{
  "best_metric": 1.0859394073486328,
  "best_model_checkpoint": "miner_id_24/checkpoint-500",
  "epoch": 0.6884681583476764,
  "eval_steps": 250,
  "global_step": 700,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0009835259404966806,
      "grad_norm": 9.03037166595459,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.1955,
      "step": 1
    },
    {
      "epoch": 0.0009835259404966806,
      "eval_loss": 1.1275324821472168,
      "eval_runtime": 484.6766,
      "eval_samples_per_second": 33.563,
      "eval_steps_per_second": 4.197,
      "step": 1
    },
    {
      "epoch": 0.001967051880993361,
      "grad_norm": 8.760616302490234,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.0633,
      "step": 2
    },
    {
      "epoch": 0.0029505778214900416,
      "grad_norm": 6.406024932861328,
      "learning_rate": 1.2e-05,
      "loss": 1.0792,
      "step": 3
    },
    {
      "epoch": 0.003934103761986722,
      "grad_norm": 6.365318298339844,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.0979,
      "step": 4
    },
    {
      "epoch": 0.004917629702483403,
      "grad_norm": 8.005640029907227,
      "learning_rate": 2e-05,
      "loss": 1.0387,
      "step": 5
    },
    {
      "epoch": 0.005901155642980083,
      "grad_norm": 7.427581787109375,
      "learning_rate": 2.4e-05,
      "loss": 1.1472,
      "step": 6
    },
    {
      "epoch": 0.0068846815834767644,
      "grad_norm": 6.178953647613525,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 1.0855,
      "step": 7
    },
    {
      "epoch": 0.007868207523973445,
      "grad_norm": 6.672806262969971,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 1.1226,
      "step": 8
    },
    {
      "epoch": 0.008851733464470126,
      "grad_norm": 6.151551723480225,
      "learning_rate": 3.6e-05,
      "loss": 1.0987,
      "step": 9
    },
    {
      "epoch": 0.009835259404966806,
      "grad_norm": 4.984577655792236,
      "learning_rate": 4e-05,
      "loss": 0.9842,
      "step": 10
    },
    {
      "epoch": 0.010818785345463487,
      "grad_norm": 4.839271068572998,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 1.0339,
      "step": 11
    },
    {
      "epoch": 0.011802311285960166,
      "grad_norm": 4.72609281539917,
      "learning_rate": 4.8e-05,
      "loss": 1.0376,
      "step": 12
    },
    {
      "epoch": 0.012785837226456848,
      "grad_norm": 4.5971360206604,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 1.0003,
      "step": 13
    },
    {
      "epoch": 0.013769363166953529,
      "grad_norm": 4.283706188201904,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 1.0122,
      "step": 14
    },
    {
      "epoch": 0.014752889107450208,
      "grad_norm": 4.256502151489258,
      "learning_rate": 6e-05,
      "loss": 1.0243,
      "step": 15
    },
    {
      "epoch": 0.01573641504794689,
      "grad_norm": 4.081477642059326,
      "learning_rate": 6.400000000000001e-05,
      "loss": 1.0621,
      "step": 16
    },
    {
      "epoch": 0.01671994098844357,
      "grad_norm": 4.428706645965576,
      "learning_rate": 6.800000000000001e-05,
      "loss": 1.0631,
      "step": 17
    },
    {
      "epoch": 0.017703466928940252,
      "grad_norm": 4.312934398651123,
      "learning_rate": 7.2e-05,
      "loss": 1.0676,
      "step": 18
    },
    {
      "epoch": 0.01868699286943693,
      "grad_norm": 4.375700950622559,
      "learning_rate": 7.6e-05,
      "loss": 1.1003,
      "step": 19
    },
    {
      "epoch": 0.01967051880993361,
      "grad_norm": 4.138792514801025,
      "learning_rate": 8e-05,
      "loss": 0.983,
      "step": 20
    },
    {
      "epoch": 0.020654044750430294,
      "grad_norm": 3.9937527179718018,
      "learning_rate": 8.4e-05,
      "loss": 1.028,
      "step": 21
    },
    {
      "epoch": 0.021637570690926974,
      "grad_norm": 3.854146718978882,
      "learning_rate": 8.800000000000001e-05,
      "loss": 1.0542,
      "step": 22
    },
    {
      "epoch": 0.022621096631423653,
      "grad_norm": 3.9819438457489014,
      "learning_rate": 9.200000000000001e-05,
      "loss": 1.0962,
      "step": 23
    },
    {
      "epoch": 0.023604622571920333,
      "grad_norm": 3.7932257652282715,
      "learning_rate": 9.6e-05,
      "loss": 1.0027,
      "step": 24
    },
    {
      "epoch": 0.024588148512417016,
      "grad_norm": 3.985522747039795,
      "learning_rate": 0.0001,
      "loss": 1.0385,
      "step": 25
    },
    {
      "epoch": 0.025571674452913695,
      "grad_norm": 4.005710601806641,
      "learning_rate": 0.00010400000000000001,
      "loss": 1.0082,
      "step": 26
    },
    {
      "epoch": 0.026555200393410375,
      "grad_norm": 4.152320861816406,
      "learning_rate": 0.00010800000000000001,
      "loss": 1.1124,
      "step": 27
    },
    {
      "epoch": 0.027538726333907058,
      "grad_norm": 4.043179988861084,
      "learning_rate": 0.00011200000000000001,
      "loss": 1.0395,
      "step": 28
    },
    {
      "epoch": 0.028522252274403737,
      "grad_norm": 3.829171895980835,
      "learning_rate": 0.000116,
      "loss": 1.0675,
      "step": 29
    },
    {
      "epoch": 0.029505778214900417,
      "grad_norm": 3.7003633975982666,
      "learning_rate": 0.00012,
      "loss": 1.1005,
      "step": 30
    },
    {
      "epoch": 0.0304893041553971,
      "grad_norm": 3.832540988922119,
      "learning_rate": 0.000124,
      "loss": 1.0442,
      "step": 31
    },
    {
      "epoch": 0.03147283009589378,
      "grad_norm": 3.9027154445648193,
      "learning_rate": 0.00012800000000000002,
      "loss": 1.1144,
      "step": 32
    },
    {
      "epoch": 0.03245635603639046,
      "grad_norm": 3.805137872695923,
      "learning_rate": 0.000132,
      "loss": 1.1183,
      "step": 33
    },
    {
      "epoch": 0.03343988197688714,
      "grad_norm": 3.929744243621826,
      "learning_rate": 0.00013600000000000003,
      "loss": 1.1401,
      "step": 34
    },
    {
      "epoch": 0.03442340791738382,
      "grad_norm": 3.6989941596984863,
      "learning_rate": 0.00014,
      "loss": 1.071,
      "step": 35
    },
    {
      "epoch": 0.035406933857880504,
      "grad_norm": 3.7118210792541504,
      "learning_rate": 0.000144,
      "loss": 1.0526,
      "step": 36
    },
    {
      "epoch": 0.036390459798377184,
      "grad_norm": 3.7488160133361816,
      "learning_rate": 0.000148,
      "loss": 1.1051,
      "step": 37
    },
    {
      "epoch": 0.03737398573887386,
      "grad_norm": 3.595310926437378,
      "learning_rate": 0.000152,
      "loss": 1.064,
      "step": 38
    },
    {
      "epoch": 0.03835751167937054,
      "grad_norm": 3.55122971534729,
      "learning_rate": 0.00015600000000000002,
      "loss": 1.0407,
      "step": 39
    },
    {
      "epoch": 0.03934103761986722,
      "grad_norm": 3.5784990787506104,
      "learning_rate": 0.00016,
      "loss": 1.102,
      "step": 40
    },
    {
      "epoch": 0.0403245635603639,
      "grad_norm": 3.5550732612609863,
      "learning_rate": 0.000164,
      "loss": 1.1022,
      "step": 41
    },
    {
      "epoch": 0.04130808950086059,
      "grad_norm": 3.4226324558258057,
      "learning_rate": 0.000168,
      "loss": 1.1179,
      "step": 42
    },
    {
      "epoch": 0.04229161544135727,
      "grad_norm": 3.2767834663391113,
      "learning_rate": 0.000172,
      "loss": 1.0915,
      "step": 43
    },
    {
      "epoch": 0.04327514138185395,
      "grad_norm": 3.420482873916626,
      "learning_rate": 0.00017600000000000002,
      "loss": 1.0812,
      "step": 44
    },
    {
      "epoch": 0.04425866732235063,
      "grad_norm": 3.3167741298675537,
      "learning_rate": 0.00018,
      "loss": 1.1189,
      "step": 45
    },
    {
      "epoch": 0.045242193262847306,
      "grad_norm": 3.238745927810669,
      "learning_rate": 0.00018400000000000003,
      "loss": 1.0461,
      "step": 46
    },
    {
      "epoch": 0.046225719203343986,
      "grad_norm": 3.281376600265503,
      "learning_rate": 0.000188,
      "loss": 1.1617,
      "step": 47
    },
    {
      "epoch": 0.047209245143840665,
      "grad_norm": 3.2215256690979004,
      "learning_rate": 0.000192,
      "loss": 1.1205,
      "step": 48
    },
    {
      "epoch": 0.04819277108433735,
      "grad_norm": 3.3025946617126465,
      "learning_rate": 0.000196,
      "loss": 1.1562,
      "step": 49
    },
    {
      "epoch": 0.04917629702483403,
      "grad_norm": 3.150087356567383,
      "learning_rate": 0.0002,
      "loss": 1.1736,
      "step": 50
    },
    {
      "epoch": 0.05015982296533071,
      "grad_norm": 3.1989223957061768,
      "learning_rate": 0.00020400000000000003,
      "loss": 1.1633,
      "step": 51
    },
    {
      "epoch": 0.05114334890582739,
      "grad_norm": 3.330007314682007,
      "learning_rate": 0.00020800000000000001,
      "loss": 1.1628,
      "step": 52
    },
    {
      "epoch": 0.05212687484632407,
      "grad_norm": 3.2046127319335938,
      "learning_rate": 0.00021200000000000003,
      "loss": 1.1249,
      "step": 53
    },
    {
      "epoch": 0.05311040078682075,
      "grad_norm": 3.350651741027832,
      "learning_rate": 0.00021600000000000002,
      "loss": 1.2083,
      "step": 54
    },
    {
      "epoch": 0.054093926727317436,
      "grad_norm": 3.347811698913574,
      "learning_rate": 0.00022000000000000003,
      "loss": 1.198,
      "step": 55
    },
    {
      "epoch": 0.055077452667814115,
      "grad_norm": 3.261667251586914,
      "learning_rate": 0.00022400000000000002,
      "loss": 1.1676,
      "step": 56
    },
    {
      "epoch": 0.056060978608310795,
      "grad_norm": 3.2465434074401855,
      "learning_rate": 0.00022799999999999999,
      "loss": 1.1817,
      "step": 57
    },
    {
      "epoch": 0.057044504548807475,
      "grad_norm": 3.2840354442596436,
      "learning_rate": 0.000232,
      "loss": 1.1665,
      "step": 58
    },
    {
      "epoch": 0.058028030489304154,
      "grad_norm": 3.216740846633911,
      "learning_rate": 0.000236,
      "loss": 1.1982,
      "step": 59
    },
    {
      "epoch": 0.059011556429800834,
      "grad_norm": 3.4775424003601074,
      "learning_rate": 0.00024,
      "loss": 1.209,
      "step": 60
    },
    {
      "epoch": 0.05999508237029752,
      "grad_norm": 3.0584778785705566,
      "learning_rate": 0.000244,
      "loss": 1.1706,
      "step": 61
    },
    {
      "epoch": 0.0609786083107942,
      "grad_norm": 3.017756938934326,
      "learning_rate": 0.000248,
      "loss": 1.167,
      "step": 62
    },
    {
      "epoch": 0.06196213425129088,
      "grad_norm": 3.0101511478424072,
      "learning_rate": 0.000252,
      "loss": 1.2131,
      "step": 63
    },
    {
      "epoch": 0.06294566019178756,
      "grad_norm": 3.021139621734619,
      "learning_rate": 0.00025600000000000004,
      "loss": 1.1957,
      "step": 64
    },
    {
      "epoch": 0.06392918613228424,
      "grad_norm": 3.0112669467926025,
      "learning_rate": 0.00026000000000000003,
      "loss": 1.1844,
      "step": 65
    },
    {
      "epoch": 0.06491271207278092,
      "grad_norm": 3.1999220848083496,
      "learning_rate": 0.000264,
      "loss": 1.2539,
      "step": 66
    },
    {
      "epoch": 0.0658962380132776,
      "grad_norm": 3.2715108394622803,
      "learning_rate": 0.000268,
      "loss": 1.2542,
      "step": 67
    },
    {
      "epoch": 0.06687976395377428,
      "grad_norm": 2.9691739082336426,
      "learning_rate": 0.00027200000000000005,
      "loss": 1.2363,
      "step": 68
    },
    {
      "epoch": 0.06786328989427096,
      "grad_norm": 3.060223340988159,
      "learning_rate": 0.000276,
      "loss": 1.237,
      "step": 69
    },
    {
      "epoch": 0.06884681583476764,
      "grad_norm": 2.989504814147949,
      "learning_rate": 0.00028,
      "loss": 1.2826,
      "step": 70
    },
    {
      "epoch": 0.06983034177526433,
      "grad_norm": 3.0857932567596436,
      "learning_rate": 0.000284,
      "loss": 1.1703,
      "step": 71
    },
    {
      "epoch": 0.07081386771576101,
      "grad_norm": 3.0549941062927246,
      "learning_rate": 0.000288,
      "loss": 1.2199,
      "step": 72
    },
    {
      "epoch": 0.07179739365625769,
      "grad_norm": 3.007155179977417,
      "learning_rate": 0.000292,
      "loss": 1.264,
      "step": 73
    },
    {
      "epoch": 0.07278091959675437,
      "grad_norm": 3.0878915786743164,
      "learning_rate": 0.000296,
      "loss": 1.2983,
      "step": 74
    },
    {
      "epoch": 0.07376444553725105,
      "grad_norm": 2.9615390300750732,
      "learning_rate": 0.00030000000000000003,
      "loss": 1.224,
      "step": 75
    },
    {
      "epoch": 0.07474797147774773,
      "grad_norm": 3.080251932144165,
      "learning_rate": 0.000304,
      "loss": 1.2241,
      "step": 76
    },
    {
      "epoch": 0.0757314974182444,
      "grad_norm": 3.0525965690612793,
      "learning_rate": 0.000308,
      "loss": 1.3308,
      "step": 77
    },
    {
      "epoch": 0.07671502335874109,
      "grad_norm": 2.922020673751831,
      "learning_rate": 0.00031200000000000005,
      "loss": 1.3482,
      "step": 78
    },
    {
      "epoch": 0.07769854929923777,
      "grad_norm": 3.15800404548645,
      "learning_rate": 0.00031600000000000004,
      "loss": 1.2608,
      "step": 79
    },
    {
      "epoch": 0.07868207523973444,
      "grad_norm": 3.229642152786255,
