,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.3133,
      "step": 8207
    },
    {
      "epoch": 2.576170112597591,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.3584,
      "step": 8208
    },
    {
      "epoch": 2.5764839734787555,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.5127,
      "step": 8209
    },
    {
      "epoch": 2.57679783435992,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 1.5952,
      "step": 8210
    },
    {
      "epoch": 2.5771116952410846,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 1.488,
      "step": 8211
    },
    {
      "epoch": 2.5774255561222486,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.472,
      "step": 8212
    },
    {
      "epoch": 2.5777394170034134,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002,
      "loss": 1.7622,
      "step": 8213
    },
    {
      "epoch": 2.5780532778845777,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.877,
      "step": 8214
    },
    {
      "epoch": 2.578367138765742,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.4925,
      "step": 8215
    },
    {
      "epoch": 2.5786809996469064,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.4816,
      "step": 8216
    },
    {
      "epoch": 2.578994860528071,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 1.6054,
      "step": 8217
    },
    {
      "epoch": 2.5793087214092356,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.3514,
      "step": 8218
    },
    {
      "epoch": 2.5796225822903995,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.2207,
      "step": 8219
    },
    {
      "epoch": 2.5799364431715643,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.5068,
      "step": 8220
    },
    {
      "epoch": 2.5802503040527287,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.7892,
      "step": 8221
    },
    {
      "epoch": 2.580564164933893,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0002,
      "loss": 2.6314,
      "step": 8222
    },
    {
      "epoch": 2.5808780258150574,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.6758,
      "step": 8223
    },
    {
      "epoch": 2.581191886696222,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.4171,
      "step": 8224
    },
    {
      "epoch": 2.5815057475773866,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.2907,
      "step": 8225
    },
    {
      "epoch": 2.5818196084585505,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.2336,
      "step": 8226
    },
    {
      "epoch": 2.5821334693397153,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2096,
      "step": 8227
    },
    {
      "epoch": 2.5824473302208797,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.2043,
      "step": 8228
    },
    {
      "epoch": 2.5824473302208797,
      "eval_loss": 1.9074493646621704,
      "eval_runtime": 123.3494,
      "eval_samples_per_second": 8.107,
      "eval_steps_per_second": 8.107,
      "step": 8228
    },
    {
      "epoch": 2.5824473302208797,
      "mmlu_eval_accuracy": 0.41454449161637824,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.36363636363636365,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.0,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.2692307692307692,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.21951219512195122,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.3125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5116279069767442,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
      "mmlu_eval_accuracy_high_school_physics": 0.47058823529411764,
      "mmlu_eval_accuracy_high_school_psychology": 0.65,
      "mmlu_eval_accuracy_high_school_statistics": 0.13043478260869565,
      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
      "mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
      "mmlu_eval_accuracy_human_sexuality": 0.25,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.68,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5,
      "mmlu_eval_accuracy_moral_disputes": 0.5,
      "mmlu_eval_accuracy_moral_scenarios": 0.29,
      "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.4492753623188406,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.2777777777777778,
      "mmlu_eval_accuracy_world_religions": 0.5789473684210527,
      "mmlu_loss": 1.1899194760426908,
      "step": 8228
    },
    {
      "epoch": 2.582761191102044,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.2567,
      "step": 8229
    },
    {
      "epoch": 2.5830750519832084,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.2747,
      "step": 8230
    },
    {
      "epoch": 2.5833889128643728,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.3675,
      "step": 8231
    },
    {
      "epoch": 2.583702773745537,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.2292,
      "step": 8232
    },
    {
      "epoch": 2.5840166346267015,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.2752,
      "step": 8233
    },
    {
      "epoch": 2.5843304955078663,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.16,
      "step": 8234
    },
    {
      "epoch": 2.5846443563890307,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.2344,
      "step": 8235
    },
    {
      "epoch": 2.584958217270195,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.244,
      "step": 8236
    },
    {
      "epoch": 2.5852720781513594,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.1552,
      "step": 8237
    },
    {
      "epoch": 2.5855859390325238,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.1759,
      "step": 8238
    },
    {
      "epoch": 2.585899799913688,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.123,
      "step": 8239
    },
    {
      "epoch": 2.5862136607948525,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.1789,
      "step": 8240
    },
    {
      "epoch": 2.5865275216760173,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.2807,
      "step": 8241
    },
    {
      "epoch": 2.5868413825571817,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.3216,
      "step": 8242
    },
    {
      "epoch": 2.587155243438346,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.1823,
      "step": 8243
    },
    {
      "epoch": 2.5874691043195104,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.0445,
      "step": 8244
    },
    {
      "epoch": 2.5877829652006747,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.1879,
      "step": 8245
    },
    {
      "epoch": 2.588096826081839,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.2182,
      "step": 8246
    },
    {
      "epoch": 2.5884106869630035,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.1286,
      "step": 8247
    },
    {
      "epoch": 2.5887245478441683,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.2157,
      "step": 8248
    },
    {
      "epoch": 2.5890384087253326,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.1246,
      "step": 8249
    },
    {
      "epoch": 2.589352269606497,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.3203,
      "step": 8250
    },
    {
      "epoch": 2.5896661304876614,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.2874,
      "step": 8251
    },
    {
      "epoch": 2.5899799913688257,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.3615,
      "step": 8252
    },
    {
      "epoch": 2.59029385224999,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.2743,
      "step": 8253
    },
    {
      "epoch": 2.5906077131311545,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.5896,
      "step": 8254
    },
    {
      "epoch": 2.5909215740123193,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.5021,
      "step": 8255
    },
    {
      "epoch": 2.5912354348934836,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 1.5988,
      "step": 8256
    },
    {
      "epoch": 2.591549295774648,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002,
      "loss": 1.7784,
      "step": 8257
    },
    {
      "epoch": 2.5918631566558124,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002,
      "loss": 1.8666,
      "step": 8258
    },
    {
      "epoch": 2.5921770175369767,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.5718,
      "step": 8259
    },
    {
      "epoch": 2.592490878418141,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.5626,
      "step": 8260
    },
    {
      "epoch": 2.5928047392993054,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 1.8873,
      "step": 8261
    },
    {
      "epoch": 2.5931186001804702,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.7911,
      "step": 8262
    },
    {
      "epoch": 2.593432461061634,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 2.0731,
      "step": 8263
    },
    {
      "epoch": 2.593746321942799,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 2.0167,
      "step": 8264
    },
    {
      "epoch": 2.5940601828239633,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.5655,
      "step": 8265
    },
    {
      "epoch": 2.5943740437051277,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.531,
      "step": 8266
    },
    {
      "epoch": 2.594687904586292,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.603,
      "step": 8267
    },
    {
      "epoch": 2.5950017654674564,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 1.8439,
      "step": 8268
    },
    {
      "epoch": 2.5953156263486212,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.5067,
      "step": 8269
    },
    {
      "epoch": 2.595629487229785,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 1.4847,
      "step": 8270
    },
    {
      "epoch": 2.59594334811095,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0002,
      "loss": 1.6979,
      "step": 8271
    },
    {
      "epoch": 2.5962572089921143,
      "grad_norm": 0.90625,
      "learning_rate": 0.0002,
      "loss": 2.0087,
      "step": 8272
    },
    {
      "epoch": 2.5965710698732787,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.5709,
      "step": 8273
    },
    {
      "epoch": 2.596884930754443,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.2485,
      "step": 8274
    },
    {
      "epoch": 2.5971987916356074,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.2437,
      "step": 8275
    },
    {
      "epoch": 2.597512652516772,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.264,
      "step": 8276
    },
    {
      "epoch": 2.597826513397936,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.3852,
      "step": 8277
    },
    {
      "epoch": 2.598140374279101,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.2733,
      "step": 8278
    },
    {
      "epoch": 2.5984542351602653,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3565,
      "step": 8279
    },
    {
      "epoch": 2.5987680960414297,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.3105,
      "step": 8280
    },
    {
      "epoch": 2.599081956922594,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.4101,
      "step": 8281
    },
    {
      "epoch": 2.5993958178037584,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.3282,
      "step": 8282
    },
    {
      "epoch": 2.5997096786849228,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.2049,
      "step": 8283
    },
    {
      "epoch": 2.600023539566087,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.202,
      "step": 8284
    },
    {
      "epoch": 2.600337400447252,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.3742,
      "step": 8285
    },
    {
      "epoch": 2.6006512613284163,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.2715,
      "step": 8286
    },
    {
      "epoch": 2.6009651222095806,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.2058,
      "step": 8287
    },
    {
      "epoch": 2.601278983090745,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2405,
      "step": 8288
    },
    {
      "epoch": 2.6015928439719094,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.1919,
      "step": 8289
    },
    {
      "epoch": 2.6019067048530737,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.0828,
      "step": 8290
    },
    {
      "epoch": 2.602220565734238,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.1654,
      "step": 8291
    },
    {
      "epoch": 2.602534426615403,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.2461,
      "step": 8292
    },
    {
      "epoch": 2.6028482874965673,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.1752,
      "step": 8293
    },
    {
      "epoch": 2.6031621483777316,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.145,
      "step": 8294
    },
    {
      "epoch": 2.603476009258896,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.1916,
      "step": 8295
    },
    {
      "epoch": 2.6037898701400604,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.3542,
      "step": 8296
    },
    {
      "epoch": 2.6041037310212247,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.3022,
      "step": 8297
    },
    {
      "epoch": 2.604417591902389,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.3625,
      "step": 8298
    },
    {
      "epoch": 2.604731452783554,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.06,
      "step": 8299
    },
    {
      "epoch": 2.6050453136647183,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.6145,
      "step": 8300
    },
    {
      "epoch": 2.6053591745458826,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.2478,
      "step": 8301
    },
    {
      "epoch": 2.605673035427047,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.3985,
      "step": 8302
    },
    {
      "epoch": 2.6059868963082113,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.4626,
      "step": 8303
    },
    {
      "epoch": 2.6063007571893757,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 1.2528,
      "step": 8304
    },
    {
      "epoch": 2.60661461807054,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.5425,
      "step": 8305
    },
    {
      "epoch": 2.606928478951705,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.4051,
      "step": 8306
    },
    {
      "epoch": 2.6072423398328692,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.2567,
      "step": 8307
    },
    {
      "epoch": 2.6075562007140336,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.4044,
      "step": 8308
    },
    {
      "epoch": 2.607870061595198,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.563,
      "step": 8309
    },
    {
      "epoch": 2.6081839224763623,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.6939,
      "step": 8310
    },
    {
      "epoch": 2.6084977833575267,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.5868,
      "step": 8311
    },
    {
      "epoch": 2.608811644238691,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0002,
      "loss": 1.9556,
      "step": 8312
    },
    {
      "epoch": 2.609125505119856,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002,
      "loss": 2.0743,
      "step": 8313
    },
    {
      "epoch": 2.6094393660010198,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.6938,
      "step": 8314
    },
    {
      "epoch": 2.6097532268821846,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.8604,
      "step": 8315
    },
    {
      "epoch": 2.610067087763349,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 2.0713,
      "step": 8316
    },
    {
      "epoch": 2.6103809486445133,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.5262,
      "step": 8317
    },
    {
      "epoch": 2.6106948095256777,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002,
      "loss": 2.0565,
      "step": 8318
    },
    {
      "epoch": 2.611008670406842,
      "grad_norm": 0.84375,
      "learning_rate": 0.0002,
      "loss": 1.8823,
      "step": 8319
    },
    {
      "epoch": 2.611322531288007,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002,
      "loss": 1.5251,
      "step": 8320
    },
    {
      "epoch": 2.6116363921691708,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002,
      "loss": 1.4625,
      "step": 8321
    },
    {
      "epoch": 2.6119502530503356,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002,
      "loss": 2.7737,
      "step": 8322
    },
    {
      "epoch": 2.6122641139315,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 1.6456,
      "step": 8323
    },
    {
      "epoch": 2.6125779748126643,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.1549,
      "step": 8324
    },
    {
      "epoch": 2.6128918356938287,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.3724,
      "step": 8325
    },
    {
      "epoch": 2.613205696574993,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.2699,
      "step": 8326
    },
    {
      "epoch": 2.613519557456158,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 1.2312,
      "step": 8327
    },
    {
      "epoch": 2.6138334183373217,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.1906,
      "step": 8328
    },
    {
      "epoch": 2.6141472792184866,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.2342,
      "step": 8329
    },
    {
      "epoch": 2.614461140099651,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.1849,
      "step": 8330
    },
    {
      "epoch": 2.6147750009808153,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2706,
      "step": 8331
    },
    {
      "epoch": 2.6150888618619796,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.3209,
      "step": 8332
    },
    {
      "epoch": 2.615402722743144,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.3661,
      "step": 8333
    },
    {
      "epoch": 2.6157165836243084,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.3045,
      "step": 8334
    },
    {
      "epoch": 2.6160304445054727,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.3723,
      "step": 8335
    },
    {
      "epoch": 2.6163443053866375,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.2643,
      "step": 8336
    },
    {
      "epoch": 2.616658166267802,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.0759,
      "step": 8337
    },
    {
      "epoch": 2.6169720271489663,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2399,
      "step": 8338
    },
    {
      "epoch": 2.6172858880301306,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.2641,
      "step": 8339
    },
    {
      "epoch": 2.617599748911295,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.1964,
      "step": 8340
    },
    {
      "epoch": 2.6179136097924594,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.0877,
      "step": 8341
    },
    {
      "epoch": 2.6182274706736237,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 0.9945,
      "step": 8342
    },
    {
      "epoch": 2.6185413315547885,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.0591,
      "step": 8343
    },
    {
      "epoch": 2.618855192435953,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.2723,
      "step": 8344
    },
    {
      "epoch": 2.6191690533171172,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.0151,
      "step": 8345
    },
    {
      "epoch": 2.6194829141982816,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.1581,
      "step": 8346
    },
    {
      "epoch": 2.619796775079446,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.2291,
      "step": 8347
    },
    {
      "epoch": 2.6201106359606103,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.2839,
      "step": 8348
    },
    {
      "epoch": 2.6204244968417747,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002,
      "loss": 1.3063,
      "step": 8349
    },
    {
      "epoch": 2.6207383577229395,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.3656,
      "step": 8350
    },
    {
      "epoch": 2.621052218604104,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.3309,
      "step": 8351
    },
    {
      "epoch": 2.6213660794852682,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 1.3024,
      "step": 8352
    },
    {
      "epoch": 2.6216799403664326,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.1654,
      "step": 8353
    },
    {
      "epoch": 2.621993801247597,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.3907,
      "step": 8354
    },
    {
      "epoch": 2.6223076621287613,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.5283,
      "step": 8355
    },
    {
      "epoch": 2.6226215230099257,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.2897,
      "step": 8356
    },
    {
      "epoch": 2.6229353838910905,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002,
      "loss": 1.6339,
      "step": 8357
    },
    {
      "epoch": 2.623249244772255,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.5963,
      "step": 8358
    },
    {
      "epoch": 2.623563105653419,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.7894,
      "step": 8359
    },
    {
      "epoch": 2.6238769665345836,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 1.7884,
      "step": 8360
    },
    {
      "epoch": 2.624190827415748,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.5161,
      "step": 8361
    },
    {
      "epoch": 2.6245046882969123,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.5421,
      "step": 8362
    },
    {
      "epoch": 2.6248185491780767,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.5299,
      "step": 8363
    },
    {
      "epoch": 2.6251324100592415,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002,
      "loss": 1.8635,
      "step": 8364
    },
    {
      "epoch": 2.6254462709404054,
      "grad_norm": 1.46875,
      "learning_rate": 0.0002,
      "loss": 1.841,
      "step": 8365
    },
    {
      "epoch": 2.62576013182157,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.7608,
      "step": 8366
    },
    {
      "epoch": 2.6260739927027346,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.3834,
      "step": 8367
    },
    {
      "epoch": 2.626387853583899,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.4079,
      "step": 8368
    },
    {
      "epoch": 2.6267017144650633,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.2541,
      "step": 8369
    },
    {
      "epoch": 2.6270155753462276,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.3591,
      "step": 8370
    },
    {
      "epoch": 2.6273294362273925,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.4543,
      "step": 8371
    },
    {
      "epoch": 2.6276432971085564,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002,
      "loss": 2.1663,
      "step": 8372
    },
    {
      "epoch": 2.627957157989721,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.6518,
      "step": 8373
    },
    {
      "epoch": 2.6282710188708855,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.3158,
      "step": 8374
    },
    {
      "epoch": 2.62858487975205,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.2642,
      "step": 8375
    },
    {
      "epoch": 2.6288987406332143,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.1811,
      "step": 8376
    },
    {
      "epoch": 2.6292126015143786,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.2948,
      "step": 8377
    },
    {
      "epoch": 2.6295264623955434,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.2624,
      "step": 8378
    },
    {
      "epoch": 2.6298403232767074,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.2263,
      "step": 8379
    },
    {
      "epoch": 2.630154184157872,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.3232,
      "step": 8380
    },
    {