      "learning_rate": 0.00032,
      "loss": 1.3027,
      "step": 80
    },
    {
      "epoch": 0.07966560118023112,
      "grad_norm": 3.2053112983703613,
      "learning_rate": 0.000324,
      "loss": 1.3266,
      "step": 81
    },
    {
      "epoch": 0.0806491271207278,
      "grad_norm": 2.8999037742614746,
      "learning_rate": 0.000328,
      "loss": 1.263,
      "step": 82
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 2.933065176010132,
      "learning_rate": 0.000332,
      "loss": 1.3056,
      "step": 83
    },
    {
      "epoch": 0.08261617900172118,
      "grad_norm": 2.87294340133667,
      "learning_rate": 0.000336,
      "loss": 1.2671,
      "step": 84
    },
    {
      "epoch": 0.08359970494221786,
      "grad_norm": 2.8643088340759277,
      "learning_rate": 0.00034,
      "loss": 1.3105,
      "step": 85
    },
    {
      "epoch": 0.08458323088271454,
      "grad_norm": 3.0644736289978027,
      "learning_rate": 0.000344,
      "loss": 1.2914,
      "step": 86
    },
    {
      "epoch": 0.08556675682321122,
      "grad_norm": 2.843308448791504,
      "learning_rate": 0.000348,
      "loss": 1.4172,
      "step": 87
    },
    {
      "epoch": 0.0865502827637079,
      "grad_norm": 3.0372660160064697,
      "learning_rate": 0.00035200000000000005,
      "loss": 1.2841,
      "step": 88
    },
    {
      "epoch": 0.08753380870420457,
      "grad_norm": 2.7811527252197266,
      "learning_rate": 0.00035600000000000003,
      "loss": 1.2991,
      "step": 89
    },
    {
      "epoch": 0.08851733464470125,
      "grad_norm": 3.0117645263671875,
      "learning_rate": 0.00036,
      "loss": 1.3515,
      "step": 90
    },
    {
      "epoch": 0.08950086058519793,
      "grad_norm": 2.8653600215911865,
      "learning_rate": 0.000364,
      "loss": 1.3834,
      "step": 91
    },
    {
      "epoch": 0.09048438652569461,
      "grad_norm": 2.8261609077453613,
      "learning_rate": 0.00036800000000000005,
      "loss": 1.305,
      "step": 92
    },
    {
      "epoch": 0.09146791246619129,
      "grad_norm": 2.7723865509033203,
      "learning_rate": 0.00037200000000000004,
      "loss": 1.3566,
      "step": 93
    },
    {
      "epoch": 0.09245143840668797,
      "grad_norm": 3.1048543453216553,
      "learning_rate": 0.000376,
      "loss": 1.3605,
      "step": 94
    },
    {
      "epoch": 0.09343496434718465,
      "grad_norm": 3.0473995208740234,
      "learning_rate": 0.00038,
      "loss": 1.3368,
      "step": 95
    },
    {
      "epoch": 0.09441849028768133,
      "grad_norm": 2.969036102294922,
      "learning_rate": 0.000384,
      "loss": 1.3693,
      "step": 96
    },
    {
      "epoch": 0.09540201622817802,
      "grad_norm": 2.808734178543091,
      "learning_rate": 0.000388,
      "loss": 1.3842,
      "step": 97
    },
    {
      "epoch": 0.0963855421686747,
      "grad_norm": 2.9717369079589844,
      "learning_rate": 0.000392,
      "loss": 1.3874,
      "step": 98
    },
    {
      "epoch": 0.09736906810917138,
      "grad_norm": 2.8518717288970947,
      "learning_rate": 0.00039600000000000003,
      "loss": 1.348,
      "step": 99
    },
    {
      "epoch": 0.09835259404966806,
      "grad_norm": 2.9979913234710693,
      "learning_rate": 0.0004,
      "loss": 1.3793,
      "step": 100
    },
    {
      "epoch": 0.09933611999016474,
      "grad_norm": 2.8096790313720703,
      "learning_rate": 0.00039999725844948536,
      "loss": 1.3618,
      "step": 101
    },
    {
      "epoch": 0.10031964593066142,
      "grad_norm": 2.811119318008423,
      "learning_rate": 0.00039998903387310247,
      "loss": 1.4195,
      "step": 102
    },
    {
      "epoch": 0.1013031718711581,
      "grad_norm": 2.7859537601470947,
      "learning_rate": 0.00039997532649633215,
      "loss": 1.3467,
      "step": 103
    },
    {
      "epoch": 0.10228669781165478,
      "grad_norm": 2.910825252532959,
      "learning_rate": 0.0003999561366949691,
      "loss": 1.4546,
      "step": 104
    },
    {
      "epoch": 0.10327022375215146,
      "grad_norm": 3.2803215980529785,
      "learning_rate": 0.00039993146499511144,
      "loss": 1.3661,
      "step": 105
    },
    {
      "epoch": 0.10425374969264814,
      "grad_norm": 2.7668137550354004,
      "learning_rate": 0.00039990131207314633,
      "loss": 1.4008,
      "step": 106
    },
    {
      "epoch": 0.10523727563314482,
      "grad_norm": 2.752786874771118,
      "learning_rate": 0.00039986567875573126,
      "loss": 1.4156,
      "step": 107
    },
    {
      "epoch": 0.1062208015736415,
      "grad_norm": 2.808030366897583,
      "learning_rate": 0.0003998245660197717,
      "loss": 1.4068,
      "step": 108
    },
    {
      "epoch": 0.10720432751413818,
      "grad_norm": 2.8663578033447266,
      "learning_rate": 0.00039977797499239404,
      "loss": 1.453,
      "step": 109
    },
    {
      "epoch": 0.10818785345463487,
      "grad_norm": 2.627333402633667,
      "learning_rate": 0.0003997259069509148,
      "loss": 1.4093,
      "step": 110
    },
    {
      "epoch": 0.10917137939513155,
      "grad_norm": 2.807516574859619,
      "learning_rate": 0.0003996683633228057,
      "loss": 1.3799,
      "step": 111
    },
    {
      "epoch": 0.11015490533562823,
      "grad_norm": 2.89886474609375,
      "learning_rate": 0.00039960534568565434,
      "loss": 1.4899,
      "step": 112
    },
    {
      "epoch": 0.11113843127612491,
      "grad_norm": 2.8411178588867188,
      "learning_rate": 0.0003995368557671211,
      "loss": 1.4382,
      "step": 113
    },
    {
      "epoch": 0.11212195721662159,
      "grad_norm": 2.9060070514678955,
      "learning_rate": 0.00039946289544489163,
      "loss": 1.4959,
      "step": 114
    },
    {
      "epoch": 0.11310548315711827,
      "grad_norm": 2.7524595260620117,
      "learning_rate": 0.0003993834667466256,
      "loss": 1.3913,
      "step": 115
    },
    {
      "epoch": 0.11408900909761495,
      "grad_norm": 3.0734031200408936,
      "learning_rate": 0.0003992985718499009,
      "loss": 1.4537,
      "step": 116
    },
    {
      "epoch": 0.11507253503811163,
      "grad_norm": 2.654207706451416,
      "learning_rate": 0.00039920821308215394,
      "loss": 1.4131,
      "step": 117
    },
    {
      "epoch": 0.11605606097860831,
      "grad_norm": 2.639305830001831,
      "learning_rate": 0.00039911239292061604,
      "loss": 1.5106,
      "step": 118
    },
    {
      "epoch": 0.11703958691910499,
      "grad_norm": 2.814849853515625,
      "learning_rate": 0.00039901111399224534,
      "loss": 1.4856,
      "step": 119
    },
    {
      "epoch": 0.11802311285960167,
      "grad_norm": 2.888063907623291,
      "learning_rate": 0.0003989043790736547,
      "loss": 1.4671,
      "step": 120
    },
    {
      "epoch": 0.11900663880009835,
      "grad_norm": 2.623919725418091,
      "learning_rate": 0.000398792191091036,
      "loss": 1.4942,
      "step": 121
    },
    {
      "epoch": 0.11999016474059504,
      "grad_norm": 2.8057353496551514,
      "learning_rate": 0.00039867455312007926,
      "loss": 1.4451,
      "step": 122
    },
    {
      "epoch": 0.12097369068109172,
      "grad_norm": 2.8340566158294678,
      "learning_rate": 0.00039855146838588913,
      "loss": 1.4547,
      "step": 123
    },
    {
      "epoch": 0.1219572166215884,
      "grad_norm": 2.714200496673584,
      "learning_rate": 0.0003984229402628956,
      "loss": 1.4661,
      "step": 124
    },
    {
      "epoch": 0.12294074256208508,
      "grad_norm": 2.887300491333008,
      "learning_rate": 0.0003982889722747621,
      "loss": 1.475,
      "step": 125
    },
    {
      "epoch": 0.12392426850258176,
      "grad_norm": 2.792158365249634,
      "learning_rate": 0.0003981495680942887,
      "loss": 1.5472,
      "step": 126
    },
    {
      "epoch": 0.12490779444307844,
      "grad_norm": 2.6247410774230957,
      "learning_rate": 0.00039800473154331153,
      "loss": 1.4102,
      "step": 127
    },
    {
      "epoch": 0.12589132038357512,
      "grad_norm": 2.675847291946411,
      "learning_rate": 0.00039785446659259773,
      "loss": 1.5081,
      "step": 128
    },
    {
      "epoch": 0.1268748463240718,
      "grad_norm": 2.617077350616455,
      "learning_rate": 0.0003976987773617367,
      "loss": 1.3768,
      "step": 129
    },
    {
      "epoch": 0.12785837226456848,
      "grad_norm": 2.740143299102783,
      "learning_rate": 0.00039753766811902755,
      "loss": 1.4738,
      "step": 130
    },
    {
      "epoch": 0.12884189820506517,
      "grad_norm": 2.608574390411377,
      "learning_rate": 0.0003973711432813615,
      "loss": 1.4572,
      "step": 131
    },
    {
      "epoch": 0.12982542414556184,
      "grad_norm": 2.8854312896728516,
      "learning_rate": 0.000397199207414101,
      "loss": 1.4841,
      "step": 132
    },
    {
      "epoch": 0.13080895008605853,
      "grad_norm": 2.779543161392212,
      "learning_rate": 0.0003970218652309548,
      "loss": 1.4778,
      "step": 133
    },
    {
      "epoch": 0.1317924760265552,
      "grad_norm": 2.8026397228240967,
      "learning_rate": 0.0003968391215938484,
      "loss": 1.4394,
      "step": 134
    },
    {
      "epoch": 0.1327760019670519,
      "grad_norm": 2.6550095081329346,
      "learning_rate": 0.000396650981512791,
      "loss": 1.5005,
      "step": 135
    },
    {
      "epoch": 0.13375952790754855,
      "grad_norm": 2.8108248710632324,
      "learning_rate": 0.0003964574501457378,
      "loss": 1.5245,
      "step": 136
    },
    {
      "epoch": 0.13474305384804525,
      "grad_norm": 2.8617358207702637,
      "learning_rate": 0.000396258532798449,
      "loss": 1.4474,
      "step": 137
    },
    {
      "epoch": 0.1357265797885419,
      "grad_norm": 2.939607620239258,
      "learning_rate": 0.00039605423492434437,
      "loss": 1.4929,
      "step": 138
    },
    {
      "epoch": 0.1367101057290386,
      "grad_norm": 2.8078112602233887,
      "learning_rate": 0.00039584456212435316,
      "loss": 1.4675,
      "step": 139
    },
    {
      "epoch": 0.13769363166953527,
      "grad_norm": 2.72772479057312,
      "learning_rate": 0.00039562952014676116,
      "loss": 1.47,
      "step": 140
    },
    {
      "epoch": 0.13867715761003196,
      "grad_norm": 2.5927319526672363,
      "learning_rate": 0.0003954091148870528,
      "loss": 1.4201,
      "step": 141
    },
    {
      "epoch": 0.13966068355052866,
      "grad_norm": 2.6626243591308594,
      "learning_rate": 0.00039518335238774953,
      "loss": 1.4841,
      "step": 142
    },
    {
      "epoch": 0.14064420949102532,
      "grad_norm": 2.7324771881103516,
      "learning_rate": 0.0003949522388382444,
      "loss": 1.4551,
      "step": 143
    },
    {
      "epoch": 0.14162773543152202,
      "grad_norm": 2.604612350463867,
      "learning_rate": 0.0003947157805746321,
      "loss": 1.4115,
      "step": 144
    },
    {
      "epoch": 0.14261126137201868,
      "grad_norm": 2.6373226642608643,
      "learning_rate": 0.0003944739840795353,
      "loss": 1.4459,
      "step": 145
    },
    {
      "epoch": 0.14359478731251538,
      "grad_norm": 2.418677568435669,
      "learning_rate": 0.00039422685598192723,
      "loss": 1.4385,
      "step": 146
    },
    {
      "epoch": 0.14457831325301204,
      "grad_norm": 2.7488105297088623,
      "learning_rate": 0.00039397440305694937,
      "loss": 1.4752,
      "step": 147
    },
    {
      "epoch": 0.14556183919350874,
      "grad_norm": 2.63874888420105,
      "learning_rate": 0.0003937166322257262,
      "loss": 1.5162,
      "step": 148
    },
    {
      "epoch": 0.1465453651340054,
      "grad_norm": 2.529266595840454,
      "learning_rate": 0.0003934535505551754,
      "loss": 1.494,
      "step": 149
    },
    {
      "epoch": 0.1475288910745021,
      "grad_norm": 2.5442469120025635,
      "learning_rate": 0.00039318516525781367,
      "loss": 1.4872,
      "step": 150
    },
    {
      "epoch": 0.14851241701499876,
      "grad_norm": 2.616391658782959,
      "learning_rate": 0.00039291148369155964,
      "loss": 1.4668,
      "step": 151
    },
    {
      "epoch": 0.14949594295549545,
      "grad_norm": 2.5333828926086426,
      "learning_rate": 0.00039263251335953166,
      "loss": 1.4733,
      "step": 152
    },
    {
      "epoch": 0.15047946889599212,
      "grad_norm": 2.450347900390625,
      "learning_rate": 0.0003923482619098423,
      "loss": 1.5808,
      "step": 153
    },
    {
      "epoch": 0.1514629948364888,
      "grad_norm": 2.4622340202331543,
      "learning_rate": 0.0003920587371353886,
      "loss": 1.5297,
      "step": 154
    },
    {
      "epoch": 0.1524465207769855,
      "grad_norm": 2.627448558807373,
      "learning_rate": 0.0003917639469736386,
      "loss": 1.4834,
      "step": 155
    },
    {
      "epoch": 0.15343004671748217,
      "grad_norm": 2.396347999572754,
      "learning_rate": 0.00039146389950641346,
      "loss": 1.4618,
      "step": 156
    },
    {
      "epoch": 0.15441357265797886,
      "grad_norm": 2.5668747425079346,
      "learning_rate": 0.00039115860295966604,
      "loss": 1.4914,
      "step": 157
    },
    {
      "epoch": 0.15539709859847553,
      "grad_norm": 3.0465917587280273,
      "learning_rate": 0.0003908480657032554,
      "loss": 1.5055,
      "step": 158
    },
    {
      "epoch": 0.15638062453897222,
      "grad_norm": 2.384202718734741,
      "learning_rate": 0.0003905322962507173,
      "loss": 1.4339,
      "step": 159
    },
    {
      "epoch": 0.1573641504794689,
      "grad_norm": 2.435328245162964,
      "learning_rate": 0.00039021130325903074,
      "loss": 1.5682,
      "step": 160
    },
    {
      "epoch": 0.15834767641996558,
      "grad_norm": 2.399083375930786,
      "learning_rate": 0.0003898850955283808,
      "loss": 1.4667,
      "step": 161
    },
    {
      "epoch": 0.15933120236046225,
      "grad_norm": 2.4345366954803467,
      "learning_rate": 0.0003895536820019172,
      "loss": 1.388,
      "step": 162
    },
    {
      "epoch": 0.16031472830095894,
      "grad_norm": 2.531015396118164,
      "learning_rate": 0.0003892170717655091,
      "loss": 1.4716,
      "step": 163
    },
    {
      "epoch": 0.1612982542414556,
      "grad_norm": 2.4581446647644043,
      "learning_rate": 0.00038887527404749624,
      "loss": 1.4462,
      "step": 164
    },
    {
      "epoch": 0.1622817801819523,
      "grad_norm": 2.587550640106201,
      "learning_rate": 0.00038852829821843575,
      "loss": 1.5327,
      "step": 165
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 2.4031155109405518,
      "learning_rate": 0.00038817615379084514,
      "loss": 1.522,
      "step": 166
    },
    {
      "epoch": 0.16424883206294566,
      "grad_norm": 2.594148874282837,
      "learning_rate": 0.0003878188504189418,
      "loss": 1.4298,
      "step": 167
    },
    {
      "epoch": 0.16523235800344235,
      "grad_norm": 2.426117420196533,
      "learning_rate": 0.0003874563978983783,
      "loss": 1.4689,
      "step": 168
    },
    {
      "epoch": 0.16621588394393902,
      "grad_norm": 2.4734981060028076,
      "learning_rate": 0.0003870888061659735,
      "loss": 1.4861,
      "step": 169
    },
    {
      "epoch": 0.1671994098844357,
      "grad_norm": 2.446207284927368,
      "learning_rate": 0.00038671608529944036,
      "loss": 1.4698,
      "step": 170
    },
    {
      "epoch": 0.16818293582493238,
      "grad_norm": 2.3899686336517334,
      "learning_rate": 0.0003863382455171098,
      "loss": 1.449,
      "step": 171
    },
    {
      "epoch": 0.16916646176542907,
      "grad_norm": 2.3972983360290527,
      "learning_rate": 0.00038595529717765027,
      "loss": 1.4645,
      "step": 172
    },
    {
      "epoch": 0.17014998770592574,
      "grad_norm": 2.5043587684631348,
      "learning_rate": 0.000385567250779784,
      "loss": 1.4809,
      "step": 173
    },
    {
      "epoch": 0.17113351364642243,
      "grad_norm": 2.449852466583252,
      "learning_rate": 0.000385174116961999,
      "loss": 1.5263,
      "step": 174
    },
    {
      "epoch": 0.1721170395869191,
      "grad_norm": 2.288686513900757,
      "learning_rate": 0.0003847759065022574,
      "loss": 1.4359,
      "step": 175
    },
    {
      "epoch": 0.1731005655274158,
      "grad_norm": 2.459357500076294,
      "learning_rate": 0.0003843726303177001,
      "loss": 1.3979,
      "step": 176
    },
    {
      "epoch": 0.17408409146791246,
      "grad_norm": 2.609467029571533,
      "learning_rate": 0.00038396429946434756,
      "loss": 1.5191,
      "step": 177
    },
    {
      "epoch": 0.17506761740840915,
      "grad_norm": 2.5471909046173096,
      "learning_rate": 0.00038355092513679624,
      "loss": 1.5023,
      "step": 178
    },
    {
      "epoch": 0.17605114334890581,
      "grad_norm": 2.462642192840576,
      "learning_rate": 0.0003831325186679123,
      "loss": 1.5181,
      "step": 179
    },
    {
      "epoch": 0.1770346692894025,
      "grad_norm": 2.422457456588745,
      "learning_rate": 0.0003827090915285202,
      "loss": 1.4656,
      "step": 180
    },
    {
      "epoch": 0.1780181952298992,
      "grad_norm": 2.5553433895111084,
      "learning_rate": 0.00038228065532708907,
      "loss": 1.3687,
      "step": 181
    },
    {
      "epoch": 0.17900172117039587,
      "grad_norm": 2.4368438720703125,
      "learning_rate": 0.00038184722180941376,
      "loss": 1.4777,
      "step": 182
    },
    {
      "epoch": 0.17998524711089256,
      "grad_norm": 2.573629140853882,
      "learning_rate": 0.000381408802858293,
      "loss": 1.4937,
      "step": 183
    },
    {
      "epoch": 0.18096877305138923,
      "grad_norm": 2.39176344871521,
      "learning_rate": 0.0003809654104932039,
      "loss": 1.5122,
      "step": 184
    },
    {
      "epoch": 0.18195229899188592,
      "grad_norm": 2.4057657718658447,
      "learning_rate": 0.00038051705686997214,
      "loss": 1.4093,
      "step": 185
    },
    {
      "epoch": 0.18293582493238258,
      "grad_norm": 2.285132646560669,
      "learning_rate": 0.00038006375428043876,
      "loss": 1.5037,
      "step": 186
    },
    {
      "epoch": 0.18391935087287928,
      "grad_norm": 2.273092269897461,
      "learning_rate": 0.00037960551515212314,
      "loss": 1.5205,
      "step": 187
    },
    {
      "epoch": 0.18490287681337594,
      "grad_norm": 2.4822018146514893,
      "learning_rate": 0.0003791423520478826,
      "loss": 1.425,
      "step": 188
    },
    {
      "epoch": 0.18588640275387264,
      "grad_norm": 2.3591558933258057,
      "learning_rate": 0.0003786742776655675,
      "loss": 1.4423,
      "step": 189
    },
    {
      "epoch": 0.1868699286943693,
      "grad_norm": 2.52620005607605,
      "learning_rate": 0.0003782013048376736,
      "loss": 1.5691,
      "step": 190
    },
    {
      "epoch": 0.187853454634866,
      "grad_norm": 2.53462290763855,
      "learning_rate": 0.0003777234465309898,
      "loss": 1.4321,
      "step": 191
    },
    {
      "epoch": 0.18883698057536266,
      "grad_norm": 2.546398878097534,
      "learning_rate": 0.00037724071584624295,
      "loss": 1.5338,
      "step": 192
    },
    {
      "epoch": 0.18982050651585936,
      "grad_norm": 2.478454828262329,
      "learning_rate": 0.00037675312601773874,
      "loss": 1.4099,
      "step": 193
    },
    {
      "epoch": 0.19080403245635605,
      "grad_norm": 2.3813908100128174,
      "learning_rate": 0.0003762606904129984,
      "loss": 1.4732,
      "step": 194
    },
    {
      "epoch": 0.19178755839685271,
      "grad_norm": 2.383133888244629,
      "learning_rate": 0.00037576342253239307,
      "loss": 1.5392,
      "step": 195
    },
    {
      "epoch": 0.1927710843373494,
      "grad_norm": 2.4089512825012207,
      "learning_rate": 0.0003752613360087727,
      "loss": 1.453,
      "step": 196
    },
    {
      "epoch": 0.19375461027784607,
      "grad_norm": 2.471236228942871,
      "learning_rate": 0.0003747544446070931,
      "loss": 1.4648,
      "step": 197
    },
    {
      "epoch": 0.19473813621834277,
      "grad_norm": 2.4847681522369385,
      "learning_rate": 0.0003742427622240379,
      "loss": 1.5095,
      "step": 198
    },
    {
      "epoch": 0.19572166215883943,
      "grad_norm": 2.397475004196167,
      "learning_rate": 0.00037372630288763826,
      "loss": 1.4638,
      "step": 199
    },
    {
      "epoch": 0.19670518809933613,
      "grad_norm": 2.5717852115631104,
      "learning_rate": 0.00037320508075688776,
      "loss": 1.5553,
      "step": 200
    },
    {
      "epoch": 0.1976887140398328,
      "grad_norm": 2.37680983543396,
      "learning_rate": 0.00037267911012135436,
      "loss": 1.4346,
      "step": 201
    },
    {
      "epoch": 0.19867223998032948,
      "grad_norm": 2.4877028465270996,
      "learning_rate": 0.0003721484054007888,
      "loss": 1.4983,
      "step": 202
    },
    {
      "epoch": 0.19965576592082615,
      "grad_norm": 2.3479998111724854,
      "learning_rate": 0.00037161298114472894,
      "loss": 1.4719,
      "step": 203
    },
    {
      "epoch": 0.20063929186132284,
      "grad_norm": 2.509665012359619,
      "learning_rate": 0.00037107285203210135,
      "loss": 1.4612,
      "step": 204
    },
    {
      "epoch": 0.2016228178018195,
      "grad_norm": 2.3066768646240234,
      "learning_rate": 0.00037052803287081844,
      "loss": 1.4312,
      "step": 205
    },
    {
      "epoch": 0.2026063437423162,
      "grad_norm": 2.68830943107605,
      "learning_rate": 0.00036997853859737284,
      "loss": 1.4549,
      "step": 206
    },
    {
      "epoch": 0.2035898696828129,
      "grad_norm": 2.4020886421203613,
      "learning_rate": 0.0003694243842764275,
      "loss": 1.4813,
      "step": 207
    },
    {
      "epoch": 0.20457339562330956,
      "grad_norm": 2.3946518898010254,
      "learning_rate": 0.00036886558510040304,
      "loss": 1.5203,
      "step": 208
    },
    {
      "epoch": 0.20555692156380626,
      "grad_norm": 2.25839900970459,
      "learning_rate": 0.00036830215638906125,
      "loss": 1.4594,
      "step": 209
    },
    {
      "epoch": 0.20654044750430292,
      "grad_norm": 2.254807710647583,
      "learning_rate": 0.00036773411358908485,
      "loss": 1.3341,
      "step": 210
    },
    {
      "epoch": 0.20752397344479961,
      "grad_norm": 2.3252182006835938,
      "learning_rate": 0.0003671614722736541,
      "loss": 1.4327,
      "step": 211
    },
    {
      "epoch": 0.20850749938529628,
      "grad_norm": 2.2827632427215576,
      "learning_rate": 0.0003665842481420199,
      "loss": 1.4615,
      "step": 212
    },
    {
      "epoch": 0.20949102532579297,
      "grad_norm": 2.349755048751831,
      "learning_rate": 0.0003660024570190735,
      "loss": 1.3929,
      "step": 213
    },
    {
      "epoch": 0.21047455126628964,
      "grad_norm": 2.340223789215088,
      "learning_rate": 0.00036541611485491234,
      "loss": 1.4348,
      "step": 214
    },
    {
      "epoch": 0.21145807720678633,
      "grad_norm": 2.3793411254882812,
      "learning_rate": 0.0003648252377244031,
      "loss": 1.5211,
      "step": 215
    },
    {
      "epoch": 0.212441603147283,
      "grad_norm": 2.3160605430603027,
      "learning_rate": 0.00036422984182674084,
      "loss": 1.5247,
      "step": 216
    },
    {
      "epoch": 0.2134251290877797,
      "grad_norm": 2.2637507915496826,
      "learning_rate": 0.0003636299434850047,
      "loss": 1.4333,
      "step": 217
    },
    {
      "epoch": 0.21440865502827636,
      "grad_norm": 2.3266677856445312,
      "learning_rate": 0.00036302555914571085,
      "loss": 1.436,
      "step": 218
    },
    {
      "epoch": 0.21539218096877305,
      "grad_norm": 2.220222234725952,
      "learning_rate": 0.00036241670537836125,
      "loss": 1.3747,
      "step": 219
    },
    {
      "epoch": 0.21637570690926974,
      "grad_norm": 2.197068691253662,
      "learning_rate": 0.0003618033988749895,
      "loss": 1.4659,
      "step": 220
    },
    {
      "epoch": 0.2173592328497664,
      "grad_norm": 2.255974054336548,
      "learning_rate": 0.00036118565644970317,
      "loss": 1.376,
      "step": 221
    },
    {
      "epoch": 0.2183427587902631,
      "grad_norm": 2.281059980392456,
      "learning_rate": 0.0003605634950382229,
      "loss": 1.5118,
      "step": 222
    },
    {
      "epoch": 0.21932628473075977,
      "grad_norm": 2.3769853115081787,
      "learning_rate": 0.00035993693169741815,
      "loss": 1.4705,
      "step": 223
    },
    {
      "epoch": 0.22030981067125646,
      "grad_norm": 2.311769962310791,
      "learning_rate": 0.00035930598360483926,
      "loss": 1.3907,
      "step": 224
    },
    {
      "epoch": 0.22129333661175313,
      "grad_norm": 2.2300007343292236,
      "learning_rate": 0.0003586706680582471,
      "loss": 1.5265,
      "step": 225
    },
    {
      "epoch": 0.22227686255224982,
      "grad_norm": 2.375342607498169,
      "learning_rate": 0.0003580310024751381,
      "loss": 1.5125,
      "step": 226
    },
    {
      "epoch": 0.2232603884927465,
      "grad_norm": 2.27280592918396,
      "learning_rate": 0.0003573870043922675,
      "loss": 1.4466,
      "step": 227
    },
    {
      "epoch": 0.22424391443324318,
      "grad_norm": 2.21986722946167,
      "learning_rate": 0.000356738691465168,
      "loss": 1.413,
      "step": 228
    },
    {
      "epoch": 0.22522744037373985,
      "grad_norm": 2.2521064281463623,
      "learning_rate": 0.000356086081467666,
      "loss": 1.3787,
      "step": 229
    },
    {
      "epoch": 0.22621096631423654,
      "grad_norm": 2.1412253379821777,
      "learning_rate": 0.0003554291922913942,
      "loss": 1.4381,
      "step": 230
    },
    {
      "epoch": 0.22719449225473323,
      "grad_norm": 2.179872989654541,
      "learning_rate": 0.0003547680419453013,
      "loss": 1.4138,
      "step": 231
    },
    {
      "epoch": 0.2281780181952299,
      "grad_norm": 2.1537280082702637,
      "learning_rate": 0.0003541026485551579,
      "loss": 1.5195,
      "step": 232
    },
    {
      "epoch": 0.2291615441357266,
      "grad_norm": 2.122849702835083,
      "learning_rate": 0.00035343303036306,
      "loss": 1.4646,
      "step": 233
    },
    {
      "epoch": 0.23014507007622326,
      "grad_norm": 2.27311372756958,
      "learning_rate": 0.0003527592057269285,
      "loss": 1.5041,
      "step": 234
    },
    {
      "epoch": 0.23112859601671995,
      "grad_norm": 2.1978180408477783,
      "learning_rate": 0.0003520811931200062,
      "loss": 1.449,
      "step": 235
    },
    {
      "epoch": 0.23211212195721662,
      "grad_norm": 2.2040724754333496,
      "learning_rate": 0.0003513990111303513,
      "loss": 1.4385,
      "step": 236
    },
    {
      "epoch": 0.2330956478977133,