      "epoch": 2.6304680450390365,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.1449,
      "step": 8381
    },
    {
      "epoch": 2.630781905920201,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.3683,
      "step": 8382
    },
    {
      "epoch": 2.6310957668013653,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.1488,
      "step": 8383
    },
    {
      "epoch": 2.6314096276825296,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.1882,
      "step": 8384
    },
    {
      "epoch": 2.631723488563694,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.1748,
      "step": 8385
    },
    {
      "epoch": 2.6320373494448583,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.2578,
      "step": 8386
    },
    {
      "epoch": 2.632351210326023,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.2396,
      "step": 8387
    },
    {
      "epoch": 2.6326650712071875,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.1114,
      "step": 8388
    },
    {
      "epoch": 2.632978932088352,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.2044,
      "step": 8389
    },
    {
      "epoch": 2.6332927929695162,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.3455,
      "step": 8390
    },
    {
      "epoch": 2.6336066538506806,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.1319,
      "step": 8391
    },
    {
      "epoch": 2.633920514731845,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.1566,
      "step": 8392
    },
    {
      "epoch": 2.6342343756130093,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.156,
      "step": 8393
    },
    {
      "epoch": 2.634548236494174,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.0712,
      "step": 8394
    },
    {
      "epoch": 2.6348620973753385,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.2226,
      "step": 8395
    },
    {
      "epoch": 2.635175958256503,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.2076,
      "step": 8396
    },
    {
      "epoch": 2.6354898191376672,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.2224,
      "step": 8397
    },
    {
      "epoch": 2.6358036800188316,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 1.3536,
      "step": 8398
    },
    {
      "epoch": 2.636117540899996,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.3716,
      "step": 8399
    },
    {
      "epoch": 2.6364314017811603,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.6506,
      "step": 8400
    },
    {
      "epoch": 2.636745262662325,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.519,
      "step": 8401
    },
    {
      "epoch": 2.6370591235434895,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.5257,
      "step": 8402
    },
    {
      "epoch": 2.637372984424654,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.3942,
      "step": 8403
    },
    {
      "epoch": 2.637686845305818,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 1.792,
      "step": 8404
    },
    {
      "epoch": 2.6380007061869826,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.801,
      "step": 8405
    },
    {
      "epoch": 2.638314567068147,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002,
      "loss": 2.1817,
      "step": 8406
    },
    {
      "epoch": 2.6386284279493113,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.4631,
      "step": 8407
    },
    {
      "epoch": 2.638942288830476,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.758,
      "step": 8408
    },
    {
      "epoch": 2.6392561497116405,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002,
      "loss": 1.6361,
      "step": 8409
    },
    {
      "epoch": 2.639570010592805,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.4907,
      "step": 8410
    },
    {
      "epoch": 2.639883871473969,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.9963,
      "step": 8411
    },
    {
      "epoch": 2.6401977323551336,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 1.8844,
      "step": 8412
    },
    {
      "epoch": 2.640511593236298,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 2.2396,
      "step": 8413
    },
    {
      "epoch": 2.6408254541174623,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 1.8646,
      "step": 8414
    },
    {
      "epoch": 2.641139314998627,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.4994,
      "step": 8415
    },
    {
      "epoch": 2.641139314998627,
      "eval_loss": 1.8945120573043823,
      "eval_runtime": 123.0456,
      "eval_samples_per_second": 8.127,
      "eval_steps_per_second": 8.127,
      "step": 8415
    },
    {
      "epoch": 2.641139314998627,
      "mmlu_eval_accuracy": 0.42608607294045037,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.36363636363636365,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.5,
      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.24390243902439024,
      "mmlu_eval_accuracy_formal_logic": 0.42857142857142855,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.21875,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.3333333333333333,
      "mmlu_eval_accuracy_high_school_european_history": 0.4444444444444444,
      "mmlu_eval_accuracy_high_school_geography": 0.6363636363636364,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.5,
      "mmlu_eval_accuracy_human_aging": 0.391304347826087,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
      "mmlu_eval_accuracy_management": 0.45454545454545453,
      "mmlu_eval_accuracy_marketing": 0.68,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5581395348837209,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.24,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.4117647058823529,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.38235294117647056,
      "mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.2856478922550854,
      "step": 8415
    },
    {
      "epoch": 2.6414531758797914,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.4189,
      "step": 8416
    },
    {
      "epoch": 2.641767036760956,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.4305,
      "step": 8417
    },
    {
      "epoch": 2.64208089764212,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.4127,
      "step": 8418
    },
    {
      "epoch": 2.6423947585232845,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0002,
      "loss": 1.9099,
      "step": 8419
    },
    {
      "epoch": 2.642708619404449,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.4152,
      "step": 8420
    },
    {
      "epoch": 2.6430224802856133,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.4639,
      "step": 8421
    },
    {
      "epoch": 2.643336341166778,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002,
      "loss": 2.7808,
      "step": 8422
    },
    {
      "epoch": 2.643650202047942,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.7995,
      "step": 8423
    },
    {
      "epoch": 2.643964062929107,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.2992,
      "step": 8424
    },
    {
      "epoch": 2.644277923810271,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.2542,
      "step": 8425
    },
    {
      "epoch": 2.6445917846914355,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.4125,
      "step": 8426
    },
    {
      "epoch": 2.6449056455726,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.337,
      "step": 8427
    },
    {
      "epoch": 2.6452195064537642,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.2766,
      "step": 8428
    },
    {
      "epoch": 2.645533367334929,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.3633,
      "step": 8429
    },
    {
      "epoch": 2.645847228216093,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.1436,
      "step": 8430
    },
    {
      "epoch": 2.646161089097258,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.3849,
      "step": 8431
    },
    {
      "epoch": 2.646474949978422,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.3329,
      "step": 8432
    },
    {
      "epoch": 2.6467888108595865,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.2212,
      "step": 8433
    },
    {
      "epoch": 2.647102671740751,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.2438,
      "step": 8434
    },
    {
      "epoch": 2.6474165326219152,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.2411,
      "step": 8435
    },
    {
      "epoch": 2.6477303935030796,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2334,
      "step": 8436
    },
    {
      "epoch": 2.648044254384244,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.3269,
      "step": 8437
    },
    {
      "epoch": 2.6483581152654088,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.1645,
      "step": 8438
    },
    {
      "epoch": 2.648671976146573,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.2039,
      "step": 8439
    },
    {
      "epoch": 2.6489858370277375,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.2391,
      "step": 8440
    },
    {
      "epoch": 2.649299697908902,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.0828,
      "step": 8441
    },
    {
      "epoch": 2.649613558790066,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.1592,
      "step": 8442
    },
    {
      "epoch": 2.6499274196712306,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.0751,
      "step": 8443
    },
    {
      "epoch": 2.650241280552395,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0002,
      "loss": 1.3597,
      "step": 8444
    },
    {
      "epoch": 2.6505551414335597,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.1523,
      "step": 8445
    },
    {
      "epoch": 2.650869002314724,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.2061,
      "step": 8446
    },
    {
      "epoch": 2.6511828631958885,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.1965,
      "step": 8447
    },
    {
      "epoch": 2.651496724077053,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.1895,
      "step": 8448
    },
    {
      "epoch": 2.651810584958217,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.2136,
      "step": 8449
    },
    {
      "epoch": 2.6521244458393816,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.3367,
      "step": 8450
    },
    {
      "epoch": 2.652438306720546,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.3442,
      "step": 8451
    },
    {
      "epoch": 2.6527521676017107,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.2243,
      "step": 8452
    },
    {
      "epoch": 2.653066028482875,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.3153,
      "step": 8453
    },
    {
      "epoch": 2.6533798893640395,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.4304,
      "step": 8454
    },
    {
      "epoch": 2.653693750245204,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.5442,
      "step": 8455
    },
    {
      "epoch": 2.654007611126368,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 1.6766,
      "step": 8456
    },
    {
      "epoch": 2.6543214720075325,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 1.6185,
      "step": 8457
    },
    {
      "epoch": 2.654635332888697,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.6418,
      "step": 8458
    },
    {
      "epoch": 2.6549491937698617,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 1.632,
      "step": 8459
    },
    {
      "epoch": 2.655263054651026,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.765,
      "step": 8460
    },
    {
      "epoch": 2.6555769155321904,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.4777,
      "step": 8461
    },
    {
      "epoch": 2.655890776413355,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.6658,
      "step": 8462
    },
    {
      "epoch": 2.656204637294519,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002,
      "loss": 1.8833,
      "step": 8463
    },
    {
      "epoch": 2.6565184981756835,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.9558,
      "step": 8464
    },
    {
      "epoch": 2.656832359056848,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002,
      "loss": 1.868,
      "step": 8465
    },
    {
      "epoch": 2.6571462199380127,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002,
      "loss": 1.7597,
      "step": 8466
    },
    {
      "epoch": 2.657460080819177,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.4994,
      "step": 8467
    },
    {
      "epoch": 2.6577739417003414,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002,
      "loss": 1.6752,
      "step": 8468
    },
    {
      "epoch": 2.658087802581506,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 1.7291,
      "step": 8469
    },
    {
      "epoch": 2.65840166346267,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002,
      "loss": 1.6224,
      "step": 8470
    },
    {
      "epoch": 2.6587155243438345,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0002,
      "loss": 1.856,
      "step": 8471
    },
    {
      "epoch": 2.659029385224999,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 2.2316,
      "step": 8472
    },
    {
      "epoch": 2.6593432461061637,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.7545,
      "step": 8473
    },
    {
      "epoch": 2.6596571069873276,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.257,
      "step": 8474
    },
    {
      "epoch": 2.6599709678684924,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.1942,
      "step": 8475
    },
    {
      "epoch": 2.6602848287496568,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.2967,
      "step": 8476
    },
    {
      "epoch": 2.660598689630821,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.1637,
      "step": 8477
    },
    {
      "epoch": 2.6609125505119855,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.4163,
      "step": 8478
    },
    {
      "epoch": 2.66122641139315,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.2409,
      "step": 8479
    },
    {
      "epoch": 2.6615402722743147,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.3627,
      "step": 8480
    },
    {
      "epoch": 2.6618541331554786,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.3037,
      "step": 8481
    },
    {
      "epoch": 2.6621679940366434,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.1783,
      "step": 8482
    },
    {
      "epoch": 2.6624818549178078,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.3501,
      "step": 8483
    },
    {
      "epoch": 2.662795715798972,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0002,
      "loss": 1.2841,
      "step": 8484
    },
    {
      "epoch": 2.6631095766801365,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.1642,
      "step": 8485
    },
    {
      "epoch": 2.663423437561301,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.1707,
      "step": 8486
    },
    {
      "epoch": 2.6637372984424657,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.2257,
      "step": 8487
    },
    {
      "epoch": 2.6640511593236296,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.2401,
      "step": 8488
    },
    {
      "epoch": 2.6643650202047944,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.1447,
      "step": 8489
    },
    {
      "epoch": 2.6646788810859587,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.1235,
      "step": 8490
    },
    {
      "epoch": 2.664992741967123,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.1939,
      "step": 8491
    },
    {
      "epoch": 2.6653066028482875,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.1468,
      "step": 8492
    },
    {
      "epoch": 2.665620463729452,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.0362,
      "step": 8493
    },
    {
      "epoch": 2.665934324610616,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.2149,
      "step": 8494
    },
    {
      "epoch": 2.6662481854917806,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.1619,
      "step": 8495
    },
    {
      "epoch": 2.6665620463729454,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.1317,
      "step": 8496
    },
    {
      "epoch": 2.6668759072541097,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.0797,
      "step": 8497
    },
    {
      "epoch": 2.667189768135274,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.3879,
      "step": 8498
    },
    {
      "epoch": 2.6675036290164384,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.2889,
      "step": 8499
    },
    {
      "epoch": 2.667817489897603,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.3283,
      "step": 8500
    },
    {
      "epoch": 2.668131350778767,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.3981,
      "step": 8501
    },
    {
      "epoch": 2.6684452116599315,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.3878,
      "step": 8502
    },
    {
      "epoch": 2.6687590725410963,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.5925,
      "step": 8503
    },
    {
      "epoch": 2.6690729334222607,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.44,
      "step": 8504
    },
    {
      "epoch": 2.669386794303425,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.6258,
      "step": 8505
    },
    {
      "epoch": 2.6697006551845894,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.4896,
      "step": 8506
    },
    {
      "epoch": 2.670014516065754,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.756,
      "step": 8507
    },
    {
      "epoch": 2.670328376946918,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002,
      "loss": 1.9529,
      "step": 8508
    },
    {
      "epoch": 2.6706422378280825,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.3928,
      "step": 8509
    },
    {
      "epoch": 2.6709560987092473,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.7558,
      "step": 8510
    },
    {
      "epoch": 2.6712699595904117,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.9714,
      "step": 8511
    },
    {
      "epoch": 2.671583820471576,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.3451,
      "step": 8512
    },
    {
      "epoch": 2.6718976813527404,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.7067,
      "step": 8513
    },
    {
      "epoch": 2.672211542233905,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.6845,
      "step": 8514
    },
    {
      "epoch": 2.672525403115069,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 1.6022,
      "step": 8515
    },
    {
      "epoch": 2.6728392639962335,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.9001,
      "step": 8516
    },
    {
      "epoch": 2.6731531248773983,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 1.679,
      "step": 8517
    },
    {
      "epoch": 2.6734669857585627,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.3794,
      "step": 8518
    },
    {
      "epoch": 2.673780846639727,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002,
      "loss": 1.6959,
      "step": 8519
    },
    {
      "epoch": 2.6740947075208914,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0002,
      "loss": 1.7331,
      "step": 8520
    },
    {
      "epoch": 2.6744085684020558,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002,
      "loss": 1.7728,
      "step": 8521
    },
    {
      "epoch": 2.67472242928322,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002,
      "loss": 2.2475,
      "step": 8522
    },
    {
      "epoch": 2.6750362901643845,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.5638,
      "step": 8523
    },
    {
      "epoch": 2.6753501510455493,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.3461,
      "step": 8524
    },
    {
      "epoch": 2.675664011926713,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.4559,
      "step": 8525
    },
    {
      "epoch": 2.675977872807878,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.3845,
      "step": 8526
    },
    {
      "epoch": 2.6762917336890424,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.3858,
      "step": 8527
    },
    {
      "epoch": 2.6766055945702067,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.4012,
      "step": 8528
    },
    {
      "epoch": 2.676919455451371,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.4453,
      "step": 8529
    },
    {
      "epoch": 2.6772333163325355,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.3059,
      "step": 8530
    },
    {
      "epoch": 2.6775471772137003,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2132,
      "step": 8531
    },
    {
      "epoch": 2.677861038094864,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.1632,
      "step": 8532
    },
    {
      "epoch": 2.678174898976029,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002,
      "loss": 1.2249,
      "step": 8533
    },
    {
      "epoch": 2.6784887598571934,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.2381,
      "step": 8534
    },
    {
      "epoch": 2.6788026207383577,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.151,
      "step": 8535
    },
    {
      "epoch": 2.679116481619522,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.1198,
      "step": 8536
    },
    {
      "epoch": 2.6794303425006865,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.4175,
      "step": 8537
    },
    {
      "epoch": 2.6797442033818513,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.1252,
      "step": 8538
    },
    {
      "epoch": 2.680058064263015,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.3283,
      "step": 8539
    },
    {
      "epoch": 2.68037192514418,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.1368,
      "step": 8540
    },
    {
      "epoch": 2.6806857860253444,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.1963,
      "step": 8541
    },
    {
      "epoch": 2.6809996469065087,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.2002,
      "step": 8542
    },
    {
      "epoch": 2.681313507787673,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.1619,
      "step": 8543
    },
    {
      "epoch": 2.6816273686688374,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.2311,
      "step": 8544
    },
    {
      "epoch": 2.681941229550002,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.2419,
      "step": 8545
    },
    {
      "epoch": 2.682255090431166,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.0106,
      "step": 8546
    },
    {
      "epoch": 2.682568951312331,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.2238,
      "step": 8547
    },
    {
      "epoch": 2.6828828121934953,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.2007,
      "step": 8548
    },
    {
      "epoch": 2.6831966730746597,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 1.1026,
      "step": 8549
    },
    {
      "epoch": 2.683510533955824,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 1.3844,
      "step": 8550
    },
    {
      "epoch": 2.6838243948369884,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.2921,
      "step": 8551
    },
    {
      "epoch": 2.684138255718153,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 1.3349,
      "step": 8552
    },
    {
      "epoch": 2.684452116599317,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.5137,
      "step": 8553
    },
    {
      "epoch": 2.684765977480482,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 1.6831,
      "step": 8554
    },
    {
      "epoch": 2.6850798383616463,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0002,
      "loss": 1.8571,
      "step": 8555
    },
    {