      "grad_norm": 2.2172176837921143,
      "learning_rate": 0.0003507126784603276,
      "loss": 1.4733,
      "step": 237
    },
    {
      "epoch": 0.23407917383820998,
      "grad_norm": 2.3342642784118652,
      "learning_rate": 0.0003500222139260919,
      "loss": 1.4793,
      "step": 238
    },
    {
      "epoch": 0.23506269977870667,
      "grad_norm": 2.190760374069214,
      "learning_rate": 0.0003493276364570783,
      "loss": 1.4386,
      "step": 239
    },
    {
      "epoch": 0.23604622571920333,
      "grad_norm": 6.0800933837890625,
      "learning_rate": 0.00034862896509547887,
      "loss": 1.3766,
      "step": 240
    },
    {
      "epoch": 0.23702975165970003,
      "grad_norm": 2.31746768951416,
      "learning_rate": 0.000347926218995722,
      "loss": 1.4216,
      "step": 241
    },
    {
      "epoch": 0.2380132776001967,
      "grad_norm": 2.432718515396118,
      "learning_rate": 0.0003472194174239469,
      "loss": 1.391,
      "step": 242
    },
    {
      "epoch": 0.2389968035406934,
      "grad_norm": 2.181988477706909,
      "learning_rate": 0.0003465085797574758,
      "loss": 1.425,
      "step": 243
    },
    {
      "epoch": 0.23998032948119008,
      "grad_norm": 2.2645370960235596,
      "learning_rate": 0.00034579372548428235,
      "loss": 1.4169,
      "step": 244
    },
    {
      "epoch": 0.24096385542168675,
      "grad_norm": 2.313570261001587,
      "learning_rate": 0.0003450748742024575,
      "loss": 1.4122,
      "step": 245
    },
    {
      "epoch": 0.24194738136218344,
      "grad_norm": 2.3265764713287354,
      "learning_rate": 0.00034435204561967245,
      "loss": 1.4993,
      "step": 246
    },
    {
      "epoch": 0.2429309073026801,
      "grad_norm": 2.2329747676849365,
      "learning_rate": 0.00034362525955263776,
      "loss": 1.42,
      "step": 247
    },
    {
      "epoch": 0.2439144332431768,
      "grad_norm": 2.2317209243774414,
      "learning_rate": 0.00034289453592656067,
      "loss": 1.3151,
      "step": 248
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 2.1244490146636963,
      "learning_rate": 0.0003421598947745985,
      "loss": 1.3863,
      "step": 249
    },
    {
      "epoch": 0.24588148512417016,
      "grad_norm": 2.0842623710632324,
      "learning_rate": 0.0003414213562373095,
      "loss": 1.3731,
      "step": 250
    },
    {
      "epoch": 0.24588148512417016,
      "eval_loss": 1.3477705717086792,
      "eval_runtime": 485.4618,
      "eval_samples_per_second": 33.508,
      "eval_steps_per_second": 4.19,
      "step": 250
    },
    {
      "epoch": 0.24686501106466682,
      "grad_norm": 2.3793063163757324,
      "learning_rate": 0.00034067894056210077,
      "loss": 1.4192,
      "step": 251
    },
    {
      "epoch": 0.24784853700516352,
      "grad_norm": 2.0896759033203125,
      "learning_rate": 0.0003399326681026731,
      "loss": 1.3935,
      "step": 252
    },
    {
      "epoch": 0.24883206294566018,
      "grad_norm": 2.192542791366577,
      "learning_rate": 0.00033918255931846285,
      "loss": 1.3791,
      "step": 253
    },
    {
      "epoch": 0.24981558888615688,
      "grad_norm": 2.1628544330596924,
      "learning_rate": 0.0003384286347740814,
      "loss": 1.4235,
      "step": 254
    },
    {
      "epoch": 0.25079911482665357,
      "grad_norm": 2.170259475708008,
      "learning_rate": 0.0003376709151387508,
      "loss": 1.4244,
      "step": 255
    },
    {
      "epoch": 0.25178264076715023,
      "grad_norm": 2.1723480224609375,
      "learning_rate": 0.00033690942118573774,
      "loss": 1.4334,
      "step": 256
    },
    {
      "epoch": 0.2527661667076469,
      "grad_norm": 2.1078920364379883,
      "learning_rate": 0.0003361441737917836,
      "loss": 1.3699,
      "step": 257
    },
    {
      "epoch": 0.2537496926481436,
      "grad_norm": 2.2880465984344482,
      "learning_rate": 0.00033537519393653215,
      "loss": 1.4131,
      "step": 258
    },
    {
      "epoch": 0.2547332185886403,
      "grad_norm": 2.042609453201294,
      "learning_rate": 0.0003346025027019547,
      "loss": 1.4546,
      "step": 259
    },
    {
      "epoch": 0.25571674452913695,
      "grad_norm": 2.2044858932495117,
      "learning_rate": 0.00033382612127177166,
      "loss": 1.4071,
      "step": 260
    },
    {
      "epoch": 0.2567002704696336,
      "grad_norm": 2.1961300373077393,
      "learning_rate": 0.00033304607093087216,
      "loss": 1.4385,
      "step": 261
    },
    {
      "epoch": 0.25768379641013034,
      "grad_norm": 2.189028739929199,
      "learning_rate": 0.00033226237306473036,
      "loss": 1.3897,
      "step": 262
    },
    {
      "epoch": 0.258667322350627,
      "grad_norm": 2.0428004264831543,
      "learning_rate": 0.0003314750491588192,
      "loss": 1.4159,
      "step": 263
    },
    {
      "epoch": 0.25965084829112367,
      "grad_norm": 2.1762430667877197,
      "learning_rate": 0.00033068412079802114,
      "loss": 1.3774,
      "step": 264
    },
    {
      "epoch": 0.26063437423162034,
      "grad_norm": 2.0909976959228516,
      "learning_rate": 0.0003298896096660367,
      "loss": 1.4519,
      "step": 265
    },
    {
      "epoch": 0.26161790017211706,
      "grad_norm": 1.9835556745529175,
      "learning_rate": 0.00032909153754479014,
      "loss": 1.4166,
      "step": 266
    },
    {
      "epoch": 0.2626014261126137,
      "grad_norm": 2.1426315307617188,
      "learning_rate": 0.00032828992631383163,
      "loss": 1.4645,
      "step": 267
    },
    {
      "epoch": 0.2635849520531104,
      "grad_norm": 2.224727153778076,
      "learning_rate": 0.000327484797949738,
      "loss": 1.5332,
      "step": 268
    },
    {
      "epoch": 0.26456847799360705,
      "grad_norm": 2.110823154449463,
      "learning_rate": 0.00032667617452551,
      "loss": 1.4196,
      "step": 269
    },
    {
      "epoch": 0.2655520039341038,
      "grad_norm": 2.097383975982666,
      "learning_rate": 0.0003258640782099675,
      "loss": 1.3925,
      "step": 270
    },
    {
      "epoch": 0.26653552987460044,
      "grad_norm": 2.1051602363586426,
      "learning_rate": 0.0003250485312671411,
      "loss": 1.4343,
      "step": 271
    },
    {
      "epoch": 0.2675190558150971,
      "grad_norm": 1.9822003841400146,
      "learning_rate": 0.00032422955605566206,
      "loss": 1.3924,
      "step": 272
    },
    {
      "epoch": 0.2685025817555938,
      "grad_norm": 2.120588541030884,
      "learning_rate": 0.00032340717502814974,
      "loss": 1.361,
      "step": 273
    },
    {
      "epoch": 0.2694861076960905,
      "grad_norm": 2.0788121223449707,
      "learning_rate": 0.0003225814107305953,
      "loss": 1.3632,
      "step": 274
    },
    {
      "epoch": 0.27046963363658716,
      "grad_norm": 2.1129138469696045,
      "learning_rate": 0.00032175228580174416,
      "loss": 1.4318,
      "step": 275
    },
    {
      "epoch": 0.2714531595770838,
      "grad_norm": 2.093747854232788,
      "learning_rate": 0.000320919822972475,
      "loss": 1.3725,
      "step": 276
    },
    {
      "epoch": 0.27243668551758055,
      "grad_norm": 1.9742975234985352,
      "learning_rate": 0.00032008404506517684,
      "loss": 1.4021,
      "step": 277
    },
    {
      "epoch": 0.2734202114580772,
      "grad_norm": 2.12450909614563,
      "learning_rate": 0.0003192449749931232,
      "loss": 1.388,
      "step": 278
    },
    {
      "epoch": 0.2744037373985739,
      "grad_norm": 2.1144449710845947,
      "learning_rate": 0.00031840263575984393,
      "loss": 1.4413,
      "step": 279
    },
    {
      "epoch": 0.27538726333907054,
      "grad_norm": 2.036271333694458,
      "learning_rate": 0.00031755705045849464,
      "loss": 1.3833,
      "step": 280
    },
    {
      "epoch": 0.27637078927956726,
      "grad_norm": 2.0477890968322754,
      "learning_rate": 0.0003167082422712235,
      "loss": 1.4138,
      "step": 281
    },
    {
      "epoch": 0.27735431522006393,
      "grad_norm": 8.66737174987793,
      "learning_rate": 0.0003158562344685358,
      "loss": 1.3188,
      "step": 282
    },
    {
      "epoch": 0.2783378411605606,
      "grad_norm": 2.157360553741455,
      "learning_rate": 0.00031500105040865574,
      "loss": 1.3898,
      "step": 283
    },
    {
      "epoch": 0.2793213671010573,
      "grad_norm": 3.9456725120544434,
      "learning_rate": 0.0003141427135368864,
      "loss": 1.4494,
      "step": 284
    },
    {
      "epoch": 0.280304893041554,
      "grad_norm": 2.061131238937378,
      "learning_rate": 0.00031328124738496657,
      "loss": 1.4211,
      "step": 285
    },
    {
      "epoch": 0.28128841898205065,
      "grad_norm": 2.090033769607544,
      "learning_rate": 0.00031241667557042614,
      "loss": 1.477,
      "step": 286
    },
    {
      "epoch": 0.2822719449225473,
      "grad_norm": 2.136894464492798,
      "learning_rate": 0.0003115490217959381,
      "loss": 1.4745,
      "step": 287
    },
    {
      "epoch": 0.28325547086304403,
      "grad_norm": 2.0802109241485596,
      "learning_rate": 0.00031067830984866885,
      "loss": 1.3622,
      "step": 288
    },
    {
      "epoch": 0.2842389968035407,
      "grad_norm": 2.0720763206481934,
      "learning_rate": 0.0003098045635996264,
      "loss": 1.4252,
      "step": 289
    },
    {
      "epoch": 0.28522252274403737,
      "grad_norm": 2.1034626960754395,
      "learning_rate": 0.00030892780700300546,
      "loss": 1.491,
      "step": 290
    },
    {
      "epoch": 0.28620604868453403,
      "grad_norm": 2.192293405532837,
      "learning_rate": 0.000308048064095531,
      "loss": 1.4509,
      "step": 291
    },
    {
      "epoch": 0.28718957462503075,
      "grad_norm": 2.013444662094116,
      "learning_rate": 0.0003071653589957993,
      "loss": 1.4391,
      "step": 292
    },
    {
      "epoch": 0.2881731005655274,
      "grad_norm": 1.9928369522094727,
      "learning_rate": 0.00030627971590361657,
      "loss": 1.3496,
      "step": 293
    },
    {
      "epoch": 0.2891566265060241,
      "grad_norm": 2.0342020988464355,
      "learning_rate": 0.00030539115909933556,
      "loss": 1.4693,
      "step": 294
    },
    {
      "epoch": 0.29014015244652075,
      "grad_norm": 2.0666186809539795,
      "learning_rate": 0.0003044997129431898,
      "loss": 1.429,
      "step": 295
    },
    {
      "epoch": 0.29112367838701747,
      "grad_norm": 1.9321917295455933,
      "learning_rate": 0.00030360540187462606,
      "loss": 1.386,
      "step": 296
    },
    {
      "epoch": 0.29210720432751414,
      "grad_norm": 1.877564549446106,
      "learning_rate": 0.00030270825041163403,
      "loss": 1.3555,
      "step": 297
    },
    {
      "epoch": 0.2930907302680108,
      "grad_norm": 2.1787307262420654,
      "learning_rate": 0.0003018082831500743,
      "loss": 1.4266,
      "step": 298
    },
    {
      "epoch": 0.2940742562085075,
      "grad_norm": 2.0690128803253174,
      "learning_rate": 0.00030090552476300385,
      "loss": 1.4662,
      "step": 299
    },
    {
      "epoch": 0.2950577821490042,
      "grad_norm": 1.927424430847168,
      "learning_rate": 0.00030000000000000003,
      "loss": 1.3459,
      "step": 300
    },
    {
      "epoch": 0.29604130808950085,
      "grad_norm": 2.1272189617156982,
      "learning_rate": 0.00029909173368648155,
      "loss": 1.4379,
      "step": 301
    },
    {
      "epoch": 0.2970248340299975,
      "grad_norm": 2.015162467956543,
      "learning_rate": 0.00029818075072302817,
      "loss": 1.3685,
      "step": 302
    },
    {
      "epoch": 0.29800835997049424,
      "grad_norm": 2.1240289211273193,
      "learning_rate": 0.0002972670760846981,
      "loss": 1.2745,
      "step": 303
    },
    {
      "epoch": 0.2989918859109909,
      "grad_norm": 2.0548651218414307,
      "learning_rate": 0.00029635073482034305,
      "loss": 1.3977,
      "step": 304
    },
    {
      "epoch": 0.2999754118514876,
      "grad_norm": 2.206278085708618,
      "learning_rate": 0.0002954317520519217,
      "loss": 1.3421,
      "step": 305
    },
    {
      "epoch": 0.30095893779198424,
      "grad_norm": 2.087751626968384,
      "learning_rate": 0.00029451015297381086,
      "loss": 1.3093,
      "step": 306
    },
    {
      "epoch": 0.30194246373248096,
      "grad_norm": 2.1880125999450684,
      "learning_rate": 0.0002935859628521147,
      "loss": 1.3954,
      "step": 307
    },
    {
      "epoch": 0.3029259896729776,
      "grad_norm": 2.062269926071167,
      "learning_rate": 0.00029265920702397236,
      "loss": 1.3383,
      "step": 308
    },
    {
      "epoch": 0.3039095156134743,
      "grad_norm": 2.01125431060791,
      "learning_rate": 0.00029172991089686305,
      "loss": 1.3647,
      "step": 309
    },
    {
      "epoch": 0.304893041553971,
      "grad_norm": 2.0151026248931885,
      "learning_rate": 0.00029079809994790937,
      "loss": 1.3653,
      "step": 310
    },
    {
      "epoch": 0.3058765674944677,
      "grad_norm": 1.9694570302963257,
      "learning_rate": 0.0002898637997231793,
      "loss": 1.3755,
      "step": 311
    },
    {
      "epoch": 0.30686009343496434,
      "grad_norm": 2.021287202835083,
      "learning_rate": 0.0002889270358369855,
      "loss": 1.3605,
      "step": 312
    },
    {
      "epoch": 0.307843619375461,
      "grad_norm": 2.1369752883911133,
      "learning_rate": 0.00028798783397118304,
      "loss": 1.3946,
      "step": 313
    },
    {
      "epoch": 0.30882714531595773,
      "grad_norm": 2.0734684467315674,
      "learning_rate": 0.00028704621987446553,
      "loss": 1.3521,
      "step": 314
    },
    {
      "epoch": 0.3098106712564544,
      "grad_norm": 1.9757975339889526,
      "learning_rate": 0.00028610221936165904,
      "loss": 1.3109,
      "step": 315
    },
    {
      "epoch": 0.31079419719695106,
      "grad_norm": 2.0245602130889893,
      "learning_rate": 0.00028515585831301456,
      "loss": 1.3544,
      "step": 316
    },
    {
      "epoch": 0.3117777231374477,
      "grad_norm": 1.9701772928237915,
      "learning_rate": 0.00028420716267349823,
      "loss": 1.3829,
      "step": 317
    },
    {
      "epoch": 0.31276124907794445,
      "grad_norm": 2.049523115158081,
      "learning_rate": 0.00028325615845208025,
      "loss": 1.323,
      "step": 318
    },
    {
      "epoch": 0.3137447750184411,
      "grad_norm": 1.9660050868988037,
      "learning_rate": 0.00028230287172102176,
      "loss": 1.3779,
      "step": 319
    },
    {
      "epoch": 0.3147283009589378,
      "grad_norm": 1.9932433366775513,
      "learning_rate": 0.0002813473286151601,
      "loss": 1.3525,
      "step": 320
    },
    {
      "epoch": 0.3157118268994345,
      "grad_norm": 1.9653009176254272,
      "learning_rate": 0.0002803895553311921,
      "loss": 1.3884,
      "step": 321
    },
    {
      "epoch": 0.31669535283993117,
      "grad_norm": 1.9061921834945679,
      "learning_rate": 0.0002794295781269561,
      "loss": 1.4422,
      "step": 322
    },
    {
      "epoch": 0.31767887878042783,
      "grad_norm": 1.9004137516021729,
      "learning_rate": 0.0002784674233207123,
      "loss": 1.2992,
      "step": 323
    },
    {
      "epoch": 0.3186624047209245,
      "grad_norm": 1.9827345609664917,
      "learning_rate": 0.0002775031172904206,
      "loss": 1.2575,
      "step": 324
    },
    {
      "epoch": 0.3196459306614212,
      "grad_norm": 2.017831802368164,
      "learning_rate": 0.000276536686473018,
      "loss": 1.3544,
      "step": 325
    },
    {
      "epoch": 0.3206294566019179,
      "grad_norm": 2.120736598968506,
      "learning_rate": 0.00027556815736369343,
      "loss": 1.3659,
      "step": 326
    },
    {
      "epoch": 0.32161298254241455,
      "grad_norm": 1.9706817865371704,
      "learning_rate": 0.0002745975565151618,
      "loss": 1.375,
      "step": 327
    },
    {
      "epoch": 0.3225965084829112,
      "grad_norm": 1.8807820081710815,
      "learning_rate": 0.00027362491053693565,
      "loss": 1.3256,
      "step": 328
    },
    {
      "epoch": 0.32358003442340794,
      "grad_norm": 2.0242066383361816,
      "learning_rate": 0.0002726502460945957,
      "loss": 1.4179,
      "step": 329
    },
    {
      "epoch": 0.3245635603639046,
      "grad_norm": 1.9892431497573853,
      "learning_rate": 0.00027167358990906,
      "loss": 1.3494,
      "step": 330
    },
    {
      "epoch": 0.32554708630440127,
      "grad_norm": 1.9627022743225098,
      "learning_rate": 0.00027069496875585145,
      "loss": 1.3339,
      "step": 331
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 2.0491323471069336,
      "learning_rate": 0.0002697144094643631,
      "loss": 1.4254,
      "step": 332
    },
    {
      "epoch": 0.32751413818539465,
      "grad_norm": 2.0213217735290527,
      "learning_rate": 0.0002687319389171233,
      "loss": 1.4028,
      "step": 333
    },
    {
      "epoch": 0.3284976641258913,
      "grad_norm": 1.9927444458007812,
      "learning_rate": 0.00026774758404905834,
      "loss": 1.358,
      "step": 334
    },
    {
      "epoch": 0.329481190066388,
      "grad_norm": 1.909227967262268,
      "learning_rate": 0.0002667613718467542,
      "loss": 1.2623,
      "step": 335
    },
    {
      "epoch": 0.3304647160068847,
      "grad_norm": 1.9084049463272095,
      "learning_rate": 0.00026577332934771665,
      "loss": 1.3149,
      "step": 336
    },
    {
      "epoch": 0.3314482419473814,
      "grad_norm": 1.9376944303512573,
      "learning_rate": 0.0002647834836396299,
      "loss": 1.3785,
      "step": 337
    },
    {
      "epoch": 0.33243176788787804,
      "grad_norm": 2.0030970573425293,
      "learning_rate": 0.00026379186185961404,
      "loss": 1.3646,
      "step": 338
    },
    {
      "epoch": 0.3334152938283747,
      "grad_norm": 1.8497205972671509,
      "learning_rate": 0.00026279849119348103,
      "loss": 1.3441,
      "step": 339
    },
    {
      "epoch": 0.3343988197688714,
      "grad_norm": 1.9849061965942383,
      "learning_rate": 0.00026180339887498953,
      "loss": 1.3611,
      "step": 340
    },
    {
      "epoch": 0.3353823457093681,
      "grad_norm": 1.9657052755355835,
      "learning_rate": 0.00026080661218509806,
      "loss": 1.2945,
      "step": 341
    },
    {
      "epoch": 0.33636587164986476,
      "grad_norm": 1.883117914199829,
      "learning_rate": 0.00025980815845121736,
      "loss": 1.2645,
      "step": 342
    },
    {
      "epoch": 0.3373493975903614,
      "grad_norm": 1.8728495836257935,
      "learning_rate": 0.0002588080650464608,
      "loss": 1.2821,
      "step": 343
    },
    {
      "epoch": 0.33833292353085814,
      "grad_norm": 1.8817960023880005,
      "learning_rate": 0.0002578063593888943,
      "loss": 1.2699,
      "step": 344
    },
    {
      "epoch": 0.3393164494713548,
      "grad_norm": 1.8904342651367188,
      "learning_rate": 0.00025680306894078456,
      "loss": 1.3165,
      "step": 345
    },
    {
      "epoch": 0.3402999754118515,
      "grad_norm": 1.8854939937591553,
      "learning_rate": 0.0002557982212078459,
      "loss": 1.3442,
      "step": 346
    },
    {
      "epoch": 0.3412835013523482,
      "grad_norm": 1.922358512878418,
      "learning_rate": 0.00025479184373848656,
      "loss": 1.3505,
      "step": 347
    },
    {
      "epoch": 0.34226702729284486,
      "grad_norm": 1.958953619003296,
      "learning_rate": 0.00025378396412305314,
      "loss": 1.3626,
      "step": 348
    },
    {
      "epoch": 0.3432505532333415,
      "grad_norm": 1.9960256814956665,
      "learning_rate": 0.0002527746099930746,
      "loss": 1.3508,
      "step": 349
    },
    {
      "epoch": 0.3442340791738382,
      "grad_norm": 1.9428281784057617,
      "learning_rate": 0.00025176380902050413,
      "loss": 1.283,
      "step": 350
    },
    {
      "epoch": 0.3452176051143349,
      "grad_norm": 1.9332592487335205,
      "learning_rate": 0.00025075158891696116,
      "loss": 1.3164,
      "step": 351
    },
    {
      "epoch": 0.3462011310548316,
      "grad_norm": 1.8473782539367676,
      "learning_rate": 0.000249737977432971,
      "loss": 1.3252,
      "step": 352
    },
    {
      "epoch": 0.34718465699532824,
      "grad_norm": 1.8220369815826416,
      "learning_rate": 0.0002487230023572045,
      "loss": 1.3026,
      "step": 353
    },
    {
      "epoch": 0.3481681829358249,
      "grad_norm": 1.8444721698760986,
      "learning_rate": 0.0002477066915157162,
      "loss": 1.2195,
      "step": 354
    },
    {
      "epoch": 0.34915170887632163,
      "grad_norm": 1.9033790826797485,
      "learning_rate": 0.00024668907277118114,
      "loss": 1.2647,
      "step": 355
    },
    {
      "epoch": 0.3501352348168183,
      "grad_norm": 1.8377138376235962,
      "learning_rate": 0.00024567017402213115,
      "loss": 1.2549,
      "step": 356
    },
    {
      "epoch": 0.35111876075731496,
      "grad_norm": 1.8567314147949219,
      "learning_rate": 0.0002446500232021903,
      "loss": 1.3251,
      "step": 357
    },
    {
      "epoch": 0.35210228669781163,
      "grad_norm": 2.01230525970459,
      "learning_rate": 0.00024362864827930856,
      "loss": 1.3699,
      "step": 358
    },
    {
      "epoch": 0.35308581263830835,
      "grad_norm": 2.007488250732422,
      "learning_rate": 0.00024260607725499534,
      "loss": 1.288,
      "step": 359
    },
    {
      "epoch": 0.354069338578805,
      "grad_norm": 1.8876938819885254,
      "learning_rate": 0.00024158233816355185,
      "loss": 1.3251,
      "step": 360
    },
    {
      "epoch": 0.3550528645193017,
      "grad_norm": 1.8500109910964966,
      "learning_rate": 0.0002405574590713025,
      "loss": 1.2793,
      "step": 361
    },
    {
      "epoch": 0.3560363904597984,
      "grad_norm": 1.9820373058319092,
      "learning_rate": 0.00023953146807582525,
      "loss": 1.3372,
      "step": 362
    },
    {
      "epoch": 0.35701991640029507,
      "grad_norm": 1.942043423652649,
      "learning_rate": 0.0002385043933051815,
      "loss": 1.2589,
      "step": 363
    },
    {
      "epoch": 0.35800344234079173,
      "grad_norm": 1.775593876838684,
      "learning_rate": 0.00023747626291714498,
      "loss": 1.2421,
      "step": 364
    },
    {
      "epoch": 0.3589869682812884,
      "grad_norm": 1.8077837228775024,
      "learning_rate": 0.00023644710509842956,
      "loss": 1.3026,
      "step": 365
    },
    {
      "epoch": 0.3599704942217851,
      "grad_norm": 1.86480712890625,
      "learning_rate": 0.0002354169480639167,
      "loss": 1.3056,
      "step": 366
    },
    {
      "epoch": 0.3609540201622818,
      "grad_norm": 1.8518476486206055,
      "learning_rate": 0.0002343858200558819,
      "loss": 1.2311,
      "step": 367
    },
    {
      "epoch": 0.36193754610277845,
      "grad_norm": 1.7495702505111694,
      "learning_rate": 0.00023335374934322048,
      "loss": 1.1926,
      "step": 368
    },
    {
      "epoch": 0.3629210720432751,
      "grad_norm": 1.8048973083496094,
      "learning_rate": 0.00023232076422067226,
      "loss": 1.286,
      "step": 369
    },
    {
      "epoch": 0.36390459798377184,
      "grad_norm": 1.860206127166748,
      "learning_rate": 0.0002312868930080462,
      "loss": 1.2948,
      "step": 370
    },
    {
      "epoch": 0.3648881239242685,
      "grad_norm": 1.9511946439743042,
      "learning_rate": 0.0002302521640494439,
      "loss": 1.2746,
      "step": 371
    },
    {
      "epoch": 0.36587164986476517,
      "grad_norm": 1.7899302244186401,
      "learning_rate": 0.00022921660571248236,
      "loss": 1.3213,
      "step": 372
    },
    {
      "epoch": 0.3668551758052619,
      "grad_norm": 1.8126307725906372,
      "learning_rate": 0.00022818024638751654,
      "loss": 1.2423,
      "step": 373
    },
    {
      "epoch": 0.36783870174575856,
      "grad_norm": 1.8500957489013672,
      "learning_rate": 0.0002271431144868609,
      "loss": 1.27,
      "step": 374
    },
    {
      "epoch": 0.3688222276862552,
      "grad_norm": 1.8999929428100586,
      "learning_rate": 0.00022610523844401038,
      "loss": 1.3039,
      "step": 375
    },
    {
      "epoch": 0.3698057536267519,
      "grad_norm": 1.7920165061950684,
      "learning_rate": 0.00022506664671286087,
      "loss": 1.2656,
      "step": 376
    },
    {
      "epoch": 0.3707892795672486,
      "grad_norm": 1.8043848276138306,
      "learning_rate": 0.00022402736776692942,
      "loss": 1.2865,
      "step": 377
    },
    {
      "epoch": 0.3717728055077453,
      "grad_norm": 1.863116979598999,
      "learning_rate": 0.00022298743009857336,
      "loss": 1.224,
      "step": 378
    },
    {
      "epoch": 0.37275633144824194,
      "grad_norm": 1.868927240371704,
      "learning_rate": 0.00022194686221820904,
      "loss": 1.2817,
      "step": 379
    },
    {
      "epoch": 0.3737398573887386,
      "grad_norm": 1.9089443683624268,
      "learning_rate": 0.00022090569265353072,
      "loss": 1.2431,
      "step": 380
    },
    {
      "epoch": 0.3747233833292353,
      "grad_norm": 1.7925503253936768,
      "learning_rate": 0.00021986394994872783,
      "loss": 1.2739,
      "step": 381
    },
    {
      "epoch": 0.375706909269732,
      "grad_norm": 1.8950023651123047,
      "learning_rate": 0.00021882166266370292,
      "loss": 1.2881,
      "step": 382
    },
    {
      "epoch": 0.37669043521022866,
      "grad_norm": 1.7854597568511963,
      "learning_rate": 0.00021777885937328834,
      "loss": 1.2756,
      "step": 383
    },
    {
      "epoch": 0.3776739611507253,
      "grad_norm": 1.822982668876648,
      "learning_rate": 0.00021673556866646317,
      "loss": 1.2949,
      "step": 384
    },
    {
      "epoch": 0.37865748709122204,
      "grad_norm": 1.853337049484253,
      "learning_rate": 0.000215691819145569,
      "loss": 1.2969,
      "step": 385
    },
    {
      "epoch": 0.3796410130317187,
      "grad_norm": 1.8197740316390991,
      "learning_rate": 0.00021464763942552636,
      "loss": 1.2784,
      "step": 386
    },
    {
      "epoch": 0.3806245389722154,
      "grad_norm": 1.7212492227554321,
      "learning_rate": 0.00021360305813304965,
      "loss": 1.2589,
      "step": 387
    },
    {
      "epoch": 0.3816080649127121,
      "grad_norm": 1.9747990369796753,
      "learning_rate": 0.0002125581039058627,
      "loss": 1.3955,
      "step": 388
    },
    {
      "epoch": 0.38259159085320876,
      "grad_norm": 1.7264504432678223,
      "learning_rate": 0.0002115128053919135,
      "loss": 1.2901,
      "step": 389
    },
    {
      "epoch": 0.38357511679370543,
      "grad_norm": 1.7033997774124146,
      "learning_rate": 0.0002104671912485888,
      "loss": 1.2674,
      "step": 390
    },
    {
      "epoch": 0.3845586427342021,