      "epoch": 2.6853936992428107,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.4632,
      "step": 8556
    },
    {
      "epoch": 2.685707560123975,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002,
      "loss": 1.7118,
      "step": 8557
    },
    {
      "epoch": 2.6860214210051394,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.5354,
      "step": 8558
    },
    {
      "epoch": 2.6863352818863038,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.4313,
      "step": 8559
    },
    {
      "epoch": 2.686649142767468,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.7895,
      "step": 8560
    },
    {
      "epoch": 2.686963003648633,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.5443,
      "step": 8561
    },
    {
      "epoch": 2.6872768645297973,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002,
      "loss": 1.5314,
      "step": 8562
    },
    {
      "epoch": 2.6875907254109617,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 2.2068,
      "step": 8563
    },
    {
      "epoch": 2.687904586292126,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 1.7831,
      "step": 8564
    },
    {
      "epoch": 2.6882184471732904,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.4933,
      "step": 8565
    },
    {
      "epoch": 2.6885323080544548,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 1.6909,
      "step": 8566
    },
    {
      "epoch": 2.688846168935619,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.9929,
      "step": 8567
    },
    {
      "epoch": 2.689160029816784,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002,
      "loss": 1.6094,
      "step": 8568
    },
    {
      "epoch": 2.6894738906979483,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.3614,
      "step": 8569
    },
    {
      "epoch": 2.6897877515791127,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 1.5155,
      "step": 8570
    },
    {
      "epoch": 2.690101612460277,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 1.6672,
      "step": 8571
    },
    {
      "epoch": 2.6904154733414414,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 2.2199,
      "step": 8572
    },
    {
      "epoch": 2.6907293342226057,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.4084,
      "step": 8573
    },
    {
      "epoch": 2.69104319510377,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.3214,
      "step": 8574
    },
    {
      "epoch": 2.691357055984935,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.2291,
      "step": 8575
    },
    {
      "epoch": 2.691670916866099,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.3664,
      "step": 8576
    },
    {
      "epoch": 2.6919847777472636,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.2657,
      "step": 8577
    },
    {
      "epoch": 2.692298638628428,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.2366,
      "step": 8578
    },
    {
      "epoch": 2.6926124995095924,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.1635,
      "step": 8579
    },
    {
      "epoch": 2.6929263603907567,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.3039,
      "step": 8580
    },
    {
      "epoch": 2.693240221271921,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.4299,
      "step": 8581
    },
    {
      "epoch": 2.693554082153086,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.2262,
      "step": 8582
    },
    {
      "epoch": 2.69386794303425,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.1847,
      "step": 8583
    },
    {
      "epoch": 2.6941818039154146,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.3661,
      "step": 8584
    },
    {
      "epoch": 2.694495664796579,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.2268,
      "step": 8585
    },
    {
      "epoch": 2.6948095256777433,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.2002,
      "step": 8586
    },
    {
      "epoch": 2.6951233865589077,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.3211,
      "step": 8587
    },
    {
      "epoch": 2.695437247440072,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.2702,
      "step": 8588
    },
    {
      "epoch": 2.695751108321237,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.1281,
      "step": 8589
    },
    {
      "epoch": 2.696064969202401,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.1183,
      "step": 8590
    },
    {
      "epoch": 2.6963788300835656,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.2107,
      "step": 8591
    },
    {
      "epoch": 2.69669269096473,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.2434,
      "step": 8592
    },
    {
      "epoch": 2.6970065518458943,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.1655,
      "step": 8593
    },
    {
      "epoch": 2.6973204127270587,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.0698,
      "step": 8594
    },
    {
      "epoch": 2.697634273608223,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.1282,
      "step": 8595
    },
    {
      "epoch": 2.6979481344893874,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.2471,
      "step": 8596
    },
    {
      "epoch": 2.698261995370552,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.2097,
      "step": 8597
    },
    {
      "epoch": 2.6985758562517166,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.2414,
      "step": 8598
    },
    {
      "epoch": 2.698889717132881,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002,
      "loss": 1.499,
      "step": 8599
    },
    {
      "epoch": 2.6992035780140453,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.347,
      "step": 8600
    },
    {
      "epoch": 2.6995174388952097,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.6096,
      "step": 8601
    },
    {
      "epoch": 2.699831299776374,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.3154,
      "step": 8602
    },
    {
      "epoch": 2.699831299776374,
      "eval_loss": 1.854783058166504,
      "eval_runtime": 123.9272,
      "eval_samples_per_second": 8.069,
      "eval_steps_per_second": 8.069,
      "step": 8602
    },
    {
      "epoch": 2.699831299776374,
      "mmlu_eval_accuracy": 0.41937990182100365,
      "mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.2727272727272727,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.5,
      "mmlu_eval_accuracy_college_chemistry": 0.25,
      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.18181818181818182,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.1951219512195122,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.6,
      "mmlu_eval_accuracy_high_school_biology": 0.25,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.3333333333333333,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.5,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5348837209302325,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
      "mmlu_eval_accuracy_high_school_psychology": 0.6833333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5,
      "mmlu_eval_accuracy_high_school_world_history": 0.5,
      "mmlu_eval_accuracy_human_aging": 0.43478260869565216,
      "mmlu_eval_accuracy_human_sexuality": 0.25,
      "mmlu_eval_accuracy_international_law": 0.6923076923076923,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.4444444444444444,
      "mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.6,
      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
      "mmlu_eval_accuracy_miscellaneous": 0.5697674418604651,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.26,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.4117647058823529,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.3333333333333333,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.2714938804225,
      "step": 8602
    },
    {
      "epoch": 2.7001451606575384,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.3572,
      "step": 8603
    },
    {
      "epoch": 2.7004590215387028,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.5309,
      "step": 8604
    },
    {
      "epoch": 2.7007728824198676,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.5098,
      "step": 8605
    },
    {
      "epoch": 2.701086743301032,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.4334,
      "step": 8606
    },
    {
      "epoch": 2.7014006041821963,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.9862,
      "step": 8607
    },
    {
      "epoch": 2.7017144650633607,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.5239,
      "step": 8608
    },
    {
      "epoch": 2.702028325944525,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.6678,
      "step": 8609
    },
    {
      "epoch": 2.7023421868256894,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.5955,
      "step": 8610
    },
    {
      "epoch": 2.7026560477068537,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.5368,
      "step": 8611
    },
    {
      "epoch": 2.7029699085880186,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.7202,
      "step": 8612
    },
    {
      "epoch": 2.703283769469183,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.43,
      "step": 8613
    },
    {
      "epoch": 2.7035976303503473,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 1.7292,
      "step": 8614
    },
    {
      "epoch": 2.7039114912315116,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0002,
      "loss": 1.8022,
      "step": 8615
    },
    {
      "epoch": 2.704225352112676,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.688,
      "step": 8616
    },
    {
      "epoch": 2.7045392129938404,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.681,
      "step": 8617
    },
    {
      "epoch": 2.7048530738750047,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 1.8319,
      "step": 8618
    },
    {
      "epoch": 2.7051669347561695,
      "grad_norm": 1.40625,
      "learning_rate": 0.0002,
      "loss": 1.6245,
      "step": 8619
    },
    {
      "epoch": 2.705480795637334,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.5478,
      "step": 8620
    },
    {
      "epoch": 2.7057946565184983,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.5435,
      "step": 8621
    },
    {
      "epoch": 2.7061085173996626,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002,
      "loss": 2.2143,
      "step": 8622
    },
    {
      "epoch": 2.706422378280827,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.8024,
      "step": 8623
    },
    {
      "epoch": 2.7067362391619914,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.1995,
      "step": 8624
    },
    {
      "epoch": 2.7070501000431557,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.2541,
      "step": 8625
    },
    {
      "epoch": 2.7073639609243205,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.3442,
      "step": 8626
    },
    {
      "epoch": 2.7076778218054844,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.3143,
      "step": 8627
    },
    {
      "epoch": 2.7079916826866492,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.2439,
      "step": 8628
    },
    {
      "epoch": 2.7083055435678136,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.1414,
      "step": 8629
    },
    {
      "epoch": 2.708619404448978,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.0883,
      "step": 8630
    },
    {
      "epoch": 2.7089332653301423,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.1546,
      "step": 8631
    },
    {
      "epoch": 2.7092471262113067,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.1885,
      "step": 8632
    },
    {
      "epoch": 2.7095609870924715,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.2473,
      "step": 8633
    },
    {
      "epoch": 2.7098748479736354,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.4387,
      "step": 8634
    },
    {
      "epoch": 2.7101887088548002,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.146,
      "step": 8635
    },
    {
      "epoch": 2.7105025697359646,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.2243,
      "step": 8636
    },
    {
      "epoch": 2.710816430617129,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.1487,
      "step": 8637
    },
    {
      "epoch": 2.7111302914982933,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.2089,
      "step": 8638
    },
    {
      "epoch": 2.7114441523794577,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.2721,
      "step": 8639
    },
    {
      "epoch": 2.7117580132606225,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.2621,
      "step": 8640
    },
    {
      "epoch": 2.7120718741417864,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.2357,
      "step": 8641
    },
    {
      "epoch": 2.712385735022951,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.1562,
      "step": 8642
    },
    {
      "epoch": 2.7126995959041156,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.2351,
      "step": 8643
    },
    {
      "epoch": 2.71301345678528,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.2223,
      "step": 8644
    },
    {
      "epoch": 2.7133273176664443,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.31,
      "step": 8645
    },
    {
      "epoch": 2.7136411785476087,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.1058,
      "step": 8646
    },
    {
      "epoch": 2.713955039428773,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 1.3404,
      "step": 8647
    },
    {
      "epoch": 2.7142689003099374,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.2708,
      "step": 8648
    },
    {
      "epoch": 2.714582761191102,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 1.177,
      "step": 8649
    },
    {
      "epoch": 2.7148966220722666,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.283,
      "step": 8650
    },
    {
      "epoch": 2.715210482953431,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.598,
      "step": 8651
    },
    {
      "epoch": 2.7155243438345953,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.2474,
      "step": 8652
    },
    {
      "epoch": 2.7158382047157597,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.5265,
      "step": 8653
    },
    {
      "epoch": 2.716152065596924,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 1.4124,
      "step": 8654
    },
    {
      "epoch": 2.7164659264780884,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.3375,
      "step": 8655
    },
    {
      "epoch": 2.716779787359253,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.524,
      "step": 8656
    },
    {
      "epoch": 2.7170936482404175,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 1.2545,
      "step": 8657
    },
    {
      "epoch": 2.717407509121582,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.5326,
      "step": 8658
    },
    {
      "epoch": 2.7177213700027463,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.6215,
      "step": 8659
    },
    {
      "epoch": 2.7180352308839106,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.4816,
      "step": 8660
    },
    {
      "epoch": 2.718349091765075,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.3333,
      "step": 8661
    },
    {
      "epoch": 2.7186629526462394,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.5719,
      "step": 8662
    },
    {
      "epoch": 2.718976813527404,
      "grad_norm": 0.75,
      "learning_rate": 0.0002,
      "loss": 1.8487,
      "step": 8663
    },
    {
      "epoch": 2.7192906744085685,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 1.7522,
      "step": 8664
    },
    {
      "epoch": 2.719604535289733,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.9769,
      "step": 8665
    },
    {
      "epoch": 2.7199183961708973,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.7196,
      "step": 8666
    },
    {
      "epoch": 2.7202322570520616,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.9253,
      "step": 8667
    },
    {
      "epoch": 2.720546117933226,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002,
      "loss": 1.5955,
      "step": 8668
    },
    {
      "epoch": 2.7208599788143903,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.5427,
      "step": 8669
    },
    {
      "epoch": 2.721173839695555,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0002,
      "loss": 1.7613,
      "step": 8670
    },
    {
      "epoch": 2.7214877005767195,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.6044,
      "step": 8671
    },
    {
      "epoch": 2.721801561457884,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0002,
      "loss": 2.3301,
      "step": 8672
    },
    {
      "epoch": 2.7221154223390482,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.7773,
      "step": 8673
    },
    {
      "epoch": 2.7224292832202126,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.1072,
      "step": 8674
    },
    {
      "epoch": 2.722743144101377,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.2099,
      "step": 8675
    },
    {
      "epoch": 2.7230570049825413,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.2762,
      "step": 8676
    },
    {
      "epoch": 2.723370865863706,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.1811,
      "step": 8677
    },
    {
      "epoch": 2.7236847267448705,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.3018,
      "step": 8678
    },
    {
      "epoch": 2.723998587626035,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.3907,
      "step": 8679
    },
    {
      "epoch": 2.7243124485071992,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.3295,
      "step": 8680
    },
    {
      "epoch": 2.7246263093883636,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.1297,
      "step": 8681
    },
    {
      "epoch": 2.724940170269528,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.2774,
      "step": 8682
    },
    {
      "epoch": 2.7252540311506923,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.2902,
      "step": 8683
    },
    {
      "epoch": 2.725567892031857,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.1716,
      "step": 8684
    },
    {
      "epoch": 2.725881752913021,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.2768,
      "step": 8685
    },
    {
      "epoch": 2.726195613794186,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.2164,
      "step": 8686
    },
    {
      "epoch": 2.72650947467535,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.2066,
      "step": 8687
    },
    {
      "epoch": 2.7268233355565146,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.1088,
      "step": 8688
    },
    {
      "epoch": 2.727137196437679,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.1148,
      "step": 8689
    },
    {
      "epoch": 2.7274510573188433,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.1722,
      "step": 8690
    },
    {
      "epoch": 2.727764918200008,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.3588,
      "step": 8691
    },
    {
      "epoch": 2.728078779081172,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.146,
      "step": 8692
    },
    {
      "epoch": 2.728392639962337,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.077,
      "step": 8693
    },
    {
      "epoch": 2.728706500843501,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.3593,
      "step": 8694
    },
    {
      "epoch": 2.7290203617246656,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.1837,
      "step": 8695
    },
    {
      "epoch": 2.72933422260583,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002,
      "loss": 1.3162,
      "step": 8696
    },
    {
      "epoch": 2.7296480834869943,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002,
      "loss": 1.398,
      "step": 8697
    },
    {
      "epoch": 2.729961944368159,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 0.9846,
      "step": 8698
    },
    {
      "epoch": 2.730275805249323,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.1597,
      "step": 8699
    },
    {
      "epoch": 2.730589666130488,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.2665,
      "step": 8700
    },
    {
      "epoch": 2.730903527011652,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.2117,
      "step": 8701
    },
    {
      "epoch": 2.7312173878928165,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.5063,
      "step": 8702
    },
    {
      "epoch": 2.731531248773981,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.4168,
      "step": 8703
    },
    {
      "epoch": 2.7318451096551453,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002,
      "loss": 1.8986,
      "step": 8704
    },
    {
      "epoch": 2.7321589705363096,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.4895,
      "step": 8705
    },
    {
      "epoch": 2.732472831417474,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.3423,
      "step": 8706
    },
    {
      "epoch": 2.732786692298639,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002,
      "loss": 1.7741,
      "step": 8707
    },
    {
      "epoch": 2.733100553179803,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 1.3557,
      "step": 8708
    },
    {
      "epoch": 2.7334144140609675,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.4249,
      "step": 8709
    },
    {
      "epoch": 2.733728274942132,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002,
      "loss": 1.7385,
      "step": 8710
    },
    {
      "epoch": 2.7340421358232962,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0002,
      "loss": 1.7876,
      "step": 8711
    },
    {
      "epoch": 2.7343559967044606,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 1.7375,
      "step": 8712
    },
    {
      "epoch": 2.734669857585625,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.5032,
      "step": 8713
    },
    {
      "epoch": 2.73498371846679,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.5102,
      "step": 8714
    },
    {
      "epoch": 2.735297579347954,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.9412,
      "step": 8715
    },
    {
      "epoch": 2.7356114402291185,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002,
      "loss": 2.1166,
      "step": 8716
    },
    {
      "epoch": 2.735925301110283,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002,
      "loss": 1.4727,
      "step": 8717
    },
    {
      "epoch": 2.7362391619914472,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.4201,
      "step": 8718
    },
    {
      "epoch": 2.7365530228726116,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.4954,
      "step": 8719
    },
    {
      "epoch": 2.736866883753776,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.4639,
      "step": 8720
    },
    {
      "epoch": 2.7371807446349408,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.8064,
      "step": 8721
    },
    {
      "epoch": 2.737494605516105,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002,
      "loss": 2.2398,
      "step": 8722
    },
    {
      "epoch": 2.7378084663972695,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.4036,
      "step": 8723
    },
    {
      "epoch": 2.738122327278434,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.2911,
      "step": 8724
    },
    {
      "epoch": 2.738436188159598,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.3762,
      "step": 8725
    },
    {
      "epoch": 2.7387500490407626,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.3825,
      "step": 8726
    },
    {
      "epoch": 2.739063909921927,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.2075,
      "step": 8727
    },