      "grad_norm": 1.7432687282562256,
      "learning_rate": 0.00020942129014192852,
      "loss": 1.2087,
      "step": 391
    },
    {
      "epoch": 0.3855421686746988,
      "grad_norm": 1.7505927085876465,
      "learning_rate": 0.00020837513074583992,
      "loss": 1.2362,
      "step": 392
    },
    {
      "epoch": 0.3865256946151955,
      "grad_norm": 1.8046858310699463,
      "learning_rate": 0.00020732874174131128,
      "loss": 1.2497,
      "step": 393
    },
    {
      "epoch": 0.38750922055569215,
      "grad_norm": 1.7545992136001587,
      "learning_rate": 0.00020628215181562568,
      "loss": 1.2616,
      "step": 394
    },
    {
      "epoch": 0.3884927464961888,
      "grad_norm": 1.7934821844100952,
      "learning_rate": 0.00020523538966157466,
      "loss": 1.2024,
      "step": 395
    },
    {
      "epoch": 0.38947627243668553,
      "grad_norm": 1.8329651355743408,
      "learning_rate": 0.0002041884839766714,
      "loss": 1.2064,
      "step": 396
    },
    {
      "epoch": 0.3904597983771822,
      "grad_norm": 1.8631696701049805,
      "learning_rate": 0.00020314146346236416,
      "loss": 1.2507,
      "step": 397
    },
    {
      "epoch": 0.39144332431767886,
      "grad_norm": 1.8060160875320435,
      "learning_rate": 0.00020209435682324916,
      "loss": 1.3107,
      "step": 398
    },
    {
      "epoch": 0.3924268502581756,
      "grad_norm": 1.8386930227279663,
      "learning_rate": 0.0002010471927662839,
      "loss": 1.2402,
      "step": 399
    },
    {
      "epoch": 0.39341037619867225,
      "grad_norm": 1.828965663909912,
      "learning_rate": 0.0002,
      "loss": 1.2084,
      "step": 400
    },
    {
      "epoch": 0.3943939021391689,
      "grad_norm": 1.787771463394165,
      "learning_rate": 0.00019895280723371608,
      "loss": 1.1763,
      "step": 401
    },
    {
      "epoch": 0.3953774280796656,
      "grad_norm": 1.8663722276687622,
      "learning_rate": 0.0001979056431767509,
      "loss": 1.2559,
      "step": 402
    },
    {
      "epoch": 0.3963609540201623,
      "grad_norm": 1.8643670082092285,
      "learning_rate": 0.0001968585365376359,
      "loss": 1.2626,
      "step": 403
    },
    {
      "epoch": 0.39734447996065897,
      "grad_norm": 1.8276618719100952,
      "learning_rate": 0.00019581151602332864,
      "loss": 1.2552,
      "step": 404
    },
    {
      "epoch": 0.39832800590115564,
      "grad_norm": 1.8075330257415771,
      "learning_rate": 0.0001947646103384254,
      "loss": 1.2119,
      "step": 405
    },
    {
      "epoch": 0.3993115318416523,
      "grad_norm": 1.7703173160552979,
      "learning_rate": 0.00019371784818437434,
      "loss": 1.2228,
      "step": 406
    },
    {
      "epoch": 0.400295057782149,
      "grad_norm": 1.815887451171875,
      "learning_rate": 0.0001926712582586887,
      "loss": 1.2183,
      "step": 407
    },
    {
      "epoch": 0.4012785837226457,
      "grad_norm": 1.7415165901184082,
      "learning_rate": 0.00019162486925416013,
      "loss": 1.2408,
      "step": 408
    },
    {
      "epoch": 0.40226210966314235,
      "grad_norm": 1.7467395067214966,
      "learning_rate": 0.0001905787098580715,
      "loss": 1.1907,
      "step": 409
    },
    {
      "epoch": 0.403245635603639,
      "grad_norm": 1.9408924579620361,
      "learning_rate": 0.00018953280875141123,
      "loss": 1.1885,
      "step": 410
    },
    {
      "epoch": 0.40422916154413574,
      "grad_norm": 1.7087867259979248,
      "learning_rate": 0.00018848719460808658,
      "loss": 1.2729,
      "step": 411
    },
    {
      "epoch": 0.4052126874846324,
      "grad_norm": 1.8479716777801514,
      "learning_rate": 0.00018744189609413734,
      "loss": 1.2582,
      "step": 412
    },
    {
      "epoch": 0.40619621342512907,
      "grad_norm": 1.7879359722137451,
      "learning_rate": 0.00018639694186695045,
      "loss": 1.2412,
      "step": 413
    },
    {
      "epoch": 0.4071797393656258,
      "grad_norm": 1.781746506690979,
      "learning_rate": 0.00018535236057447372,
      "loss": 1.2207,
      "step": 414
    },
    {
      "epoch": 0.40816326530612246,
      "grad_norm": 1.6866943836212158,
      "learning_rate": 0.00018430818085443104,
      "loss": 1.1498,
      "step": 415
    },
    {
      "epoch": 0.4091467912466191,
      "grad_norm": 1.804580807685852,
      "learning_rate": 0.00018326443133353693,
      "loss": 1.2262,
      "step": 416
    },
    {
      "epoch": 0.4101303171871158,
      "grad_norm": 1.7881170511245728,
      "learning_rate": 0.0001822211406267117,
      "loss": 1.2344,
      "step": 417
    },
    {
      "epoch": 0.4111138431276125,
      "grad_norm": 1.8688361644744873,
      "learning_rate": 0.00018117833733629715,
      "loss": 1.2515,
      "step": 418
    },
    {
      "epoch": 0.4120973690681092,
      "grad_norm": 1.7373318672180176,
      "learning_rate": 0.00018013605005127225,
      "loss": 1.2245,
      "step": 419
    },
    {
      "epoch": 0.41308089500860584,
      "grad_norm": 1.7880922555923462,
      "learning_rate": 0.00017909430734646935,
      "loss": 1.2435,
      "step": 420
    },
    {
      "epoch": 0.4140644209491025,
      "grad_norm": 1.7909986972808838,
      "learning_rate": 0.00017805313778179095,
      "loss": 1.2218,
      "step": 421
    },
    {
      "epoch": 0.41504794688959923,
      "grad_norm": 1.826875925064087,
      "learning_rate": 0.00017701256990142671,
      "loss": 1.2784,
      "step": 422
    },
    {
      "epoch": 0.4160314728300959,
      "grad_norm": 1.7037346363067627,
      "learning_rate": 0.0001759726322330706,
      "loss": 1.1891,
      "step": 423
    },
    {
      "epoch": 0.41701499877059256,
      "grad_norm": 1.8292216062545776,
      "learning_rate": 0.00017493335328713913,
      "loss": 1.2159,
      "step": 424
    },
    {
      "epoch": 0.4179985247110893,
      "grad_norm": 1.8551576137542725,
      "learning_rate": 0.00017389476155598975,
      "loss": 1.1899,
      "step": 425
    },
    {
      "epoch": 0.41898205065158595,
      "grad_norm": 1.7008535861968994,
      "learning_rate": 0.00017285688551313915,
      "loss": 1.2458,
      "step": 426
    },
    {
      "epoch": 0.4199655765920826,
      "grad_norm": 1.7437422275543213,
      "learning_rate": 0.00017181975361248348,
      "loss": 1.2477,
      "step": 427
    },
    {
      "epoch": 0.4209491025325793,
      "grad_norm": 1.7090297937393188,
      "learning_rate": 0.0001707833942875177,
      "loss": 1.2404,
      "step": 428
    },
    {
      "epoch": 0.421932628473076,
      "grad_norm": 1.7244435548782349,
      "learning_rate": 0.00016974783595055616,
      "loss": 1.2698,
      "step": 429
    },
    {
      "epoch": 0.42291615441357266,
      "grad_norm": 1.6990344524383545,
      "learning_rate": 0.00016871310699195379,
      "loss": 1.197,
      "step": 430
    },
    {
      "epoch": 0.42389968035406933,
      "grad_norm": 1.715484857559204,
      "learning_rate": 0.00016767923577932782,
      "loss": 1.196,
      "step": 431
    },
    {
      "epoch": 0.424883206294566,
      "grad_norm": 1.880728840827942,
      "learning_rate": 0.00016664625065677957,
      "loss": 1.2737,
      "step": 432
    },
    {
      "epoch": 0.4258667322350627,
      "grad_norm": 1.796807885169983,
      "learning_rate": 0.00016561417994411808,
      "loss": 1.2194,
      "step": 433
    },
    {
      "epoch": 0.4268502581755594,
      "grad_norm": 1.7898166179656982,
      "learning_rate": 0.00016458305193608336,
      "loss": 1.1725,
      "step": 434
    },
    {
      "epoch": 0.42783378411605605,
      "grad_norm": 1.692898154258728,
      "learning_rate": 0.0001635528949015705,
      "loss": 1.1763,
      "step": 435
    },
    {
      "epoch": 0.4288173100565527,
      "grad_norm": 1.8150991201400757,
      "learning_rate": 0.00016252373708285504,
      "loss": 1.1773,
      "step": 436
    },
    {
      "epoch": 0.42980083599704944,
      "grad_norm": 1.6946556568145752,
      "learning_rate": 0.00016149560669481857,
      "loss": 1.2195,
      "step": 437
    },
    {
      "epoch": 0.4307843619375461,
      "grad_norm": 1.809956669807434,
      "learning_rate": 0.00016046853192417477,
      "loss": 1.2104,
      "step": 438
    },
    {
      "epoch": 0.43176788787804277,
      "grad_norm": 1.729413390159607,
      "learning_rate": 0.00015944254092869757,
      "loss": 1.2392,
      "step": 439
    },
    {
      "epoch": 0.4327514138185395,
      "grad_norm": 1.7889063358306885,
      "learning_rate": 0.00015841766183644817,
      "loss": 1.2077,
      "step": 440
    },
    {
      "epoch": 0.43373493975903615,
      "grad_norm": 1.794033169746399,
      "learning_rate": 0.0001573939227450047,
      "loss": 1.1854,
      "step": 441
    },
    {
      "epoch": 0.4347184656995328,
      "grad_norm": 1.7685823440551758,
      "learning_rate": 0.00015637135172069154,
      "loss": 1.2152,
      "step": 442
    },
    {
      "epoch": 0.4357019916400295,
      "grad_norm": 1.743825078010559,
      "learning_rate": 0.00015534997679780976,
      "loss": 1.1843,
      "step": 443
    },
    {
      "epoch": 0.4366855175805262,
      "grad_norm": 1.795933723449707,
      "learning_rate": 0.00015432982597786884,
      "loss": 1.221,
      "step": 444
    },
    {
      "epoch": 0.43766904352102287,
      "grad_norm": 1.720615029335022,
      "learning_rate": 0.000153310927228819,
      "loss": 1.2354,
      "step": 445
    },
    {
      "epoch": 0.43865256946151954,
      "grad_norm": 1.676834225654602,
      "learning_rate": 0.00015229330848428387,
      "loss": 1.1811,
      "step": 446
    },
    {
      "epoch": 0.4396360954020162,
      "grad_norm": 1.7972372770309448,
      "learning_rate": 0.00015127699764279552,
      "loss": 1.2239,
      "step": 447
    },
    {
      "epoch": 0.4406196213425129,
      "grad_norm": 1.6563159227371216,
      "learning_rate": 0.00015026202256702908,
      "loss": 1.1704,
      "step": 448
    },
    {
      "epoch": 0.4416031472830096,
      "grad_norm": 1.6985008716583252,
      "learning_rate": 0.00014924841108303888,
      "loss": 1.2249,
      "step": 449
    },
    {
      "epoch": 0.44258667322350626,
      "grad_norm": 1.6926326751708984,
      "learning_rate": 0.00014823619097949583,
      "loss": 1.2331,
      "step": 450
    },
    {
      "epoch": 0.443570199164003,
      "grad_norm": 1.748921513557434,
      "learning_rate": 0.00014722539000692548,
      "loss": 1.1964,
      "step": 451
    },
    {
      "epoch": 0.44455372510449964,
      "grad_norm": 1.7176635265350342,
      "learning_rate": 0.00014621603587694688,
      "loss": 1.2599,
      "step": 452
    },
    {
      "epoch": 0.4455372510449963,
      "grad_norm": 1.6862727403640747,
      "learning_rate": 0.00014520815626151352,
      "loss": 1.206,
      "step": 453
    },
    {
      "epoch": 0.446520776985493,
      "grad_norm": 1.7514681816101074,
      "learning_rate": 0.00014420177879215416,
      "loss": 1.123,
      "step": 454
    },
    {
      "epoch": 0.4475043029259897,
      "grad_norm": 1.713309407234192,
      "learning_rate": 0.00014319693105921549,
      "loss": 1.1877,
      "step": 455
    },
    {
      "epoch": 0.44848782886648636,
      "grad_norm": 1.7206729650497437,
      "learning_rate": 0.00014219364061110565,
      "loss": 1.16,
      "step": 456
    },
    {
      "epoch": 0.449471354806983,
      "grad_norm": 1.6439610719680786,
      "learning_rate": 0.00014119193495353924,
      "loss": 1.1667,
      "step": 457
    },
    {
      "epoch": 0.4504548807474797,
      "grad_norm": 1.7147465944290161,
      "learning_rate": 0.00014019184154878269,
      "loss": 1.2132,
      "step": 458
    },
    {
      "epoch": 0.4514384066879764,
      "grad_norm": 1.7215044498443604,
      "learning_rate": 0.00013919338781490193,
      "loss": 1.1353,
      "step": 459
    },
    {
      "epoch": 0.4524219326284731,
      "grad_norm": 1.7469907999038696,
      "learning_rate": 0.00013819660112501054,
      "loss": 1.1428,
      "step": 460
    },
    {
      "epoch": 0.45340545856896974,
      "grad_norm": 1.69190514087677,
      "learning_rate": 0.00013720150880651902,
      "loss": 1.1902,
      "step": 461
    },
    {
      "epoch": 0.45438898450946646,
      "grad_norm": 1.6948457956314087,
      "learning_rate": 0.000136208138140386,
      "loss": 1.1947,
      "step": 462
    },
    {
      "epoch": 0.45537251044996313,
      "grad_norm": 1.81825852394104,
      "learning_rate": 0.00013521651636037017,
      "loss": 1.248,
      "step": 463
    },
    {
      "epoch": 0.4563560363904598,
      "grad_norm": 1.6670900583267212,
      "learning_rate": 0.00013422667065228337,
      "loss": 1.1704,
      "step": 464
    },
    {
      "epoch": 0.45733956233095646,
      "grad_norm": 1.736108422279358,
      "learning_rate": 0.00013323862815324585,
      "loss": 1.2052,
      "step": 465
    },
    {
      "epoch": 0.4583230882714532,
      "grad_norm": 1.6330331563949585,
      "learning_rate": 0.00013225241595094173,
      "loss": 1.1926,
      "step": 466
    },
    {
      "epoch": 0.45930661421194985,
      "grad_norm": 1.6478614807128906,
      "learning_rate": 0.0001312680610828768,
      "loss": 1.1915,
      "step": 467
    },
    {
      "epoch": 0.4602901401524465,
      "grad_norm": 1.6963731050491333,
      "learning_rate": 0.000130285590535637,
      "loss": 1.1436,