    {
      "epoch": 2.7393777708030917,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.2913,
      "step": 8728
    },
    {
      "epoch": 2.739691631684256,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.2767,
      "step": 8729
    },
    {
      "epoch": 2.7400054925654205,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.3512,
      "step": 8730
    },
    {
      "epoch": 2.740319353446585,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.1116,
      "step": 8731
    },
    {
      "epoch": 2.740633214327749,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.3591,
      "step": 8732
    },
    {
      "epoch": 2.7409470752089136,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.3166,
      "step": 8733
    },
    {
      "epoch": 2.741260936090078,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2564,
      "step": 8734
    },
    {
      "epoch": 2.7415747969712427,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.2005,
      "step": 8735
    },
    {
      "epoch": 2.7418886578524067,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.2311,
      "step": 8736
    },
    {
      "epoch": 2.7422025187335715,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.13,
      "step": 8737
    },
    {
      "epoch": 2.742516379614736,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.3589,
      "step": 8738
    },
    {
      "epoch": 2.7428302404959,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.3245,
      "step": 8739
    },
    {
      "epoch": 2.7431441013770645,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 1.2068,
      "step": 8740
    },
    {
      "epoch": 2.743457962258229,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.2018,
      "step": 8741
    },
    {
      "epoch": 2.7437718231393937,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.1263,
      "step": 8742
    },
    {
      "epoch": 2.7440856840205576,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.2845,
      "step": 8743
    },
    {
      "epoch": 2.7443995449017224,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.1948,
      "step": 8744
    },
    {
      "epoch": 2.744713405782887,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.2739,
      "step": 8745
    },
    {
      "epoch": 2.745027266664051,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.1993,
      "step": 8746
    },
    {
      "epoch": 2.7453411275452155,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.1774,
      "step": 8747
    },
    {
      "epoch": 2.74565498842638,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.2651,
      "step": 8748
    },
    {
      "epoch": 2.7459688493075447,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.2833,
      "step": 8749
    },
    {
      "epoch": 2.7462827101887086,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.4769,
      "step": 8750
    },
    {
      "epoch": 2.7465965710698734,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.5504,
      "step": 8751
    },
    {
      "epoch": 2.746910431951038,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.1777,
      "step": 8752
    },
    {
      "epoch": 2.747224292832202,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.576,
      "step": 8753
    },
    {
      "epoch": 2.7475381537133665,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.3775,
      "step": 8754
    },
    {
      "epoch": 2.747852014594531,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.4437,
      "step": 8755
    },
    {
      "epoch": 2.7481658754756952,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.6717,
      "step": 8756
    },
    {
      "epoch": 2.7484797363568596,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 1.7627,
      "step": 8757
    },
    {
      "epoch": 2.7487935972380244,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.5599,
      "step": 8758
    },
    {
      "epoch": 2.7491074581191888,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.6528,
      "step": 8759
    },
    {
      "epoch": 2.749421319000353,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.5964,
      "step": 8760
    },
    {
      "epoch": 2.7497351798815175,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.9533,
      "step": 8761
    },
    {
      "epoch": 2.750049040762682,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 1.6586,
      "step": 8762
    },
    {
      "epoch": 2.7503629016438462,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.8038,
      "step": 8763
    },
    {
      "epoch": 2.7506767625250106,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.8532,
      "step": 8764
    },
    {
      "epoch": 2.7509906234061754,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002,
      "loss": 2.0524,
      "step": 8765
    },
    {
      "epoch": 2.7513044842873398,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002,
      "loss": 2.0539,
      "step": 8766
    },
    {
      "epoch": 2.751618345168504,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.825,
      "step": 8767
    },
    {
      "epoch": 2.7519322060496685,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.5484,
      "step": 8768
    },
    {
      "epoch": 2.752246066930833,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.593,
      "step": 8769
    },
    {
      "epoch": 2.752559927811997,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.2336,
      "step": 8770
    },
    {
      "epoch": 2.7528737886931616,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.4015,
      "step": 8771
    },
    {
      "epoch": 2.7531876495743264,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002,
      "loss": 2.1122,
      "step": 8772
    },
    {
      "epoch": 2.7535015104554907,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.4368,
      "step": 8773
    },
    {
      "epoch": 2.753815371336655,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.3118,
      "step": 8774
    },
    {
      "epoch": 2.7541292322178195,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.2818,
      "step": 8775
    },
    {
      "epoch": 2.754443093098984,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.3451,
      "step": 8776
    },
    {
      "epoch": 2.754756953980148,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 1.3147,
      "step": 8777
    },
    {
      "epoch": 2.7550708148613126,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.4598,
      "step": 8778
    },
    {
      "epoch": 2.7553846757424774,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.2215,
      "step": 8779
    },
    {
      "epoch": 2.7556985366236417,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.394,
      "step": 8780
    },
    {
      "epoch": 2.756012397504806,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.2969,
      "step": 8781
    },
    {
      "epoch": 2.7563262583859705,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.3555,
      "step": 8782
    },
    {
      "epoch": 2.756640119267135,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.3092,
      "step": 8783
    },
    {
      "epoch": 2.756953980148299,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.3651,
      "step": 8784
    },
    {
      "epoch": 2.7572678410294635,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.2268,
      "step": 8785
    },
    {
      "epoch": 2.7575817019106283,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.3241,
      "step": 8786
    },
    {
      "epoch": 2.7578955627917923,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.1267,
      "step": 8787
    },
    {
      "epoch": 2.758209423672957,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.1252,
      "step": 8788
    },
    {
      "epoch": 2.7585232845541214,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.1699,
      "step": 8789
    },
    {
      "epoch": 2.7585232845541214,
      "eval_loss": 1.8641449213027954,
      "eval_runtime": 123.7639,
      "eval_samples_per_second": 8.08,
      "eval_steps_per_second": 8.08,
      "step": 8789
    },
    {
      "epoch": 2.7585232845541214,
      "mmlu_eval_accuracy": 0.4237500740786868,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
      "mmlu_eval_accuracy_college_biology": 0.375,
      "mmlu_eval_accuracy_college_chemistry": 0.0,
      "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": 0.45454545454545453,
      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
      "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
      "mmlu_eval_accuracy_econometrics": 0.25,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.24390243902439024,
      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
      "mmlu_eval_accuracy_global_facts": 0.3,
      "mmlu_eval_accuracy_high_school_biology": 0.25,
      "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": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.46511627906976744,
      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
      "mmlu_eval_accuracy_high_school_physics": 0.47058823529411764,
      "mmlu_eval_accuracy_high_school_psychology": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_statistics": 0.13043478260869565,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.4230769230769231,
      "mmlu_eval_accuracy_human_aging": 0.5217391304347826,
      "mmlu_eval_accuracy_human_sexuality": 0.25,
      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.68,
      "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
      "mmlu_eval_accuracy_miscellaneous": 0.5116279069767442,
      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
      "mmlu_eval_accuracy_moral_scenarios": 0.3,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.35294117647058826,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
      "mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.2777777777777778,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.1843531046775648,
      "step": 8789
    },
    {
      "epoch": 2.758837145435286,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.1949,
      "step": 8790
    },
    {
      "epoch": 2.75915100631645,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.3141,
      "step": 8791
    },
    {
      "epoch": 2.7594648671976145,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.1525,
      "step": 8792
    },
    {
      "epoch": 2.7597787280787793,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.098,
      "step": 8793
    },
    {
      "epoch": 2.7600925889599432,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.1569,
      "step": 8794
    },
    {
      "epoch": 2.760406449841108,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.2112,
      "step": 8795
    },
    {
      "epoch": 2.7607203107222724,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.2274,
      "step": 8796
    },
    {
      "epoch": 2.761034171603437,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 1.118,
      "step": 8797
    },
    {
      "epoch": 2.761348032484601,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.1078,
      "step": 8798
    },
    {
      "epoch": 2.7616618933657655,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.2266,
      "step": 8799
    },
    {
      "epoch": 2.7619757542469303,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.2176,
      "step": 8800
    },
    {
      "epoch": 2.7622896151280942,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.4884,
      "step": 8801
    },
    {
      "epoch": 2.762603476009259,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.5496,
      "step": 8802
    },
    {
      "epoch": 2.7629173368904234,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.5419,
      "step": 8803
    },
    {
      "epoch": 2.7632311977715878,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.4183,
      "step": 8804
    },
    {
      "epoch": 2.763545058652752,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.6199,
      "step": 8805
    },
    {
      "epoch": 2.7638589195339165,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.4599,
      "step": 8806
    },
    {
      "epoch": 2.764172780415081,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.8484,
      "step": 8807
    },
    {
      "epoch": 2.764486641296245,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 1.4786,
      "step": 8808
    },
    {
      "epoch": 2.76480050217741,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.5282,
      "step": 8809
    },
    {
      "epoch": 2.7651143630585744,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.724,
      "step": 8810
    },
    {
      "epoch": 2.7654282239397387,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.8674,
      "step": 8811
    },
    {
      "epoch": 2.765742084820903,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 1.9523,
      "step": 8812
    },
    {
      "epoch": 2.7660559457020675,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0002,
      "loss": 1.8308,
      "step": 8813
    },
    {
      "epoch": 2.766369806583232,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002,
      "loss": 1.7908,
      "step": 8814
    },
    {
      "epoch": 2.766683667464396,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.5607,
      "step": 8815
    },
    {
      "epoch": 2.766997528345561,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.3884,
      "step": 8816
    },
    {
      "epoch": 2.7673113892267254,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 2.0593,
      "step": 8817
    },
    {
      "epoch": 2.7676252501078897,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.4634,
      "step": 8818
    },
    {
      "epoch": 2.767939110989054,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 1.4195,
      "step": 8819
    },
    {
      "epoch": 2.7682529718702185,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.6989,
      "step": 8820
    },
    {
      "epoch": 2.768566832751383,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.6499,
      "step": 8821
    },
    {
      "epoch": 2.768880693632547,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 2.0474,
      "step": 8822
    },
    {
      "epoch": 2.769194554513712,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.6071,
      "step": 8823
    },
    {
      "epoch": 2.7695084153948764,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.2207,
      "step": 8824
    },
    {
      "epoch": 2.7698222762760407,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.3999,
      "step": 8825
    },
    {
      "epoch": 2.770136137157205,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.2489,
      "step": 8826
    },
    {
      "epoch": 2.7704499980383694,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.4141,
      "step": 8827
    },
    {
      "epoch": 2.770763858919534,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.218,
      "step": 8828
    },
    {
      "epoch": 2.771077719800698,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.3598,
      "step": 8829
    },
    {
      "epoch": 2.771391580681863,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.2875,
      "step": 8830
    },
    {
      "epoch": 2.7717054415630273,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.2131,
      "step": 8831
    },
    {
      "epoch": 2.7720193024441917,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.3535,
      "step": 8832
    },
    {
      "epoch": 2.772333163325356,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.33,
      "step": 8833
    },
    {
      "epoch": 2.7726470242065204,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.4125,
      "step": 8834
    },
    {
      "epoch": 2.772960885087685,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.3441,
      "step": 8835
    },
    {
      "epoch": 2.773274745968849,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.2537,
      "step": 8836
    },
    {
      "epoch": 2.773588606850014,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.1943,
      "step": 8837
    },
    {
      "epoch": 2.773902467731178,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.0238,
      "step": 8838
    },
    {
      "epoch": 2.7742163286123427,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.2097,
      "step": 8839
    },
    {
      "epoch": 2.774530189493507,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.0903,
      "step": 8840
    },
    {
      "epoch": 2.7748440503746714,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.2512,
      "step": 8841
    },
    {
      "epoch": 2.7751579112558358,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.2876,
      "step": 8842
    },
    {
      "epoch": 2.775471772137,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.1895,
      "step": 8843
    },
    {
      "epoch": 2.775785633018165,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.2201,
      "step": 8844
    },
    {
      "epoch": 2.776099493899329,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.5158,
      "step": 8845
    },
    {
      "epoch": 2.7764133547804937,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.1696,
      "step": 8846
    },
    {
      "epoch": 2.776727215661658,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.1493,
      "step": 8847
    },
    {
      "epoch": 2.7770410765428224,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 1.2045,
      "step": 8848
    },
    {
      "epoch": 2.7773549374239868,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.1858,
      "step": 8849
    },
    {
      "epoch": 2.777668798305151,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.25,
      "step": 8850
    },
    {
      "epoch": 2.777982659186316,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 1.3839,
      "step": 8851
    },
    {
      "epoch": 2.77829652006748,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.2745,
      "step": 8852
    },
    {
      "epoch": 2.7786103809486447,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.6005,
      "step": 8853
    },
    {
      "epoch": 2.778924241829809,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.3623,
      "step": 8854
    },
    {
      "epoch": 2.7792381027109734,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.5142,
      "step": 8855
    },
    {
      "epoch": 2.7795519635921377,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.6793,
      "step": 8856
    },
    {
      "epoch": 2.779865824473302,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.5301,
      "step": 8857
    },
    {
      "epoch": 2.7801796853544665,
      "grad_norm": 1.875,
      "learning_rate": 0.0002,
      "loss": 1.7962,
      "step": 8858
    },
    {
      "epoch": 2.780493546235631,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002,
      "loss": 1.4824,
      "step": 8859
    },
    {
      "epoch": 2.7808074071167956,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.8273,
      "step": 8860
    },
    {
      "epoch": 2.78112126799796,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.6435,
      "step": 8861
    },
    {
      "epoch": 2.7814351288791244,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 1.5797,
      "step": 8862
    },
    {
      "epoch": 2.7817489897602887,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 2.2116,
      "step": 8863
    },
    {
      "epoch": 2.782062850641453,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.72,
      "step": 8864
    },
    {
      "epoch": 2.7823767115226175,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002,
      "loss": 1.8395,
      "step": 8865
    },
    {
      "epoch": 2.782690572403782,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002,
      "loss": 1.8089,
      "step": 8866
    },
    {
      "epoch": 2.7830044332849466,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.4679,
      "step": 8867
    },
    {
      "epoch": 2.783318294166111,
      "grad_norm": 0.75,
      "learning_rate": 0.0002,
      "loss": 1.5088,
      "step": 8868
    },
    {
      "epoch": 2.7836321550472753,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 1.5487,
      "step": 8869
    },
    {
      "epoch": 2.7839460159284397,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0002,
      "loss": 1.7214,
      "step": 8870
    },
    {
      "epoch": 2.784259876809604,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002,
      "loss": 1.8389,
      "step": 8871
    },
    {
      "epoch": 2.7845737376907684,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002,
      "loss": 2.1598,
      "step": 8872
    },
    {
      "epoch": 2.784887598571933,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.6685,
      "step": 8873
    },
    {
      "epoch": 2.7852014594530976,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.3307,
      "step": 8874
    },
    {
      "epoch": 2.785515320334262,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.4261,
      "step": 8875
    },
    {
      "epoch": 2.7858291812154263,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.2749,
      "step": 8876
    },
    {
      "epoch": 2.7861430420965907,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 1.3086,
      "step": 8877
    },
    {
      "epoch": 2.786456902977755,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.238,
      "step": 8878
    },
    {
      "epoch": 2.7867707638589194,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.2043,
      "step": 8879
    },
    {
      "epoch": 2.787084624740084,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0002,
      "loss": 1.2158,
      "step": 8880
    },
    {
      "epoch": 2.7873984856212486,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.2323,
      "step": 8881
    },
    {
      "epoch": 2.787712346502413,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.3248,
      "step": 8882
    },
    {
      "epoch": 2.7880262073835773,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.2562,
      "step": 8883
    },
    {
      "epoch": 2.7883400682647417,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.2253,
      "step": 8884
    },
    {
      "epoch": 2.788653929145906,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.2782,
      "step": 8885
    },
    {
      "epoch": 2.7889677900270704,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.1852,
      "step": 8886
    },
    {
      "epoch": 2.7892816509082348,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.1523,
      "step": 8887
    },
    {
      "epoch": 2.7895955117893996,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.1112,
      "step": 8888
    },
    {
      "epoch": 2.789909372670564,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.2292,
      "step": 8889
    },
    {
      "epoch": 2.7902232335517283,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.3274,
      "step": 8890
    },
    {
      "epoch": 2.7905370944328927,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.0913,
      "step": 8891
    },
    {
      "epoch": 2.790850955314057,
      "grad_norm": 0.3125,
      "learning_rate": 0.0002,
      "loss": 1.1233,
      "step": 8892
    },
    {
      "epoch": 2.7911648161952214,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.0499,
      "step": 8893
    },
    {
      "epoch": 2.7914786770763858,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.2103,
      "step": 8894
    },
    {
      "epoch": 2.7917925379575506,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 1.2019,
      "step": 8895
    },
    {
      "epoch": 2.7921063988387145,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 0.9751,
      "step": 8896
    },
    {
      "epoch": 2.7924202597198793,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.1311,
      "step": 8897
    },
    {
      "epoch": 2.7927341206010436,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.3694,
      "step": 8898
    },
    {
      "epoch": 2.793047981482208,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 1.3614,
      "step": 8899
    },
    {
      "epoch": 2.7933618423633724,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.3947,
      "step": 8900
    },
    {
      "epoch": 2.7936757032445367,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.2302,
      "step": 8901
    },
    {
      "epoch": 2.7939895641257015,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.2091,
      "step": 8902
    },
    {
      "epoch": 2.7943034250068655,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.5978,
      "step": 8903
    },
    {
      "epoch": 2.7946172858880303,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.4075,
      "step": 8904
    },
    {
      "epoch": 2.7949311467691946,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.4637,
      "step": 8905
    },
    {
      "epoch": 2.795245007650359,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 1.5931,
      "step": 8906
    },
    {
      "epoch": 2.7955588685315234,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002,
      "loss": 1.6932,
      "step": 8907
    },
    {
      "epoch": 2.7958727294126877,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0002,
      "loss": 1.3757,
      "step": 8908
    },
    {
      "epoch": 2.7961865902938525,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.7842,
      "step": 8909
    },
    {
      "epoch": 2.7965004511750164,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.8431,
      "step": 8910
    },
    {
      "epoch": 2.7968143120561813,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.7284,
      "step": 8911
    },
    {
      "epoch": 2.7971281729373456,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.5956,
      "step": 8912
    },
    {
      "epoch": 2.79744203381851,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.3859,
      "step": 8913
    },
    {
      "epoch": 2.7977558946996743,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 1.7986,
      "step": 8914
    },
    {
      "epoch": 2.7980697555808387,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.4169,
      "step": 8915
    },
    {
      "epoch": 2.798383616462003,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 1.7558,
      "step": 8916
    },
    {
      "epoch": 2.7986974773431674,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.8279,
      "step": 8917
    },
    {
      "epoch": 2.7990113382243322,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.501,
      "step": 8918
    },
    {
      "epoch": 2.7993251991054966,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.3465,
      "step": 8919
    },
    {
      "epoch": 2.799639059986661,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.6286,
      "step": 8920
    },
    {
      "epoch": 2.7999529208678253,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.461,
      "step": 8921
    },
    {
      "epoch": 2.8002667817489897,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002,
      "loss": 2.1564,
      "step": 8922
    },
    {
      "epoch": 2.800580642630154,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.6156,
      "step": 8923
    },
    {
      "epoch": 2.8008945035113184,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.3794,
      "step": 8924
    },
    {
      "epoch": 2.801208364392483,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.2563,
      "step": 8925
    },
    {
      "epoch": 2.8015222252736476,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.3096,
      "step": 8926
    },
    {
      "epoch": 2.801836086154812,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.3823,
      "step": 8927
    },
    {
      "epoch": 2.8021499470359763,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.2788,
      "step": 8928
    },
    {
      "epoch": 2.8024638079171407,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.2653,
      "step": 8929
    },
    {
      "epoch": 2.802777668798305,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.219,
      "step": 8930
    },
    {
      "epoch": 2.8030915296794694,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.3238,
      "step": 8931
    },
    {
      "epoch": 2.803405390560634,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.2924,
      "step": 8932
    },
    {
      "epoch": 2.8037192514417986,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.1432,
      "step": 8933
    },
    {
      "epoch": 2.804033112322963,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.1038,
      "step": 8934
    },
    {
      "epoch": 2.8043469732041273,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.2166,
      "step": 8935
    },
    {
      "epoch": 2.8046608340852917,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.364,
      "step": 8936
    },
    {
      "epoch": 2.804974694966456,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2222,
      "step": 8937
    },
    {
      "epoch": 2.8052885558476204,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.197,
      "step": 8938
    },
    {
      "epoch": 2.805602416728785,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.0971,
      "step": 8939
    },
    {
      "epoch": 2.8059162776099495,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.1964,
      "step": 8940
    },
    {
      "epoch": 2.806230138491114,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.2351,
      "step": 8941
    },
    {
      "epoch": 2.8065439993722783,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.2927,
      "step": 8942
    },
    {
      "epoch": 2.8068578602534426,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.431,
      "step": 8943
    },
    {
      "epoch": 2.807171721134607,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.2947,
      "step": 8944
    },
    {
      "epoch": 2.8074855820157714,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.2102,
      "step": 8945
    },
    {
      "epoch": 2.807799442896936,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.2571,
      "step": 8946
    },
    {
      "epoch": 2.8081133037781,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.1792,
      "step": 8947
    },
    {
      "epoch": 2.808427164659265,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.0224,
      "step": 8948
    },
    {
      "epoch": 2.8087410255404293,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 1.2309,
      "step": 8949
    },
    {
      "epoch": 2.8090548864215936,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.2926,
      "step": 8950
    },
    {
      "epoch": 2.809368747302758,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.4216,
      "step": 8951
    },
    {
      "epoch": 2.8096826081839223,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002,
      "loss": 1.3783,
      "step": 8952
    },
    {
      "epoch": 2.809996469065087,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.3547,
      "step": 8953
    },
    {
      "epoch": 2.810310329946251,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.4433,
      "step": 8954
    },
    {
      "epoch": 2.810624190827416,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.7898,
      "step": 8955
    },
    {
      "epoch": 2.8109380517085802,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002,
      "loss": 1.6209,
      "step": 8956
    },
    {
      "epoch": 2.8112519125897446,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.4303,
      "step": 8957
    },
    {
      "epoch": 2.811565773470909,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.8104,
      "step": 8958
    },
    {
      "epoch": 2.8118796343520733,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.4039,
      "step": 8959
    },
    {
      "epoch": 2.812193495233238,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.4615,
      "step": 8960
    },
    {
      "epoch": 2.812507356114402,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002,
      "loss": 1.9492,
      "step": 8961
    },
    {
      "epoch": 2.812821216995567,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0002,
      "loss": 2.0747,
      "step": 8962
    },
    {
      "epoch": 2.8131350778767312,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0002,
      "loss": 1.9825,
      "step": 8963
    },
    {
      "epoch": 2.8134489387578956,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.585,
      "step": 8964
    },
    {
      "epoch": 2.81376279963906,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0002,
      "loss": 1.6743,
      "step": 8965
    },
    {
      "epoch": 2.8140766605202243,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.5009,
      "step": 8966
    },
    {
      "epoch": 2.8143905214013887,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002,
      "loss": 2.0763,
      "step": 8967
    },
    {
      "epoch": 2.814704382282553,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.8759,
      "step": 8968
    },
    {
      "epoch": 2.815018243163718,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.5192,
      "step": 8969
    },
    {
      "epoch": 2.815332104044882,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002,
      "loss": 1.8401,
      "step": 8970
    },
    {
      "epoch": 2.8156459649260466,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002,
      "loss": 1.3803,
      "step": 8971
    },
    {
      "epoch": 2.815959825807211,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002,
      "loss": 2.1544,
      "step": 8972
    },
    {
      "epoch": 2.8162736866883753,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.914,
      "step": 8973
    },
    {
      "epoch": 2.8165875475695397,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.0397,
      "step": 8974
    },
    {
      "epoch": 2.816901408450704,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.2654,
      "step": 8975
    },
    {
      "epoch": 2.817215269331869,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.3141,
      "step": 8976
    },
    {
      "epoch": 2.817215269331869,
      "eval_loss": 1.8765534162521362,
      "eval_runtime": 123.8379,
      "eval_samples_per_second": 8.075,
      "eval_steps_per_second": 8.075,
      "step": 8976
    },
    {
      "epoch": 2.817215269331869,
      "mmlu_eval_accuracy": 0.43177612834492324,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.2727272727272727,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
      "mmlu_eval_accuracy_college_biology": 0.4375,
      "mmlu_eval_accuracy_college_chemistry": 0.5,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.5,
      "mmlu_eval_accuracy_elementary_mathematics": 0.21951219512195122,
      "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": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.47619047619047616,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5348837209302325,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.5,
      "mmlu_eval_accuracy_high_school_physics": 0.47058823529411764,
      "mmlu_eval_accuracy_high_school_psychology": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.4230769230769231,
      "mmlu_eval_accuracy_human_aging": 0.34782608695652173,
      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.64,
      "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
      "mmlu_eval_accuracy_miscellaneous": 0.5465116279069767,
      "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.48484848484848486,
      "mmlu_eval_accuracy_philosophy": 0.38235294117647056,
      "mmlu_eval_accuracy_prehistory": 0.37142857142857144,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.3,
      "mmlu_eval_accuracy_professional_medicine": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_psychology": 0.36231884057971014,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
      "mmlu_eval_accuracy_virology": 0.3333333333333333,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.1526709688324932,
      "step": 8976
    },
    {
      "epoch": 2.817529130213033,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.3155,
      "step": 8977
    },
    {
      "epoch": 2.8178429910941976,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.3944,
      "step": 8978
    },
    {
      "epoch": 2.818156851975362,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.3358,
      "step": 8979
    },
    {
      "epoch": 2.8184707128565263,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.3133,
      "step": 8980
    },
    {
      "epoch": 2.8187845737376906,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.1682,
      "step": 8981
    },
    {
      "epoch": 2.819098434618855,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.3518,
      "step": 8982
    },
    {
      "epoch": 2.81941229550002,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.2919,
      "step": 8983
    },
    {
      "epoch": 2.819726156381184,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.112,
      "step": 8984
    },
    {
      "epoch": 2.8200400172623485,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.3675,
      "step": 8985
    },
    {
      "epoch": 2.820353878143513,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.2422,
      "step": 8986
    },
    {
      "epoch": 2.8206677390246773,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.2066,
      "step": 8987
    },
    {
      "epoch": 2.8209815999058416,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.2395,
      "step": 8988
    },
    {
      "epoch": 2.821295460787006,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.2021,
      "step": 8989
    },
    {
      "epoch": 2.821609321668171,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.3239,
      "step": 8990
    },
    {
      "epoch": 2.821923182549335,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.3449,
      "step": 8991
    },
    {
      "epoch": 2.8222370434304995,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.0573,
      "step": 8992
    },
    {
      "epoch": 2.822550904311664,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.0472,
      "step": 8993
    },
    {
      "epoch": 2.8228647651928283,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 1.1821,
      "step": 8994
    },
    {
      "epoch": 2.8231786260739926,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.057,
      "step": 8995
    },
    {
      "epoch": 2.823492486955157,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.2382,
      "step": 8996
    },
    {
      "epoch": 2.823806347836322,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.3752,
      "step": 8997
    },
    {
      "epoch": 2.8241202087174857,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.4838,
      "step": 8998
    },
    {
      "epoch": 2.8244340695986505,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 1.191,
      "step": 8999
    },
    {
      "epoch": 2.824747930479815,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 1.3935,
      "step": 9000
    },
    {
      "epoch": 2.8250617913609792,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.6454,
      "step": 9001
    },
    {
      "epoch": 2.8253756522421436,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.2102,
      "step": 9002
    },
    {
      "epoch": 2.825689513123308,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.2245,
      "step": 9003
    },
    {
      "epoch": 2.8260033740044728,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.2379,
      "step": 9004
    },
    {
      "epoch": 2.8263172348856367,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.4051,
      "step": 9005
    },
    {
      "epoch": 2.8266310957668015,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.8193,
      "step": 9006
    },
    {
      "epoch": 2.826944956647966,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.6175,
      "step": 9007
    },
    {
      "epoch": 2.82725881752913,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002,
      "loss": 1.7663,
      "step": 9008
    },
    {
      "epoch": 2.8275726784102946,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002,
      "loss": 1.7074,
      "step": 9009
    },
    {
      "epoch": 2.827886539291459,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.7753,
      "step": 9010
    },
    {
      "epoch": 2.8282004001726238,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.559,
      "step": 9011
    },
    {
      "epoch": 2.8285142610537877,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002,
      "loss": 1.8392,
      "step": 9012
    },
    {
      "epoch": 2.8288281219349525,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.9288,
      "step": 9013
    },
    {
      "epoch": 2.829141982816117,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.8832,
      "step": 9014
    },
    {
      "epoch": 2.829455843697281,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.6077,
      "step": 9015
    },
    {
      "epoch": 2.8297697045784456,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 2.1331,
      "step": 9016
    },
    {
      "epoch": 2.83008356545961,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.5486,
      "step": 9017
    },
    {
      "epoch": 2.8303974263407743,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.58,
      "step": 9018
    },
    {
      "epoch": 2.8307112872219387,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.4177,
      "step": 9019
    },
    {
      "epoch": 2.8310251481031035,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.5542,
      "step": 9020
    },
    {
      "epoch": 2.831339008984268,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.5834,
      "step": 9021
    },
    {
      "epoch": 2.831652869865432,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002,
      "loss": 2.646,
      "step": 9022
    },
    {
      "epoch": 2.8319667307465966,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 1.8011,
      "step": 9023
    },
    {
      "epoch": 2.832280591627761,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.2585,
      "step": 9024
    },
    {
      "epoch": 2.8325944525089253,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.2951,
      "step": 9025
    },
    {
      "epoch": 2.8329083133900896,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.3611,
      "step": 9026
    },
    {
      "epoch": 2.8332221742712544,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 1.4556,
      "step": 9027
    },
    {
      "epoch": 2.833536035152419,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.2796,
      "step": 9028
    },
    {
      "epoch": 2.833849896033583,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.2206,
      "step": 9029
    },
    {
      "epoch": 2.8341637569147475,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.2956,
      "step": 9030
    },
    {
      "epoch": 2.834477617795912,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2332,
      "step": 9031
    },
    {
      "epoch": 2.8347914786770763,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.3109,
      "step": 9032
    },
    {
      "epoch": 2.8351053395582406,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2161,
      "step": 9033
    },
    {
      "epoch": 2.8354192004394054,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.3615,
      "step": 9034
    },
    {
      "epoch": 2.83573306132057,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.3017,
      "step": 9035
    },
    {
      "epoch": 2.836046922201734,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.167,
      "step": 9036
    },
    {
      "epoch": 2.8363607830828985,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.2524,
      "step": 9037
    },
    {
      "epoch": 2.836674643964063,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.1429,
      "step": 9038
    },
    {
      "epoch": 2.8369885048452272,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.3353,
      "step": 9039
    },
    {
      "epoch": 2.8373023657263916,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0002,
      "loss": 1.1741,
      "step": 9040
    },
    {
      "epoch": 2.8376162266075564,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.2731,
      "step": 9041
    },
    {
      "epoch": 2.8379300874887208,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.155,
      "step": 9042
    },
    {
      "epoch": 2.838243948369885,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.2684,
      "step": 9043
    },
    {
      "epoch": 2.8385578092510495,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.2094,
      "step": 9044
    },
    {
      "epoch": 2.838871670132214,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0002,
      "loss": 1.1623,
      "step": 9045
    },
    {
      "epoch": 2.8391855310133782,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.2455,
      "step": 9046
    },
    {
      "epoch": 2.8394993918945426,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.1499,
      "step": 9047
    },
    {
      "epoch": 2.8398132527757074,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.3241,
      "step": 9048
    },
    {
      "epoch": 2.8401271136568713,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 1.2392,
      "step": 9049
    },
    {
      "epoch": 2.840440974538036,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.347,
      "step": 9050
    },
    {
      "epoch": 2.8407548354192005,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 1.3251,
      "step": 9051
    },
    {
      "epoch": 2.841068696300365,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.0969,
      "step": 9052
    },
    {
      "epoch": 2.841382557181529,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.49,
      "step": 9053
    },
    {
      "epoch": 2.8416964180626936,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.3396,
      "step": 9054
    },
    {
      "epoch": 2.8420102789438584,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.5866,
      "step": 9055
    },
    {
      "epoch": 2.8423241398250223,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.6271,
      "step": 9056
    },
    {
      "epoch": 2.842638000706187,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.3321,
      "step": 9057
    },
    {
      "epoch": 2.8429518615873515,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.5362,
      "step": 9058
    },
    {
      "epoch": 2.843265722468516,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.414,
      "step": 9059
    },
    {
      "epoch": 2.84357958334968,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.3834,
      "step": 9060
    },
    {
      "epoch": 2.8438934442308446,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.6058,
      "step": 9061
    },
    {
      "epoch": 2.8442073051120094,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.3741,
      "step": 9062
    },
    {
      "epoch": 2.8445211659931733,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 2.5764,
      "step": 9063
    },
    {
      "epoch": 2.844835026874338,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 1.4562,
      "step": 9064
    },
    {
      "epoch": 2.8451488877555025,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.5226,
      "step": 9065
    },
    {
      "epoch": 2.845462748636667,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.4985,
      "step": 9066
    },
    {
      "epoch": 2.845776609517831,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.6727,
      "step": 9067
    },
    {
      "epoch": 2.8460904703989955,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.2927,
      "step": 9068
    },
    {
      "epoch": 2.84640433128016,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002,
      "loss": 1.5055,
      "step": 9069