      "step": 468
    },
    {
      "epoch": 0.4612736660929432,
      "grad_norm": 1.7935936450958252,
      "learning_rate": 0.00012930503124414862,
      "loss": 1.1631,
      "step": 469
    },
    {
      "epoch": 0.4622571920334399,
      "grad_norm": 1.664107322692871,
      "learning_rate": 0.00012832641009093994,
      "loss": 1.2375,
      "step": 470
    },
    {
      "epoch": 0.46324071797393657,
      "grad_norm": 1.6989500522613525,
      "learning_rate": 0.00012734975390540435,
      "loss": 1.1917,
      "step": 471
    },
    {
      "epoch": 0.46422424391443323,
      "grad_norm": 1.6915833950042725,
      "learning_rate": 0.00012637508946306442,
      "loss": 1.166,
      "step": 472
    },
    {
      "epoch": 0.4652077698549299,
      "grad_norm": 1.6191935539245605,
      "learning_rate": 0.0001254024434848382,
      "loss": 1.2512,
      "step": 473
    },
    {
      "epoch": 0.4661912957954266,
      "grad_norm": 1.7693103551864624,
      "learning_rate": 0.0001244318426363066,
      "loss": 1.1593,
      "step": 474
    },
    {
      "epoch": 0.4671748217359233,
      "grad_norm": 1.6574156284332275,
      "learning_rate": 0.00012346331352698205,
      "loss": 1.2136,
      "step": 475
    },
    {
      "epoch": 0.46815834767641995,
      "grad_norm": 1.6650640964508057,
      "learning_rate": 0.0001224968827095794,
      "loss": 1.1612,
      "step": 476
    },
    {
      "epoch": 0.46914187361691667,
      "grad_norm": 1.7591502666473389,
      "learning_rate": 0.00012153257667928777,
      "loss": 1.2084,
      "step": 477
    },
    {
      "epoch": 0.47012539955741334,
      "grad_norm": 1.683616280555725,
      "learning_rate": 0.00012057042187304391,
      "loss": 1.1276,
      "step": 478
    },
    {
      "epoch": 0.47110892549791,
      "grad_norm": 1.7206487655639648,
      "learning_rate": 0.000119610444668808,
      "loss": 1.1288,
      "step": 479
    },
    {
      "epoch": 0.47209245143840667,
      "grad_norm": 1.6581438779830933,
      "learning_rate": 0.00011865267138483999,
      "loss": 1.1834,
      "step": 480
    },
    {
      "epoch": 0.4730759773789034,
      "grad_norm": 1.7817555665969849,
      "learning_rate": 0.00011769712827897825,
      "loss": 1.2167,
      "step": 481
    },
    {
      "epoch": 0.47405950331940006,
      "grad_norm": 1.7214207649230957,
      "learning_rate": 0.0001167438415479198,
      "loss": 1.1318,
      "step": 482
    },
    {
      "epoch": 0.4750430292598967,
      "grad_norm": 1.6356815099716187,
      "learning_rate": 0.00011579283732650181,
      "loss": 1.1829,
      "step": 483
    },
    {
      "epoch": 0.4760265552003934,
      "grad_norm": 1.6239361763000488,
      "learning_rate": 0.00011484414168698547,
      "loss": 1.1651,
      "step": 484
    },
    {
      "epoch": 0.4770100811408901,
      "grad_norm": 1.6749591827392578,
      "learning_rate": 0.00011389778063834094,
      "loss": 1.0833,
      "step": 485
    },
    {
      "epoch": 0.4779936070813868,
      "grad_norm": 1.711772084236145,
      "learning_rate": 0.00011295378012553453,
      "loss": 1.1464,
      "step": 486
    },
    {
      "epoch": 0.47897713302188344,
      "grad_norm": 1.7215913534164429,
      "learning_rate": 0.00011201216602881697,
      "loss": 1.1792,
      "step": 487
    },
    {
      "epoch": 0.47996065896238016,
      "grad_norm": 1.6231255531311035,
      "learning_rate": 0.00011107296416301456,
      "loss": 1.2042,
      "step": 488
    },
    {
      "epoch": 0.4809441849028768,
      "grad_norm": 1.71650230884552,
      "learning_rate": 0.00011013620027682073,
      "loss": 1.1876,
      "step": 489
    },
    {
      "epoch": 0.4819277108433735,
      "grad_norm": 1.6076571941375732,
      "learning_rate": 0.00010920190005209065,
      "loss": 1.175,
      "step": 490
    },
    {
      "epoch": 0.48291123678387016,
      "grad_norm": 1.6567050218582153,
      "learning_rate": 0.00010827008910313709,
      "loss": 1.1988,
      "step": 491
    },
    {
      "epoch": 0.4838947627243669,
      "grad_norm": 1.6528984308242798,
      "learning_rate": 0.00010734079297602771,
      "loss": 1.1188,
      "step": 492
    },
    {
      "epoch": 0.48487828866486354,
      "grad_norm": 1.6168268918991089,
      "learning_rate": 0.00010641403714788536,
      "loss": 1.1963,
      "step": 493
    },
    {
      "epoch": 0.4858618146053602,
      "grad_norm": 1.6533197164535522,
      "learning_rate": 0.00010548984702618922,
      "loss": 1.1302,
      "step": 494
    },
    {
      "epoch": 0.4868453405458569,
      "grad_norm": 1.6218030452728271,
      "learning_rate": 0.00010456824794807831,
      "loss": 1.1593,
      "step": 495
    },
    {
      "epoch": 0.4878288664863536,
      "grad_norm": 1.6518218517303467,
      "learning_rate": 0.00010364926517965692,
      "loss": 1.179,
      "step": 496
    },
    {
      "epoch": 0.48881239242685026,
      "grad_norm": 1.7298614978790283,
      "learning_rate": 0.00010273292391530191,
      "loss": 1.1633,
      "step": 497
    },
    {
      "epoch": 0.4897959183673469,
      "grad_norm": 1.6431043148040771,
      "learning_rate": 0.00010181924927697184,
      "loss": 1.1147,
      "step": 498
    },
    {
      "epoch": 0.4907794443078436,
      "grad_norm": 1.6055521965026855,
      "learning_rate": 0.00010090826631351848,
      "loss": 1.151,
      "step": 499
    },
    {
      "epoch": 0.4917629702483403,
      "grad_norm": 1.6847048997879028,
      "learning_rate": 0.00010000000000000005,
      "loss": 1.1308,
      "step": 500
    },
    {
      "epoch": 0.4917629702483403,
      "eval_loss": 1.0859394073486328,
      "eval_runtime": 484.884,
      "eval_samples_per_second": 33.548,
      "eval_steps_per_second": 4.195,
      "step": 500
    },
    {
      "epoch": 0.492746496188837,
      "grad_norm": 1.7722065448760986,
      "learning_rate": 9.909447523699617e-05,
      "loss": 1.142,
      "step": 501
    },
    {
      "epoch": 0.49373002212933365,
      "grad_norm": 1.684760332107544,
      "learning_rate": 9.819171684992574e-05,
      "loss": 1.1404,
      "step": 502
    },
    {
      "epoch": 0.49471354806983037,
      "grad_norm": 1.6577117443084717,
      "learning_rate": 9.729174958836604e-05,
      "loss": 1.1942,
      "step": 503
    },
    {
      "epoch": 0.49569707401032703,
      "grad_norm": 1.6315124034881592,
      "learning_rate": 9.639459812537399e-05,
      "loss": 1.1396,
      "step": 504
    },
    {
      "epoch": 0.4966805999508237,
      "grad_norm": 1.611763834953308,
      "learning_rate": 9.550028705681025e-05,
      "loss": 1.0717,
      "step": 505
    },
    {
      "epoch": 0.49766412589132036,
      "grad_norm": 1.792077660560608,
      "learning_rate": 9.460884090066449e-05,
      "loss": 1.1422,
      "step": 506
    },
    {
      "epoch": 0.4986476518318171,
      "grad_norm": 1.6460882425308228,
      "learning_rate": 9.37202840963834e-05,
      "loss": 1.1549,
      "step": 507
    },
    {
      "epoch": 0.49963117777231375,
      "grad_norm": 1.6016443967819214,
      "learning_rate": 9.283464100420063e-05,
      "loss": 1.1433,
      "step": 508
    },
    {
      "epoch": 0.5006147037128105,
      "grad_norm": 1.642370581626892,
      "learning_rate": 9.1951935904469e-05,
      "loss": 1.0781,
      "step": 509
    },
    {
      "epoch": 0.5015982296533071,
      "grad_norm": 1.6989452838897705,
      "learning_rate": 9.107219299699459e-05,
      "loss": 1.1015,
      "step": 510
    },
    {
      "epoch": 0.5025817555938038,
      "grad_norm": 1.6444441080093384,
      "learning_rate": 9.019543640037363e-05,
      "loss": 1.1494,
      "step": 511
    },
    {
      "epoch": 0.5035652815343005,
      "grad_norm": 1.650043249130249,
      "learning_rate": 8.93216901513312e-05,
      "loss": 1.1497,
      "step": 512
    },
    {
      "epoch": 0.5045488074747971,
      "grad_norm": 1.5892958641052246,
      "learning_rate": 8.845097820406199e-05,
      "loss": 1.0955,
      "step": 513
    },
    {
      "epoch": 0.5055323334152938,
      "grad_norm": 1.606795072555542,
      "learning_rate": 8.758332442957394e-05,
      "loss": 1.096,
      "step": 514
    },
    {
      "epoch": 0.5065158593557905,
      "grad_norm": 1.6573283672332764,
      "learning_rate": 8.671875261503348e-05,
      "loss": 1.1306,
      "step": 515
    },
    {
      "epoch": 0.5074993852962872,
      "grad_norm": 1.6091228723526,
      "learning_rate": 8.585728646311369e-05,
      "loss": 1.0845,
      "step": 516
    },
    {
      "epoch": 0.5084829112367839,
      "grad_norm": 1.6882151365280151,
      "learning_rate": 8.499894959134436e-05,
      "loss": 1.1502,
      "step": 517
    },
    {
      "epoch": 0.5094664371772806,
      "grad_norm": 1.6060574054718018,
      "learning_rate": 8.414376553146428e-05,
      "loss": 1.1295,
      "step": 518
    },
    {
      "epoch": 0.5104499631177772,
      "grad_norm": 1.6418123245239258,
      "learning_rate": 8.329175772877653e-05,
      "loss": 1.1259,
      "step": 519
    },
    {
      "epoch": 0.5114334890582739,
      "grad_norm": 1.6071466207504272,
      "learning_rate": 8.24429495415054e-05,
      "loss": 1.1085,
      "step": 520
    },
    {
      "epoch": 0.5124170149987706,
      "grad_norm": 1.5179975032806396,
      "learning_rate": 8.159736424015609e-05,
      "loss": 1.0485,
      "step": 521
    },
    {
      "epoch": 0.5134005409392672,
      "grad_norm": 1.6379435062408447,
      "learning_rate": 8.075502500687681e-05,
      "loss": 1.1094,
      "step": 522
    },
    {
      "epoch": 0.5143840668797639,
      "grad_norm": 1.6749610900878906,
      "learning_rate": 7.991595493482323e-05,
      "loss": 1.0844,
      "step": 523
    },
    {
      "epoch": 0.5153675928202607,
      "grad_norm": 1.696667194366455,
      "learning_rate": 7.908017702752504e-05,
      "loss": 1.1329,
      "step": 524
    },
    {
      "epoch": 0.5163511187607573,
      "grad_norm": 1.619667649269104,
      "learning_rate": 7.824771419825587e-05,
      "loss": 1.1005,
      "step": 525
    },
    {
      "epoch": 0.517334644701254,
      "grad_norm": 1.891265630722046,
      "learning_rate": 7.741858926940475e-05,
      "loss": 1.1328,
      "step": 526
    },
    {
      "epoch": 0.5183181706417507,
      "grad_norm": 1.612331509590149,
      "learning_rate": 7.65928249718503e-05,
      "loss": 1.1873,
      "step": 527
    },
    {
      "epoch": 0.5193016965822473,
      "grad_norm": 1.7046937942504883,
      "learning_rate": 7.577044394433794e-05,
      "loss": 1.0657,
      "step": 528
    },
    {
      "epoch": 0.520285222522744,
      "grad_norm": 1.6257797479629517,
      "learning_rate": 7.495146873285903e-05,
      "loss": 1.1406,
      "step": 529
    },
    {
      "epoch": 0.5212687484632407,
      "grad_norm": 1.6082725524902344,
      "learning_rate": 7.413592179003255e-05,
      "loss": 1.092,
      "step": 530
    },
    {
      "epoch": 0.5222522744037374,
      "grad_norm": 1.638418436050415,
      "learning_rate": 7.332382547449e-05,
      "loss": 1.1757,
      "step": 531
    },
    {
      "epoch": 0.5232358003442341,
      "grad_norm": 1.6094216108322144,
      "learning_rate": 7.251520205026205e-05,
      "loss": 1.1619,
      "step": 532
    },
    {
      "epoch": 0.5242193262847308,
      "grad_norm": 1.666649580001831,
      "learning_rate": 7.171007368616842e-05,
      "loss": 1.0869,
      "step": 533
    },
    {
      "epoch": 0.5252028522252274,
      "grad_norm": 1.6826838254928589,
      "learning_rate": 7.090846245520986e-05,
      "loss": 1.1334,
      "step": 534
    },
    {
      "epoch": 0.5261863781657241,
      "grad_norm": 1.6268287897109985,
      "learning_rate": 7.011039033396329e-05,
      "loss": 1.1287,
      "step": 535
    },
    {
      "epoch": 0.5271699041062208,
      "grad_norm": 1.6804072856903076,
      "learning_rate": 6.93158792019789e-05,
      "loss": 1.1468,
      "step": 536
    },
    {
      "epoch": 0.5281534300467174,
      "grad_norm": 1.6158145666122437,
      "learning_rate": 6.852495084118083e-05,
      "loss": 1.0231,
      "step": 537
    },
    {
      "epoch": 0.5291369559872141,
      "grad_norm": 1.7181380987167358,
      "learning_rate": 6.773762693526965e-05,
      "loss": 1.1221,
      "step": 538
    },
    {
      "epoch": 0.5301204819277109,
      "grad_norm": 1.662822961807251,
      "learning_rate": 6.695392906912786e-05,
      "loss": 1.1764,
      "step": 539
    },
    {
      "epoch": 0.5311040078682076,
      "grad_norm": 1.5750726461410522,
      "learning_rate": 6.617387872822842e-05,
      "loss": 1.0695,
      "step": 540
    },
    {
      "epoch": 0.5320875338087042,
      "grad_norm": 1.6667808294296265,
      "learning_rate": 6.539749729804538e-05,
      "loss": 1.1717,
      "step": 541
    },
    {
      "epoch": 0.5330710597492009,
      "grad_norm": 1.6217957735061646,
      "learning_rate": 6.462480606346788e-05,
      "loss": 1.0892,
      "step": 542
    },
    {
      "epoch": 0.5340545856896975,
      "grad_norm": 1.5805596113204956,
      "learning_rate": 6.385582620821644e-05,
      "loss": 1.1495,
      "step": 543
    },
    {
      "epoch": 0.5350381116301942,
      "grad_norm": 1.619666576385498,
      "learning_rate": 6.309057881426226e-05,
      "loss": 1.1372,
      "step": 544
    },
    {
      "epoch": 0.5360216375706909,
      "grad_norm": 1.5543806552886963,