    },
    {
      "epoch": 2.8467181921613243,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.4934,
      "step": 9070
    },
    {
      "epoch": 2.847032053042489,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002,
      "loss": 1.565,
      "step": 9071
    },
    {
      "epoch": 2.8473459139236534,
      "grad_norm": 1.125,
      "learning_rate": 0.0002,
      "loss": 2.0119,
      "step": 9072
    },
    {
      "epoch": 2.847659774804818,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.6404,
      "step": 9073
    },
    {
      "epoch": 2.847973635685982,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002,
      "loss": 1.498,
      "step": 9074
    },
    {
      "epoch": 2.8482874965671465,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.295,
      "step": 9075
    },
    {
      "epoch": 2.848601357448311,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.448,
      "step": 9076
    },
    {
      "epoch": 2.8489152183294753,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.2643,
      "step": 9077
    },
    {
      "epoch": 2.84922907921064,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.1701,
      "step": 9078
    },
    {
      "epoch": 2.8495429400918044,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.3511,
      "step": 9079
    },
    {
      "epoch": 2.849856800972969,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.2698,
      "step": 9080
    },
    {
      "epoch": 2.850170661854133,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.32,
      "step": 9081
    },
    {
      "epoch": 2.8504845227352975,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.1967,
      "step": 9082
    },
    {
      "epoch": 2.850798383616462,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.2424,
      "step": 9083
    },
    {
      "epoch": 2.8511122444976262,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2262,
      "step": 9084
    },
    {
      "epoch": 2.851426105378791,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.2606,
      "step": 9085
    },
    {
      "epoch": 2.8517399662599554,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.2633,
      "step": 9086
    },
    {
      "epoch": 2.8520538271411198,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.1226,
      "step": 9087
    },
    {
      "epoch": 2.852367688022284,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.1934,
      "step": 9088
    },
    {
      "epoch": 2.8526815489034485,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.3545,
      "step": 9089
    },
    {
      "epoch": 2.852995409784613,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 1.2261,
      "step": 9090
    },
    {
      "epoch": 2.853309270665777,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.1806,
      "step": 9091
    },
    {
      "epoch": 2.853623131546942,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0002,
      "loss": 1.3735,
      "step": 9092
    },
    {
      "epoch": 2.8539369924281064,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.224,
      "step": 9093
    },
    {
      "epoch": 2.8542508533092708,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.2,
      "step": 9094
    },
    {
      "epoch": 2.854564714190435,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.1722,
      "step": 9095
    },
    {
      "epoch": 2.8548785750715995,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.3908,
      "step": 9096
    },
    {
      "epoch": 2.855192435952764,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.1836,
      "step": 9097
    },
    {
      "epoch": 2.855506296833928,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 1.1041,
      "step": 9098
    },
    {
      "epoch": 2.855820157715093,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.4832,
      "step": 9099
    },
    {
      "epoch": 2.8561340185962574,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0002,
      "loss": 1.6583,
      "step": 9100
    },
    {
      "epoch": 2.8564478794774217,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.3819,
      "step": 9101
    },
    {
      "epoch": 2.856761740358586,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.5399,
      "step": 9102
    },
    {
      "epoch": 2.8570756012397505,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.4292,
      "step": 9103
    },
    {
      "epoch": 2.857389462120915,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.4899,
      "step": 9104
    },
    {
      "epoch": 2.857703323002079,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.6549,
      "step": 9105
    },
    {
      "epoch": 2.858017183883244,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.6108,
      "step": 9106
    },
    {
      "epoch": 2.858331044764408,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.6387,
      "step": 9107
    },
    {
      "epoch": 2.8586449056455727,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.6155,
      "step": 9108
    },
    {
      "epoch": 2.858958766526737,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.5385,
      "step": 9109
    },
    {
      "epoch": 2.8592726274079014,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 1.5619,
      "step": 9110
    },
    {
      "epoch": 2.859586488289066,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.6163,
      "step": 9111
    },
    {
      "epoch": 2.85990034917023,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002,
      "loss": 1.7083,
      "step": 9112
    },
    {
      "epoch": 2.860214210051395,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 2.0019,
      "step": 9113
    },
    {
      "epoch": 2.860528070932559,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 1.7176,
      "step": 9114
    },
    {
      "epoch": 2.8608419318137237,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.7489,
      "step": 9115
    },
    {
      "epoch": 2.861155792694888,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.9442,
      "step": 9116
    },
    {
      "epoch": 2.8614696535760524,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 1.7019,
      "step": 9117
    },
    {
      "epoch": 2.861783514457217,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002,
      "loss": 1.6234,
      "step": 9118
    },
    {
      "epoch": 2.862097375338381,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 1.3765,
      "step": 9119
    },
    {
      "epoch": 2.862411236219546,
      "grad_norm": 0.84375,
      "learning_rate": 0.0002,
      "loss": 1.532,
      "step": 9120
    },
    {
      "epoch": 2.86272509710071,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.594,
      "step": 9121
    },
    {
      "epoch": 2.8630389579818747,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002,
      "loss": 2.1353,
      "step": 9122
    },
    {
      "epoch": 2.863352818863039,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002,
      "loss": 1.7767,
      "step": 9123
    },
    {
      "epoch": 2.8636666797442034,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.2067,
      "step": 9124
    },
    {
      "epoch": 2.8639805406253678,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.2977,
      "step": 9125
    },
    {
      "epoch": 2.864294401506532,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.3069,
      "step": 9126
    },
    {
      "epoch": 2.8646082623876965,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.4639,
      "step": 9127
    },
    {
      "epoch": 2.864922123268861,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.1895,
      "step": 9128
    },
    {
      "epoch": 2.8652359841500257,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.3323,
      "step": 9129
    },
    {
      "epoch": 2.86554984503119,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 1.2823,
      "step": 9130
    },
    {
      "epoch": 2.8658637059123544,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.2297,
      "step": 9131
    },
    {
      "epoch": 2.8661775667935188,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.3366,
      "step": 9132
    },
    {
      "epoch": 2.866491427674683,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.3351,
      "step": 9133
    },
    {
      "epoch": 2.8668052885558475,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.2682,
      "step": 9134
    },
    {
      "epoch": 2.867119149437012,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 1.3796,
      "step": 9135
    },
    {
      "epoch": 2.8674330103181767,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.1383,
      "step": 9136
    },
    {
      "epoch": 2.867746871199341,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.3909,
      "step": 9137
    },
    {
      "epoch": 2.8680607320805054,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2909,
      "step": 9138
    },
    {
      "epoch": 2.8683745929616697,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.2686,
      "step": 9139
    },
    {
      "epoch": 2.868688453842834,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.1662,
      "step": 9140
    },
    {
      "epoch": 2.8690023147239985,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.0779,
      "step": 9141
    },
    {
      "epoch": 2.869316175605163,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0002,
      "loss": 1.2465,
      "step": 9142
    },
    {
      "epoch": 2.8696300364863276,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 1.134,
      "step": 9143
    },
    {
      "epoch": 2.869943897367492,
      "grad_norm": 0.34375,
      "learning_rate": 0.0002,
      "loss": 1.1521,
      "step": 9144
    },
    {
      "epoch": 2.8702577582486564,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.073,
      "step": 9145
    },
    {
      "epoch": 2.8705716191298207,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.252,
      "step": 9146
    },
    {
      "epoch": 2.870885480010985,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.1757,
      "step": 9147
    },
    {
      "epoch": 2.8711993408921495,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002,
      "loss": 1.2349,
      "step": 9148
    },
    {
      "epoch": 2.871513201773314,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.1728,
      "step": 9149
    },
    {
      "epoch": 2.8718270626544786,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.3679,
      "step": 9150
    },
    {
      "epoch": 2.872140923535643,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.4438,
      "step": 9151
    },
    {
      "epoch": 2.8724547844168073,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.2971,
      "step": 9152
    },
    {
      "epoch": 2.8727686452979717,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.3071,
      "step": 9153
    },
    {
      "epoch": 2.873082506179136,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.401,
      "step": 9154
    },
    {
      "epoch": 2.8733963670603004,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.4128,
      "step": 9155
    },
    {
      "epoch": 2.873710227941465,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.833,
      "step": 9156
    },
    {
      "epoch": 2.8740240888226296,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.4724,
      "step": 9157
    },
    {
      "epoch": 2.8743379497037935,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.3588,
      "step": 9158
    },
    {
      "epoch": 2.8746518105849583,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002,
      "loss": 1.8677,
      "step": 9159
    },
    {
      "epoch": 2.8749656714661227,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.2482,
      "step": 9160
    },
    {
      "epoch": 2.875279532347287,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002,
      "loss": 1.9113,
      "step": 9161
    },
    {
      "epoch": 2.8755933932284514,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0002,
      "loss": 1.7697,
      "step": 9162
    },
    {
      "epoch": 2.875907254109616,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.7191,
      "step": 9163
    },
    {
      "epoch": 2.875907254109616,
      "eval_loss": 1.8761990070343018,
      "eval_runtime": 123.7901,
      "eval_samples_per_second": 8.078,
      "eval_steps_per_second": 8.078,
      "step": 9163
    },
    {
      "epoch": 2.875907254109616,
      "mmlu_eval_accuracy": 0.4258227348042516,
      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.375,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
      "mmlu_eval_accuracy_college_biology": 0.3125,
      "mmlu_eval_accuracy_college_chemistry": 0.125,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
      "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.4375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.21951219512195122,
      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
      "mmlu_eval_accuracy_global_facts": 0.5,
      "mmlu_eval_accuracy_high_school_biology": 0.25,
      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.47058823529411764,
      "mmlu_eval_accuracy_high_school_psychology": 0.6333333333333333,
      "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
      "mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.5217391304347826,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
      "mmlu_eval_accuracy_management": 0.36363636363636365,
      "mmlu_eval_accuracy_marketing": 0.6,
      "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
      "mmlu_eval_accuracy_miscellaneous": 0.5116279069767442,
      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
      "mmlu_eval_accuracy_moral_scenarios": 0.31,
      "mmlu_eval_accuracy_nutrition": 0.48484848484848486,
      "mmlu_eval_accuracy_philosophy": 0.35294117647058826,
      "mmlu_eval_accuracy_prehistory": 0.2857142857142857,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.3225806451612903,
      "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.5454545454545454,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.6842105263157895,
      "mmlu_loss": 1.1657283430703709,
      "step": 9163
    },
    {
      "epoch": 2.8762211149907806,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.4422,
      "step": 9164
    },
    {
      "epoch": 2.8765349758719445,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 1.641,
      "step": 9165
    },
    {
      "epoch": 2.8768488367531093,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.5573,
      "step": 9166
    },
    {
      "epoch": 2.8771626976342737,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 1.5835,
      "step": 9167
    },
    {
      "epoch": 2.877476558515438,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.7923,
      "step": 9168
    },
    {
      "epoch": 2.8777904193966024,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.3367,
      "step": 9169
    },
    {
      "epoch": 2.8781042802777668,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.5132,
      "step": 9170
    },
    {
      "epoch": 2.8784181411589316,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002,
      "loss": 1.8228,
      "step": 9171
    },
    {
      "epoch": 2.8787320020400955,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002,
      "loss": 2.0626,
      "step": 9172
    },
    {
      "epoch": 2.8790458629212603,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 1.5042,
      "step": 9173
    },
    {
      "epoch": 2.8793597238024247,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.2516,
      "step": 9174
    },
    {
      "epoch": 2.879673584683589,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.3063,
      "step": 9175
    },
    {
      "epoch": 2.8799874455647534,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.4874,
      "step": 9176
    },
    {
      "epoch": 2.8803013064459178,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.2835,
      "step": 9177
    },
    {
      "epoch": 2.880615167327082,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.2245,
      "step": 9178
    },
    {
      "epoch": 2.8809290282082465,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 1.3496,
      "step": 9179
    },
    {
      "epoch": 2.8812428890894113,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.1062,
      "step": 9180
    },
    {
      "epoch": 2.8815567499705756,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.4002,
      "step": 9181
    },
    {
      "epoch": 2.88187061085174,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 1.2155,
      "step": 9182
    },
    {
      "epoch": 2.8821844717329044,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.1771,
      "step": 9183
    },
    {
      "epoch": 2.8824983326140687,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.3097,
      "step": 9184
    },
    {
      "epoch": 2.882812193495233,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.184,
      "step": 9185
    },
    {
      "epoch": 2.8831260543763975,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.2425,
      "step": 9186
    },
    {
      "epoch": 2.8834399152575623,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.2401,
      "step": 9187
    },
    {
      "epoch": 2.8837537761387266,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.1407,
      "step": 9188
    },
    {
      "epoch": 2.884067637019891,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.2738,
      "step": 9189
    },
    {
      "epoch": 2.8843814979010554,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 1.2728,
      "step": 9190
    },
    {
      "epoch": 2.8846953587822197,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.2418,
      "step": 9191
    },
    {
      "epoch": 2.885009219663384,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.2583,
      "step": 9192
    },
    {
      "epoch": 2.8853230805445484,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 1.2751,
      "step": 9193
    },
    {
      "epoch": 2.8856369414257133,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.1985,
      "step": 9194
    },
    {
      "epoch": 2.8859508023068776,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.2711,
      "step": 9195
    },
    {
      "epoch": 2.886264663188042,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.1077,
      "step": 9196
    },
    {
      "epoch": 2.8865785240692063,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.3773,
      "step": 9197
    },
    {
      "epoch": 2.8868923849503707,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.1897,
      "step": 9198
    },
    {
      "epoch": 2.887206245831535,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 1.1688,
      "step": 9199
    },
    {
      "epoch": 2.8875201067126994,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 1.5106,
      "step": 9200
    },
    {
      "epoch": 2.8878339675938642,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.3914,
      "step": 9201
    },
    {
      "epoch": 2.8881478284750286,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.5654,
      "step": 9202
    },
    {
      "epoch": 2.888461689356193,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 1.4306,
      "step": 9203
    },
    {
      "epoch": 2.8887755502373573,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 1.4539,
      "step": 9204
    },
    {
      "epoch": 2.8890894111185217,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.4266,
      "step": 9205
    },
    {
      "epoch": 2.889403271999686,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.6812,
      "step": 9206
    },
    {
      "epoch": 2.8897171328808504,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.5194,
      "step": 9207
    },
    {
      "epoch": 2.890030993762015,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.8071,
      "step": 9208
    },
    {
      "epoch": 2.890344854643179,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002,
      "loss": 1.7653,
      "step": 9209
    },
    {
      "epoch": 2.890658715524344,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.465,
      "step": 9210
    },
    {
      "epoch": 2.8909725764055083,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.6069,
      "step": 9211
    },
    {
      "epoch": 2.8912864372866727,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002,
      "loss": 2.0836,
      "step": 9212
    },
    {
      "epoch": 2.891600298167837,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 1.7322,
      "step": 9213
    },
    {
      "epoch": 2.8919141590490014,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 2.0488,
      "step": 9214
    },
    {
      "epoch": 2.892228019930166,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 1.703,
      "step": 9215
    },
    {
      "epoch": 2.89254188081133,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 1.9165,
      "step": 9216
    },
    {
      "epoch": 2.892855741692495,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 1.7885,
      "step": 9217
    },
    {
      "epoch": 2.8931696025736593,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.0853,
      "step": 9218
    },
    {
      "epoch": 2.8934834634548237,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.1929,
      "step": 9219
    },
    {
      "epoch": 2.893797324335988,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0002,
      "loss": 1.7236,
      "step": 9220
    },
    {
      "epoch": 2.8941111852171524,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.7137,
      "step": 9221
    },
    {
      "epoch": 2.894425046098317,
      "grad_norm": 1.421875,
      "learning_rate": 0.0002,
      "loss": 2.7347,
      "step": 9222
    },
    {
      "epoch": 2.894738906979481,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.4389,
      "step": 9223
    },
    {
      "epoch": 2.895052767860646,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.2317,
      "step": 9224
    },
    {
      "epoch": 2.8953666287418103,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 1.3827,
      "step": 9225
    },
    {
      "epoch": 2.8956804896229746,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.3125,
      "step": 9226
    },
    {
      "epoch": 2.895994350504139,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 1.2708,
      "step": 9227
    },
    {
      "epoch": 2.8963082113853034,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.3622,
      "step": 9228
    },
    {
      "epoch": 2.8966220722664677,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.2798,
      "step": 9229
    },
    {
      "epoch": 2.896935933147632,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.4568,
      "step": 9230
    },
    {
      "epoch": 2.897249794028797,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.3127,
      "step": 9231
    },
    {
      "epoch": 2.8975636549099613,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.186,
      "step": 9232
    },
    {
      "epoch": 2.8978775157911256,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.044,
      "step": 9233
    },
    {
      "epoch": 2.89819137667229,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.2048,
      "step": 9234
    },
    {
      "epoch": 2.8985052375534544,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.2282,
      "step": 9235
    },
    {
      "epoch": 2.8988190984346187,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.066,
      "step": 9236
    },
    {
      "epoch": 2.899132959315783,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.078,
      "step": 9237
    },
    {
      "epoch": 2.899446820196948,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 1.2239,
      "step": 9238
    },
    {
      "epoch": 2.8997606810781122,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.2316,
      "step": 9239
    },
    {
      "epoch": 2.9000745419592766,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.2616,
      "step": 9240
    },
    {
      "epoch": 2.900388402840441,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 1.1115,
      "step": 9241
    },
    {
      "epoch": 2.9007022637216053,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.1179,
      "step": 9242
    },
    {
      "epoch": 2.9010161246027697,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.1528,
      "step": 9243
    },
    {
      "epoch": 2.901329985483934,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.2991,
      "step": 9244
    },
    {
      "epoch": 2.901643846365099,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.373,
      "step": 9245
    },
    {
      "epoch": 2.9019577072462632,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.2299,
      "step": 9246
    },
    {
      "epoch": 2.9022715681274276,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 1.1942,
      "step": 9247
    },
    {
      "epoch": 2.902585429008592,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.1646,
      "step": 9248
    },
    {
      "epoch": 2.9028992898897563,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.2777,
      "step": 9249
    },
    {
      "epoch": 2.9032131507709207,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.9132,
      "step": 9250
    },
    {
      "epoch": 2.903527011652085,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.4572,
      "step": 9251
    },
    {
      "epoch": 2.90384087253325,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.6071,
      "step": 9252
    },
    {
      "epoch": 2.904154733414414,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.6622,
      "step": 9253
    },
    {
      "epoch": 2.9044685942955786,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 1.5777,
      "step": 9254
    },
    {
      "epoch": 2.904782455176743,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.526,
      "step": 9255
    },
    {
      "epoch": 2.9050963160579073,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.5428,
      "step": 9256
    },
    {
      "epoch": 2.9054101769390717,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 1.6678,
      "step": 9257
    },
    {
      "epoch": 2.905724037820236,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.5418,
      "step": 9258
    },
    {
      "epoch": 2.906037898701401,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.4801,
      "step": 9259
    },
    {
      "epoch": 2.9063517595825648,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.6799,
      "step": 9260
    },
    {
      "epoch": 2.9066656204637296,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002,
      "loss": 1.9311,
      "step": 9261
    },
    {
      "epoch": 2.906979481344894,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.6379,
      "step": 9262
    },
    {
      "epoch": 2.9072933422260583,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 1.7348,
      "step": 9263
    },
    {
      "epoch": 2.9076072031072226,
      "grad_norm": 0.9375,
      "learning_rate": 0.0002,
      "loss": 1.82,
      "step": 9264
    },
    {
      "epoch": 2.907921063988387,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.7209,
      "step": 9265
    },
    {
      "epoch": 2.908234924869552,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 1.7559,
      "step": 9266
    },
    {
      "epoch": 2.9085487857507157,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.6945,
      "step": 9267
    },
    {
      "epoch": 2.9088626466318805,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.5186,
      "step": 9268
    },
    {
      "epoch": 2.909176507513045,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.473,
      "step": 9269
    },
    {
      "epoch": 2.9094903683942093,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.3865,
      "step": 9270
    },
    {
      "epoch": 2.9098042292753736,
      "grad_norm": 0.9375,
      "learning_rate": 0.0002,
      "loss": 1.6226,
      "step": 9271
    },
    {
      "epoch": 2.910118090156538,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0002,
      "loss": 2.4095,
      "step": 9272
    },
    {
      "epoch": 2.910431951037703,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.6056,
      "step": 9273
    },
    {
      "epoch": 2.9107458119188667,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.3538,
      "step": 9274
    },
    {
      "epoch": 2.9110596728000315,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.3232,
      "step": 9275
    },
    {
      "epoch": 2.911373533681196,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.2636,
      "step": 9276
    },
    {
      "epoch": 2.9116873945623603,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 1.3079,
      "step": 9277
    },
    {
      "epoch": 2.9120012554435246,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.5118,
      "step": 9278
    },
    {
      "epoch": 2.912315116324689,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.3031,
      "step": 9279
    },
    {
      "epoch": 2.9126289772058533,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.2654,
      "step": 9280
    },
    {
      "epoch": 2.9129428380870177,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2706,
      "step": 9281
    },
    {
      "epoch": 2.9132566989681825,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 1.1949,
      "step": 9282
    },
    {
      "epoch": 2.913570559849347,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2541,
      "step": 9283
    },
    {
      "epoch": 2.9138844207305112,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.1771,
      "step": 9284
    },
    {
      "epoch": 2.9141982816116756,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 1.1466,
      "step": 9285
    },
    {
      "epoch": 2.91451214249284,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.2282,
      "step": 9286
    },
    {
      "epoch": 2.9148260033740043,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.2609,
      "step": 9287
    },
    {
      "epoch": 2.9151398642551687,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.278,
      "step": 9288
    },
    {
      "epoch": 2.9154537251363335,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.2025,
      "step": 9289
    },
    {
      "epoch": 2.915767586017498,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 1.2289,
      "step": 9290
    },
    {
      "epoch": 2.916081446898662,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.1323,
      "step": 9291
    },
    {
      "epoch": 2.9163953077798266,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.2033,
      "step": 9292
    },
    {
      "epoch": 2.916709168660991,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002,
      "loss": 1.2382,
      "step": 9293
    },
    {
      "epoch": 2.9170230295421553,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.0987,
      "step": 9294
    },
    {
      "epoch": 2.9173368904233197,
      "grad_norm": 0.296875,
      "learning_rate": 0.0002,
      "loss": 1.061,
      "step": 9295
    },
    {
      "epoch": 2.9176507513044845,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.2109,
      "step": 9296
    },
    {
      "epoch": 2.917964612185649,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.183,
      "step": 9297
    },
    {
      "epoch": 2.918278473066813,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.156,
      "step": 9298
    },
    {
      "epoch": 2.9185923339479776,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 1.2748,
      "step": 9299
    },
    {
      "epoch": 2.918906194829142,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002,
      "loss": 1.2532,
      "step": 9300
    },
    {
      "epoch": 2.9192200557103063,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 1.3469,
      "step": 9301
    },
    {
      "epoch": 2.9195339165914707,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 1.2769,
      "step": 9302
    },
    {
      "epoch": 2.9198477774726355,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.285,
      "step": 9303
    },
    {
      "epoch": 2.9201616383538,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.3363,
      "step": 9304
    },
    {
      "epoch": 2.920475499234964,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 1.4428,
      "step": 9305
    },
    {
      "epoch": 2.9207893601161286,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.3417,
      "step": 9306
    },
    {
      "epoch": 2.921103220997293,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.7082,
      "step": 9307
    },
    {
      "epoch": 2.9214170818784573,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 1.8225,
      "step": 9308
    },
    {
      "epoch": 2.9217309427596216,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 1.6316,
      "step": 9309
    },
    {
      "epoch": 2.9220448036407864,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002,
      "loss": 1.712,
      "step": 9310
    },
    {
      "epoch": 2.922358664521951,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.8636,
      "step": 9311
    },
    {
      "epoch": 2.922672525403115,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 1.6578,
      "step": 9312
    },
    {
      "epoch": 2.9229863862842795,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.5501,
      "step": 9313
    },
    {
      "epoch": 2.923300247165444,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.8732,
      "step": 9314
    },
    {
      "epoch": 2.9236141080466083,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 1.702,
      "step": 9315
    },
    {
      "epoch": 2.9239279689277726,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 1.6833,
      "step": 9316
    },
    {
      "epoch": 2.9242418298089374,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.5025,
      "step": 9317
    },
    {
      "epoch": 2.9245556906901014,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002,
      "loss": 1.7514,
      "step": 9318
    },
    {
      "epoch": 2.924869551571266,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.518,
      "step": 9319
    },
    {
      "epoch": 2.9251834124524305,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.5729,
      "step": 9320
    },
    {
      "epoch": 2.925497273333595,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 1.6051,
      "step": 9321
    },
    {
      "epoch": 2.9258111342147592,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 1.9974,
      "step": 9322
    },
    {
      "epoch": 2.9261249950959236,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.5181,
      "step": 9323
    },
    {
      "epoch": 2.9264388559770884,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.2338,
      "step": 9324
    },
    {
      "epoch": 2.9267527168582523,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.3301,
      "step": 9325
    },
    {
      "epoch": 2.927066577739417,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.258,
      "step": 9326
    },
    {
      "epoch": 2.9273804386205815,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.3584,
      "step": 9327
    },
    {
      "epoch": 2.927694299501746,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.4484,
      "step": 9328
    },
    {
      "epoch": 2.9280081603829102,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.153,
      "step": 9329
    },
    {
      "epoch": 2.9283220212640746,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.2141,
      "step": 9330
    },
    {
      "epoch": 2.9286358821452394,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.2206,
      "step": 9331
    },
    {
      "epoch": 2.9289497430264033,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.1699,
      "step": 9332
    },
    {
      "epoch": 2.929263603907568,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.2987,
      "step": 9333
    },
    {
      "epoch": 2.9295774647887325,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.1484,
      "step": 9334
    },
    {
      "epoch": 2.929891325669897,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 1.3435,
      "step": 9335
    },
    {
      "epoch": 2.930205186551061,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.2913,
      "step": 9336
    },
    {
      "epoch": 2.9305190474322256,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.2911,
      "step": 9337
    },
    {
      "epoch": 2.93083290831339,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.1074,
      "step": 9338
    },
    {
      "epoch": 2.9311467691945543,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.1217,
      "step": 9339
    },
    {
      "epoch": 2.931460630075719,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 1.2084,
      "step": 9340
    },
    {
      "epoch": 2.9317744909568835,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.1905,
      "step": 9341
    },
    {
      "epoch": 2.932088351838048,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 1.262,
      "step": 9342
    },
    {
      "epoch": 2.932402212719212,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.0687,
      "step": 9343
    },
    {
      "epoch": 2.9327160736003766,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.0989,
      "step": 9344
    },
    {
      "epoch": 2.933029934481541,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.0,
      "step": 9345
    },
    {
      "epoch": 2.9333437953627053,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.2588,
      "step": 9346
    },
    {
      "epoch": 2.93365765624387,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002,
      "loss": 1.5604,
      "step": 9347
    },
    {
      "epoch": 2.9339715171250345,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 1.2224,
      "step": 9348
    },
    {
      "epoch": 2.934285378006199,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.1948,
      "step": 9349
    },
    {
      "epoch": 2.934599238887363,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.3169,
      "step": 9350
    },
    {
      "epoch": 2.934599238887363,
      "eval_loss": 1.8553611040115356,
      "eval_runtime": 123.1225,
      "eval_samples_per_second": 8.122,
      "eval_steps_per_second": 8.122,
      "step": 9350
    },
    {
      "epoch": 2.934599238887363,
      "mmlu_eval_accuracy": 0.42224076268422145,
      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
      "mmlu_eval_accuracy_astronomy": 0.4375,
      "mmlu_eval_accuracy_business_ethics": 0.36363636363636365,
      "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": 0.36363636363636365,
      "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": 0.2727272727272727,
      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
      "mmlu_eval_accuracy_electrical_engineering": 0.25,
      "mmlu_eval_accuracy_elementary_mathematics": 0.24390243902439024,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.6,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_geography": 0.5454545454545454,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.4883720930232558,
      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
      "mmlu_eval_accuracy_high_school_physics": 0.4117647058823529,
      "mmlu_eval_accuracy_high_school_psychology": 0.6666666666666666,
      "mmlu_eval_accuracy_high_school_statistics": 0.08695652173913043,
      "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
      "mmlu_eval_accuracy_human_aging": 0.5652173913043478,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
      "mmlu_eval_accuracy_logical_fallacies": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.68,
      "mmlu_eval_accuracy_medical_genetics": 0.5454545454545454,
      "mmlu_eval_accuracy_miscellaneous": 0.5232558139534884,
      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
      "mmlu_eval_accuracy_moral_scenarios": 0.31,
      "mmlu_eval_accuracy_nutrition": 0.45454545454545453,
      "mmlu_eval_accuracy_philosophy": 0.4117647058823529,
      "mmlu_eval_accuracy_prehistory": 0.3142857142857143,
      "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
      "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
      "mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
      "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
      "mmlu_eval_accuracy_sociology": 0.45454545454545453,
      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
      "mmlu_eval_accuracy_virology": 0.4444444444444444,
      "mmlu_eval_accuracy_world_religions": 0.631578947368421,
      "mmlu_loss": 1.4160843094644167,
      "step": 9350
    },
    {
      "epoch": 2.9349130997685275,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.2188,
      "step": 9351
    },
    {
      "epoch": 2.935226960649692,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 1.4218,
      "step": 9352
    },
    {
      "epoch": 2.9355408215308563,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.4033,
      "step": 9353
    },
    {
      "epoch": 2.935854682412021,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 1.2627,
      "step": 9354
    },
    {
      "epoch": 2.9361685432931854,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002,
      "loss": 1.6191,
      "step": 9355
    },
    {
      "epoch": 2.93648240417435,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.6385,
      "step": 9356
    },
    {
      "epoch": 2.936796265055514,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002,
      "loss": 1.4936,
      "step": 9357
    },
    {
      "epoch": 2.9371101259366785,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 1.459,
      "step": 9358
    },
    {
      "epoch": 2.937423986817843,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.5441,
      "step": 9359
    },
    {
      "epoch": 2.9377378476990073,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.6919,
      "step": 9360
    },
    {
      "epoch": 2.938051708580172,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.44,
      "step": 9361
    },
    {
      "epoch": 2.9383655694613364,
      "grad_norm": 0.75,
      "learning_rate": 0.0002,
      "loss": 1.5979,
      "step": 9362
    },
    {
      "epoch": 2.938679430342501,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.8018,
      "step": 9363
    },
    {
      "epoch": 2.938993291223665,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002,
      "loss": 1.6639,
      "step": 9364
    },
    {
      "epoch": 2.9393071521048295,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002,
      "loss": 1.8697,
      "step": 9365
    },
    {
      "epoch": 2.939621012985994,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002,
      "loss": 1.6695,
      "step": 9366
    },
    {
      "epoch": 2.9399348738671582,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002,
      "loss": 2.0399,
      "step": 9367
    },
    {
      "epoch": 2.940248734748323,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.5309,
      "step": 9368
    },
    {
      "epoch": 2.940562595629487,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.8864,
      "step": 9369
    },
    {
      "epoch": 2.9408764565106518,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 1.7542,
      "step": 9370
    },
    {
      "epoch": 2.941190317391816,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.7105,
      "step": 9371
    },
    {
      "epoch": 2.9415041782729805,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002,
      "loss": 1.9176,
      "step": 9372
    },
    {
      "epoch": 2.941818039154145,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 1.8492,
      "step": 9373
    },
    {
      "epoch": 2.9421319000353092,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.2629,
      "step": 9374
    },
    {
      "epoch": 2.942445760916474,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 1.2557,
      "step": 9375
    },
    {
      "epoch": 2.942759621797638,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.2919,
      "step": 9376
    },
    {
      "epoch": 2.9430734826788028,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.132,
      "step": 9377
    },
    {
      "epoch": 2.943387343559967,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.4452,
      "step": 9378
    },
    {
      "epoch": 2.9437012044411315,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.1625,
      "step": 9379
    },
    {
      "epoch": 2.944015065322296,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3731,
      "step": 9380
    },
    {
      "epoch": 2.94432892620346,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 1.2291,
      "step": 9381
    },
    {
      "epoch": 2.944642787084625,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 1.3691,
      "step": 9382
    },
    {
      "epoch": 2.944956647965789,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.168,
      "step": 9383
    },
    {
      "epoch": 2.9452705088469537,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.3094,
      "step": 9384
    },
    {
      "epoch": 2.945584369728118,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 1.394,
      "step": 9385
    },
    {
      "epoch": 2.9458982306092825,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.1918,
      "step": 9386
    },
    {
      "epoch": 2.946212091490447,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.165,
      "step": 9387
    },
    {
      "epoch": 2.946525952371611,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.2,
      "step": 9388
    },
    {
      "epoch": 2.9468398132527756,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 1.2335,
      "step": 9389
    },
    {
      "epoch": 2.94715367413394,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0002,
      "loss": 1.1117,
      "step": 9390
    },
    {
      "epoch": 2.9474675350151047,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 1.2898,
      "step": 9391
    },
    {
      "epoch": 2.947781395896269,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.1785,
      "step": 9392
    },
    {
      "epoch": 2.9480952567774334,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0002,
      "loss": 1.4124,
      "step": 9393
    },
    {
      "epoch": 2.948409117658598,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 1.0698,
      "step": 9394
    },
    {
      "epoch": 2.948722978539762,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002,
      "loss": 1.0283,
      "step": 9395
    },
    {
      "epoch": 2.9490368394209265,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.2019,
      "step": 9396
    },
    {
      "epoch": 2.949350700302091,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 1.3432,
      "step": 9397
    },
    {
      "epoch": 2.9496645611832557,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 1.2265,
      "step": 9398
    },
    {
      "epoch": 2.94997842206442,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 1.3992,
      "step": 9399
    },
    {
      "epoch": 2.9502922829455844,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002,
      "loss": 1.4437,
      "step": 9400
    },
    {
      "epoch": 2.950606143826749,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 1.156,
      "step": 9401
    },
    {
      "epoch": 2.950920004707913,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002,
      "loss": 1.3625,
      "step": 9402
    },
    {
      "epoch": 2.9512338655890775,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 1.3376,
      "step": 9403
    },
    {
      "epoch": 2.951547726470242,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.12,
      "step": 9404
    },
    {
      "epoch": 2.9518615873514067,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.3309,
      "step": 9405
    },
    {
      "epoch": 2.952175448232571,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.3755,
      "step": 9406
    },
    {
      "epoch": 2.9524893091137354,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 1.5538,
      "step": 9407
    },
    {
      "epoch": 2.9528031699949,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.2567,
      "step": 9408
    },
    {
      "epoch": 2.953117030876064,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.7727,
      "step": 9409
    },
    {
      "epoch": 2.9534308917572285,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 1.6904,
      "step": 9410
    },
    {
      "epoch": 2.953744752638393,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002,
      "loss": 1.7954,
      "step": 9411
    },
    {
      "epoch": 2.9540586135195577,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002,
      "loss": 1.6599,
      "step": 9412
    },
    {
      "epoch": 2.954372474400722,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 1.9667,
      "step": 9413
    },
    {
      "epoch": 2.9546863352818864,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 1.517,
      "step": 9414
    },
    {
      "epoch": 2.9550001961630508,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.6352,
      "step": 9415
    },
    {
      "epoch": 2.955314057044215,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0002,
      "loss": 1.8427,
      "step": 9416
    },
    {
      "epoch": 2.9556279179253795,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002,
      "loss": 1.7782,
      "step": 9417
    },
    {
      "epoch": 2.955941778806544,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.5864,
      "step": 9418
    },
    {
      "epoch": 2.9562556396877087,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 1.3267,
      "step": 9419
    },
    {
      "epoch": 2.9565695005688726,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002,
      "loss": 1.7013,
      "step": 9420
    },
    {
      "epoch": 2.9568833614500374,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.4806,
      "step": 9421
    },
    {
      "epoch": 2.9571972223312017,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002,
      "loss": 1.9826,
      "step": 9422
    },
    {
      "epoch": 2.957511083212366,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 1.759,
      "step": 9423
    },
    {
      "epoch": 2.9578249440935305,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.2318,
      "step": 9424
    },
    {
      "epoch": 2.958138804974695,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 1.3547,
      "step": 9425
    },
    {
      "epoch": 2.9584526658558596,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 1.2914,
      "step": 9426
    },
    {
      "epoch": 2.9587665267370236,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.3337,
      "step": 9427
    },
    {
      "epoch": 2.9590803876181884,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.377,
      "step": 9428
    },
    {
      "epoch": 2.9593942484993527,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.1936,
      "step": 9429
    },
    {
      "epoch": 2.959708109380517,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 1.3915,
      "step": 9430
    },
    {
      "epoch": 2.9600219702616815,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2573,
      "step": 9431
    },
    {
      "epoch": 2.960335831142846,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.1949,
      "step": 9432
    },
    {
      "epoch": 2.9606496920240106,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.3062,
      "step": 9433
    },
    {
      "epoch": 2.9609635529051745,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.2476,
      "step": 9434
    },
    {
      "epoch": 2.9612774137863394,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.0846,
      "step": 9435
    },
    {
      "epoch": 2.9615912746675037,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.1504,
      "step": 9436
    },
    {
      "epoch": 2.961905135548668,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.1983,
      "step": 9437
    },
    {
      "epoch": 2.9622189964298324,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 1.1937,
      "step": 9438
    },
    {
      "epoch": 2.962532857310997,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.2957,
      "step": 9439
    },
    {
      "epoch": 2.962846718192161,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0002,
      "loss": 1.2687,
      "step": 9440
    },
    {
      "epoch": 2.9631605790733255,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.0766,
      "step": 9441
    },
    {
      "epoch": 2.9634744399544903,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 1.1046,
      "step": 9442
    },
    {
      "epoch": 2.9637883008356547,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 1.1553,
      "step": 9443
    },
    {
      "epoch": 2.964102161716819,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 1.2688,
      "step": 9444
    },
    {
      "epoch": 2.9644160225979834,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.2691,
      "step": 9445
    },
    {
      "epoch": 2.964729883479148,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.2599,
      "step": 9446
    },
    {
      "epoch": 2.965043744360312,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 1.377,
      "step": 9447
    },
    {
      "epoch": 2.9653576052414765,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002,
      "loss": 1.2108,
      "step": 9448
    },
    {
      "epoch": 2.9656714661226413,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0002,
      "loss": 1.2736,
      "step": 9449
    },
    {
      "epoch": 2.9659853270038057,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002,
      "loss": 1.2695,
      "step": 9450
    },
    {
      "epoch": 2.96629918788497,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.2906,
      "step": 9451
    },
    {
      "epoch": 2.9666130487661344,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.5703,
      "step": 9452
    },
    {
      "epoch": 2.9669269096472988,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 1.4633,
      "step": 9453
    },
    {
      "epoch": 2.967240770528463,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.4221,
      "step": 9454
    },
    {
      "epoch": 2.9675546314096275,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002,
      "loss": 1.7478,
      "step": 9455
    },
    {
      "epoch": 2.9678684922907923,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 1.4688,
      "step": 9456
    },
    {
      "epoch": 2.9681823531719567,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.7941,
      "step": 9457
    },
    {
      "epoch": 2.968496214053121,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002,
      "loss": 1.7128,
      "step": 9458
    },
    {
      "epoch": 2.9688100749342854,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002,
      "loss": 1.8195,
      "step": 9459
    },
    {
      "epoch": 2.9691239358154498,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.8326,
      "step": 9460
    },
    {
      "epoch": 2.969437796696614,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002,
      "loss": 1.5712,
      "step": 9461
    },
    {
      "epoch": 2.9697516575777785,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002,
      "loss": 1.4856,
      "step": 9462
    },
    {
      "epoch": 2.9700655184589433,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.9067,
      "step": 9463
    },
    {
      "epoch": 2.9703793793401077,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002,
      "loss": 1.6307,
      "step": 9464
    },
    {
      "epoch": 2.970693240221272,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 1.3992,
      "step": 9465
    },
    {
      "epoch": 2.9710071011024364,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002,
      "loss": 1.7564,
      "step": 9466
    },
    {
      "epoch": 2.9713209619836007,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 1.8826,
      "step": 9467
    },
    {
      "epoch": 2.971634822864765,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.495,
      "step": 9468
    },
    {
      "epoch": 2.9719486837459295,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.5836,
      "step": 9469
    },
    {
      "epoch": 2.9722625446270943,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.5534,
      "step": 9470
    },
    {
      "epoch": 2.972576405508258,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002,
      "loss": 1.8829,
      "step": 9471
    },
    {
      "epoch": 2.972890266389423,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002,
      "loss": 2.3762,
      "step": 9472
    },
    {
      "epoch": 2.9732041272705874,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.553,
      "step": 9473
    },
    {
      "epoch": 2.9735179881517517,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.3238,
      "step": 9474
    },
    {
      "epoch": 2.973831849032916,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 1.4552,
      "step": 9475
    },
    {
      "epoch": 2.9741457099140804,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.3197,
      "step": 9476
    },
    {
      "epoch": 2.9744595707952453,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.3568,
      "step": 9477
    },
    {
      "epoch": 2.974773431676409,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 1.2488,
      "step": 9478
    },
    {
      "epoch": 2.975087292557574,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 1.3929,
      "step": 9479
    },
    {
      "epoch": 2.9754011534387383,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2632,
      "step": 9480
    },
    {
      "epoch": 2.9757150143199027,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.3085,
      "step": 9481
    },
    {
      "epoch": 2.976028875201067,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.2421,
      "step": 9482
    },
    {
      "epoch": 2.9763427360822314,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 1.2425,
      "step": 9483
    },
    {
      "epoch": 2.9766565969633962,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.2138,
      "step": 9484
    },
    {
      "epoch": 2.97697045784456,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.1889,
      "step": 9485
    },
    {
      "epoch": 2.977284318725725,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.2086,
      "step": 9486
    },
    {
      "epoch": 2.9775981796068893,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.2376,
      "step": 9487
    },
    {
      "epoch": 2.9779120404880537,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 1.2716,
      "step": 9488
    },
    {
      "epoch": 2.978225901369218,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.2318,
      "step": 9489
    },
    {
      "epoch": 2.9785397622503824,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.1857,
      "step": 9490
    },
    {
      "epoch": 2.978853623131547,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 1.1519,
      "step": 9491
    },
    {
      "epoch": 2.979167484012711,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 1.1038,
      "step": 9492
    },
    {
      "epoch": 2.979481344893876,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0002,
      "loss": 1.1719,
      "step": 9493
    },
    {
      "epoch": 2.9797952057750403,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 1.2695,
      "step": 9494
    },
    {
      "epoch": 2.9801090666562047,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0002,
      "loss": 1.1702,
      "step": 9495
    },
    {
      "epoch": 2.980422927537369,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 1.2703,
      "step": 9496
    },
    {
      "epoch": 2.9807367884185334,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 1.1476,
      "step": 9497
    },
    {
      "epoch": 2.9810506492996978,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 1.1776,
      "step": 9498
    },
    {
      "epoch": 2.981364510180862,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 1.187,
      "step": 9499
    },
    {
      "epoch": 2.981678371062027,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.1321,
      "step": 9500
    },
    {
      "epoch": 2.9819922319431913,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.3621,
      "step": 9501
    },
    {
      "epoch": 2.9823060928243557,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.4042,
      "step": 9502
    },
    {
      "epoch": 2.98261995370552,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 1.4612,
      "step": 9503
    },
    {
      "epoch": 2.9829338145866844,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 1.3821,
      "step": 9504
    },
    {
      "epoch": 2.9832476754678487,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 1.4627,
      "step": 9505
    },
    {
      "epoch": 2.983561536349013,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002,
      "loss": 1.4459,
      "step": 9506
    },
    {
      "epoch": 2.983875397230178,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 1.7426,
      "step": 9507
    },
    {
      "epoch": 2.9841892581113423,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002,
      "loss": 1.5772,
      "step": 9508
    },
    {
      "epoch": 2.9845031189925066,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 1.611,
      "step": 9509
    },
    {
      "epoch": 2.984816979873671,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002,
      "loss": 1.9711,
      "step": 9510
    },
    {
      "epoch": 2.9851308407548354,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002,
      "loss": 2.0818,
      "step": 9511
    },
    {
      "epoch": 2.9854447016359997,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 1.6855,
      "step": 9512
    },
    {
      "epoch": 2.985758562517164,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.5575,
      "step": 9513
    },
    {
      "epoch": 2.986072423398329,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 1.5207,
      "step": 9514
    },
    {
      "epoch": 2.9863862842794933,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002,
      "loss": 1.6288,
      "step": 9515
    },
    {
      "epoch": 2.9867001451606576,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 1.7329,
      "step": 9516
    },
    {
      "epoch": 2.987014006041822,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002,
      "loss": 1.2674,
      "step": 9517
    },
    {
      "epoch": 2.9873278669229864,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 1.6206,
      "step": 9518
    },
    {
      "epoch": 2.9876417278041507,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002,
      "loss": 1.7021,
      "step": 9519
    },
    {
      "epoch": 2.987955588685315,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002,
      "loss": 1.5577,
      "step": 9520
    },
    {
      "epoch": 2.98826944956648,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.1903,
      "step": 9521
    },
    {
      "epoch": 2.9885833104476442,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0002,
      "loss": 2.6006,
      "step": 9522
    },
    {
      "epoch": 2.9888971713288086,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.5539,
      "step": 9523
    },
    {
      "epoch": 2.989211032209973,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 1.3249,
      "step": 9524
    },
    {
      "epoch": 2.9895248930911373,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.3134,
      "step": 9525
    },
    {
      "epoch": 2.9898387539723017,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 1.2997,
      "step": 9526
    },
    {
      "epoch": 2.990152614853466,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002,
      "loss": 1.2315,
      "step": 9527
    },
    {
      "epoch": 2.990466475734631,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.2112,
      "step": 9528
    },
    {
      "epoch": 2.990780336615795,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.2611,
      "step": 9529
    },
    {
      "epoch": 2.9910941974969596,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 1.3462,
      "step": 9530
    },
    {
      "epoch": 2.991408058378124,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 1.2274,
      "step": 9531
    },
    {
      "epoch": 2.9917219192592883,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 1.2384,
      "step": 9532
    },
    {
      "epoch": 2.9920357801404527,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 1.3161,
      "step": 9533
    },
    {
      "epoch": 2.992349641021617,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 1.1396,
      "step": 9534
    },
    {
      "epoch": 2.992663501902782,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 1.1765,
      "step": 9535
    },
    {
      "epoch": 2.9929773627839458,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 1.2897,
      "step": 9536
    },
    {
      "epoch": 2.9932912236651106,
      "grad_norm": 0.390625,
      "learning_rate": 0.0002,
      "loss": 1.1747,
      "step": 9537
    },
    {
      "epoch": 2.9932912236651106,
      "eval_loss": 1.8602570295333862,
      "eval_runtime": 123.0153,
      "eval_samples_per_second": 8.129,
      "eval_steps_per_second": 8.129,
      "step": 9537
    },
    {
      "epoch": 2.9932912236651106,
      "mmlu_eval_accuracy": 0.42444306629488804,
      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
      "mmlu_eval_accuracy_anatomy": 0.5,
      "mmlu_eval_accuracy_astronomy": 0.5,
      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
      "mmlu_eval_accuracy_college_biology": 0.25,
      "mmlu_eval_accuracy_college_chemistry": 0.375,
      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
      "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
      "mmlu_eval_accuracy_computer_security": 0.18181818181818182,
      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
      "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
      "mmlu_eval_accuracy_electrical_engineering": 0.375,
      "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
      "mmlu_eval_accuracy_global_facts": 0.6,
      "mmlu_eval_accuracy_high_school_biology": 0.28125,
      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
      "mmlu_eval_accuracy_high_school_computer_science": 0.3333333333333333,
      "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
      "mmlu_eval_accuracy_high_school_geography": 0.5909090909090909,
      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5238095238095238,
      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5116279069767442,
      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
      "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
      "mmlu_eval_accuracy_high_school_physics": 0.47058823529411764,
      "mmlu_eval_accuracy_high_school_psychology": 0.6,
      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
      "mmlu_eval_accuracy_high_school_us_history": 0.5,
      "mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
      "mmlu_eval_accuracy_human_aging": 0.43478260869565216,
      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
      "mmlu_eval_accuracy_logical_fallacies": 0.5,
      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
      "mmlu_eval_accuracy_management": 0.5454545454545454,
      "mmlu_eval_accuracy_marketing": 0.72,
      "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
      "mmlu_eval_accuracy_miscellaneous": 0.5116279069767442,
      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
      "mmlu_eval_accuracy_moral_scenarios": 0.23,
      "mmlu_eval_accuracy_nutrition": 0.48484848484848486,
      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
      "mmlu_eval_accuracy_prehistory": 0.34285714285714286,
      "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
      "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
      "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
      "mmlu_eval_accuracy_public_relations": 0.5,
      "mmlu_eval_accuracy_security_studies": 0.37037037037037035,
      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
      "mmlu_eval_accuracy_virology": 0.2777777777777778,
      "mmlu_eval_accuracy_world_religions": 0.5789473684210527,
      "mmlu_loss": 1.2572945500416821,
      "step": 9537
    },
    {
      "epoch": 2.993605084546275,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.0661,
      "step": 9538
    },
    {
      "epoch": 2.9939189454274393,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 1.1755,
      "step": 9539
    },
    {
      "epoch": 2.9942328063086037,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0002,
      "loss": 1.1012,
      "step": 9540
    },
    {
      "epoch": 2.994546667189768,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 1.0115,
      "step": 9541
    },
    {
      "epoch": 2.9948605280709324,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 1.3731,
      "step": 9542
    },
    {
      "epoch": 2.9951743889520968,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 1.1223,
      "step": 9543
    },
    {
      "epoch": 2.9954882498332616,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002,
      "loss": 1.6787,
      "step": 9544
    },
    {
      "epoch": 2.995802110714426,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 1.7123,
      "step": 9545
    },
    {
      "epoch": 2.9961159715955903,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 1.5949,
      "step": 9546
    },
    {
      "epoch": 2.9964298324767547,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 1.7979,
      "step": 9547
    },
    {
      "epoch": 2.996743693357919,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 1.9313,
      "step": 9548
    },
    {
      "epoch": 2.9970575542390834,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002,
      "loss": 1.3465,
      "step": 9549
    },
    {
      "epoch": 2.9973714151202477,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 1.2209,
      "step": 9550
    },
    {
      "epoch": 2.9976852760014125,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 1.3464,
      "step": 9551
    },
    {
      "epoch": 2.997999136882577,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 1.8226,
      "step": 9552
    },
    {
      "epoch": 2.9983129977637413,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 1.2924,
      "step": 9553
    },
    {
      "epoch": 2.9986268586449056,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002,
      "loss": 1.7133,
      "step": 9554
    },
    {
      "epoch": 2.99894071952607,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002,
      "loss": 1.7165,
      "step": 9555
    },
    {
      "epoch": 2.9992545804072344,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002,
      "loss": 1.6035,
      "step": 9556
    },
    {
      "epoch": 2.9995684412883987,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0002,
      "loss": 1.8044,
      "step": 9557
    },
    {
      "epoch": 2.9998823021695635,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 1.9065,
      "step": 9558
    },
    {
      "epoch": 2.9998823021695635,
      "step": 9558,
      "total_flos": 5.173921474210529e+17,
      "train_loss": 1.616389752793796,
      "train_runtime": 177951.3432,
      "train_samples_per_second": 0.859,
      "train_steps_per_second": 0.054
    }
  ],
  "logging_steps": 1,
  "max_steps": 9558,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.173921474210529e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}