      "learning_rate": 6.232908486124918e-05,
      "loss": 1.0527,
      "step": 545
    },
    {
      "epoch": 0.5370051635111877,
      "grad_norm": 1.5906850099563599,
      "learning_rate": 6.157136522591866e-05,
      "loss": 1.0886,
      "step": 546
    },
    {
      "epoch": 0.5379886894516843,
      "grad_norm": 1.7546547651290894,
      "learning_rate": 6.0817440681537144e-05,
      "loss": 1.0984,
      "step": 547
    },
    {
      "epoch": 0.538972215392181,
      "grad_norm": 1.7061015367507935,
      "learning_rate": 6.0067331897326895e-05,
      "loss": 1.0988,
      "step": 548
    },
    {
      "epoch": 0.5399557413326777,
      "grad_norm": 1.6275510787963867,
      "learning_rate": 5.9321059437899254e-05,
      "loss": 1.0858,
      "step": 549
    },
    {
      "epoch": 0.5409392672731743,
      "grad_norm": 1.6035560369491577,
      "learning_rate": 5.857864376269051e-05,
      "loss": 1.1456,
      "step": 550
    },
    {
      "epoch": 0.541922793213671,
      "grad_norm": 1.6571654081344604,
      "learning_rate": 5.784010522540151e-05,
      "loss": 1.1019,
      "step": 551
    },
    {
      "epoch": 0.5429063191541676,
      "grad_norm": 1.6013000011444092,
      "learning_rate": 5.7105464073439375e-05,
      "loss": 1.1046,
      "step": 552
    },
    {
      "epoch": 0.5438898450946644,
      "grad_norm": 1.5961564779281616,
      "learning_rate": 5.6374740447362264e-05,
      "loss": 1.1063,
      "step": 553
    },
    {
      "epoch": 0.5448733710351611,
      "grad_norm": 1.6290183067321777,
      "learning_rate": 5.564795438032757e-05,
      "loss": 1.1154,
      "step": 554
    },
    {
      "epoch": 0.5458568969756578,
      "grad_norm": 1.731971263885498,
      "learning_rate": 5.492512579754252e-05,
      "loss": 1.1149,
      "step": 555
    },
    {
      "epoch": 0.5468404229161544,
      "grad_norm": 1.6082978248596191,
      "learning_rate": 5.4206274515717736e-05,
      "loss": 1.1585,
      "step": 556
    },
    {
      "epoch": 0.5478239488566511,
      "grad_norm": 1.618871808052063,
      "learning_rate": 5.349142024252427e-05,
      "loss": 1.048,
      "step": 557
    },
    {
      "epoch": 0.5488074747971478,
      "grad_norm": 1.626085638999939,
      "learning_rate": 5.278058257605314e-05,
      "loss": 1.1256,
      "step": 558
    },
    {
      "epoch": 0.5497910007376444,
      "grad_norm": 1.5993646383285522,
      "learning_rate": 5.207378100427804e-05,
      "loss": 1.0744,
      "step": 559
    },
    {
      "epoch": 0.5507745266781411,
      "grad_norm": 1.6538002490997314,
      "learning_rate": 5.1371034904521134e-05,
      "loss": 1.103,
      "step": 560
    },
    {
      "epoch": 0.5517580526186379,
      "grad_norm": 1.6243610382080078,
      "learning_rate": 5.067236354292175e-05,
      "loss": 1.0917,
      "step": 561
    },
    {
      "epoch": 0.5527415785591345,
      "grad_norm": 1.570988416671753,
      "learning_rate": 4.9977786073908086e-05,
      "loss": 1.1085,
      "step": 562
    },
    {
      "epoch": 0.5537251044996312,
      "grad_norm": 1.603846788406372,
      "learning_rate": 4.928732153967246e-05,
      "loss": 1.1432,
      "step": 563
    },
    {
      "epoch": 0.5547086304401279,
      "grad_norm": 1.584790825843811,
      "learning_rate": 4.8600988869648745e-05,
      "loss": 1.0537,
      "step": 564
    },
    {
      "epoch": 0.5556921563806245,
      "grad_norm": 1.5726522207260132,
      "learning_rate": 4.7918806879993814e-05,
      "loss": 1.0789,
      "step": 565
    },
    {
      "epoch": 0.5566756823211212,
      "grad_norm": 1.608412504196167,
      "learning_rate": 4.724079427307162e-05,
      "loss": 1.1146,
      "step": 566
    },
    {
      "epoch": 0.5576592082616179,
      "grad_norm": 1.5979876518249512,
      "learning_rate": 4.656696963694012e-05,
      "loss": 1.0771,
      "step": 567
    },
    {
      "epoch": 0.5586427342021146,
      "grad_norm": 1.5700702667236328,
      "learning_rate": 4.589735144484217e-05,
      "loss": 1.0369,
      "step": 568
    },
    {
      "epoch": 0.5596262601426113,
      "grad_norm": 1.6290395259857178,
      "learning_rate": 4.5231958054698774e-05,
      "loss": 1.1129,
      "step": 569
    },
    {
      "epoch": 0.560609786083108,
      "grad_norm": 1.664450764656067,
      "learning_rate": 4.4570807708605825e-05,
      "loss": 1.0261,
      "step": 570
    },
    {
      "epoch": 0.5615933120236046,
      "grad_norm": 1.5685371160507202,
      "learning_rate": 4.391391853233404e-05,
      "loss": 1.1199,
      "step": 571
    },
    {
      "epoch": 0.5625768379641013,
      "grad_norm": 1.6812089681625366,
      "learning_rate": 4.326130853483206e-05,
      "loss": 1.0344,
      "step": 572
    },
    {
      "epoch": 0.563560363904598,
      "grad_norm": 1.622582197189331,
      "learning_rate": 4.261299560773255e-05,
      "loss": 1.061,
      "step": 573
    },
    {
      "epoch": 0.5645438898450946,
      "grad_norm": 1.660653829574585,
      "learning_rate": 4.196899752486192e-05,
      "loss": 1.1085,
      "step": 574
    },
    {
      "epoch": 0.5655274157855913,
      "grad_norm": 1.6321308612823486,
      "learning_rate": 4.132933194175299e-05,
      "loss": 1.0944,
      "step": 575
    },
    {
      "epoch": 0.5665109417260881,
      "grad_norm": 1.6691020727157593,
      "learning_rate": 4.069401639516075e-05,
      "loss": 1.1395,
      "step": 576
    },
    {
      "epoch": 0.5674944676665847,
      "grad_norm": 1.5686017274856567,
      "learning_rate": 4.0063068302581885e-05,
      "loss": 1.0362,
      "step": 577
    },
    {
      "epoch": 0.5684779936070814,
      "grad_norm": 1.6180577278137207,
      "learning_rate": 3.943650496177713e-05,
      "loss": 1.0636,
      "step": 578
    },
    {
      "epoch": 0.5694615195475781,
      "grad_norm": 1.5809311866760254,
      "learning_rate": 3.881434355029687e-05,
      "loss": 1.0899,
      "step": 579
    },
    {
      "epoch": 0.5704450454880747,
      "grad_norm": 1.5602284669876099,
      "learning_rate": 3.819660112501053e-05,
      "loss": 1.0964,
      "step": 580
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 1.555755376815796,
      "learning_rate": 3.758329462163874e-05,
      "loss": 1.1217,
      "step": 581
    },
    {
      "epoch": 0.5724120973690681,
      "grad_norm": 1.6677347421646118,
      "learning_rate": 3.697444085428914e-05,
      "loss": 1.1332,
      "step": 582
    },
    {
      "epoch": 0.5733956233095648,
      "grad_norm": 1.6057909727096558,
      "learning_rate": 3.637005651499528e-05,
      "loss": 1.0566,
      "step": 583
    },
    {
      "epoch": 0.5743791492500615,
      "grad_norm": 1.5967357158660889,
      "learning_rate": 3.57701581732592e-05,
      "loss": 1.1161,
      "step": 584
    },
    {
      "epoch": 0.5753626751905582,
      "grad_norm": 1.6041682958602905,
      "learning_rate": 3.5174762275596864e-05,
      "loss": 1.1068,
      "step": 585
    },
    {
      "epoch": 0.5763462011310548,
      "grad_norm": 1.5802150964736938,
      "learning_rate": 3.4583885145087615e-05,
      "loss": 1.0469,
      "step": 586
    },
    {
      "epoch": 0.5773297270715515,
      "grad_norm": 1.57747483253479,
      "learning_rate": 3.3997542980926524e-05,
      "loss": 1.0733,
      "step": 587
    },
    {
      "epoch": 0.5783132530120482,
      "grad_norm": 1.5697864294052124,
      "learning_rate": 3.3415751857980115e-05,
      "loss": 1.0767,
      "step": 588
    },
    {
      "epoch": 0.5792967789525448,
      "grad_norm": 1.550041675567627,
      "learning_rate": 3.2838527726345993e-05,
      "loss": 1.063,
      "step": 589
    },
    {
      "epoch": 0.5802803048930415,
      "grad_norm": 1.5676511526107788,
      "learning_rate": 3.226588641091521e-05,
      "loss": 1.0693,
      "step": 590
    },
    {
      "epoch": 0.5812638308335383,
      "grad_norm": 1.578323245048523,
      "learning_rate": 3.1697843610938794e-05,
      "loss": 1.0403,
      "step": 591
    },
    {
      "epoch": 0.5822473567740349,
      "grad_norm": 1.607877492904663,
      "learning_rate": 3.113441489959703e-05,
      "loss": 1.0868,
      "step": 592
    },
    {
      "epoch": 0.5832308827145316,
      "grad_norm": 1.5824037790298462,
      "learning_rate": 3.057561572357262e-05,
      "loss": 1.1237,
      "step": 593
    },
    {
      "epoch": 0.5842144086550283,
      "grad_norm": 1.5974537134170532,
      "learning_rate": 3.002146140262725e-05,
      "loss": 1.1145,
      "step": 594
    },
    {
      "epoch": 0.5851979345955249,
      "grad_norm": 1.5872732400894165,
      "learning_rate": 2.9471967129181565e-05,
      "loss": 0.9714,
      "step": 595
    },
    {
      "epoch": 0.5861814605360216,
      "grad_norm": 1.597594976425171,
      "learning_rate": 2.8927147967898682e-05,
      "loss": 1.0821,
      "step": 596
    },
    {
      "epoch": 0.5871649864765183,
      "grad_norm": 1.6081088781356812,
      "learning_rate": 2.8387018855271085e-05,
      "loss": 1.0513,
      "step": 597
    },
    {
      "epoch": 0.588148512417015,
      "grad_norm": 1.5636123418807983,
      "learning_rate": 2.7851594599211293e-05,
      "loss": 1.0317,
      "step": 598
    },
    {
      "epoch": 0.5891320383575117,
      "grad_norm": 1.5617702007293701,
      "learning_rate": 2.732088987864567e-05,
      "loss": 1.0692,
      "step": 599
    },
    {
      "epoch": 0.5901155642980084,
      "grad_norm": 1.6476192474365234,
      "learning_rate": 2.679491924311226e-05,
      "loss": 1.1095,
      "step": 600
    },
    {
      "epoch": 0.591099090238505,
      "grad_norm": 1.5220457315444946,
      "learning_rate": 2.6273697112361784e-05,
      "loss": 1.083,
      "step": 601
    },
    {
      "epoch": 0.5920826161790017,
      "grad_norm": 1.5808863639831543,
      "learning_rate": 2.575723777596213e-05,
      "loss": 1.1458,
      "step": 602
    },
    {
      "epoch": 0.5930661421194984,
      "grad_norm": 1.5608521699905396,
      "learning_rate": 2.524555539290696e-05,
      "loss": 1.049,
      "step": 603
    },
    {
      "epoch": 0.594049668059995,
      "grad_norm": 1.6394426822662354,
      "learning_rate": 2.473866399122733e-05,
      "loss": 1.0653,
      "step": 604
    },
    {
      "epoch": 0.5950331940004918,
      "grad_norm": 1.7107138633728027,
      "learning_rate": 2.4236577467606946e-05,
      "loss": 1.1567,
      "step": 605
    },
    {
      "epoch": 0.5960167199409885,
      "grad_norm": 1.5471723079681396,
      "learning_rate": 2.3739309587001567e-05,
      "loss": 1.0468,
      "step": 606
    },
    {
      "epoch": 0.5970002458814851,
      "grad_norm": 1.5568169355392456,
      "learning_rate": 2.324687398226131e-05,
      "loss": 1.0912,
      "step": 607
    },
    {
      "epoch": 0.5979837718219818,
      "grad_norm": 1.602166771888733,
      "learning_rate": 2.2759284153757053e-05,
      "loss": 1.1232,
      "step": 608
    },
    {
      "epoch": 0.5989672977624785,
      "grad_norm": 1.6254791021347046,
      "learning_rate": 2.2276553469010208e-05,
      "loss": 1.0962,
      "step": 609
    },
    {
      "epoch": 0.5999508237029751,
      "grad_norm": 1.5922654867172241,
      "learning_rate": 2.1798695162326442e-05,
      "loss": 1.0786,
      "step": 610
    },
    {
      "epoch": 0.6009343496434718,
      "grad_norm": 1.5703258514404297,
      "learning_rate": 2.1325722334432485e-05,
      "loss": 1.0496,
      "step": 611
    },
    {
      "epoch": 0.6019178755839685,
      "grad_norm": 1.5971964597702026,
      "learning_rate": 2.085764795211742e-05,
      "loss": 1.0631,
      "step": 612
    },
    {
      "epoch": 0.6029014015244653,
      "grad_norm": 1.6196777820587158,
      "learning_rate": 2.0394484847876892e-05,
      "loss": 1.1323,
      "step": 613
    },
    {
      "epoch": 0.6038849274649619,
      "grad_norm": 1.5888458490371704,
      "learning_rate": 1.9936245719561296e-05,
      "loss": 1.0985,
      "step": 614
    },
    {
      "epoch": 0.6048684534054586,
      "grad_norm": 1.6777266263961792,
      "learning_rate": 1.948294313002792e-05,
      "loss": 1.1057,
      "step": 615
    },
    {
      "epoch": 0.6058519793459552,
      "grad_norm": 1.5191071033477783,
      "learning_rate": 1.903458950679613e-05,
      "loss": 1.0123,
      "step": 616
    },
    {
      "epoch": 0.6068355052864519,
      "grad_norm": 1.5552016496658325,
      "learning_rate": 1.8591197141707027e-05,
      "loss": 1.0797,
      "step": 617
    },
    {
      "epoch": 0.6078190312269486,
      "grad_norm": 1.5961596965789795,
      "learning_rate": 1.8152778190586296e-05,
      "loss": 1.1184,
      "step": 618
    },
    {
      "epoch": 0.6088025571674452,
      "grad_norm": 1.5839762687683105,
      "learning_rate": 1.771934467291094e-05,
      "loss": 1.1104,
      "step": 619
    },
    {
      "epoch": 0.609786083107942,
      "grad_norm": 1.589618444442749,
      "learning_rate": 1.7290908471479805e-05,
      "loss": 1.1358,
      "step": 620
    },
    {
      "epoch": 0.6107696090484387,
      "grad_norm": 1.6342886686325073,
      "learning_rate": 1.6867481332087797e-05,
      "loss": 1.0282,
      "step": 621
    },
    {
      "epoch": 0.6117531349889354,
      "grad_norm": 1.5776050090789795,
      "learning_rate": 1.6449074863203772e-05,
      "loss": 1.0728,
      "step": 622
    },
    {
      "epoch": 0.612736660929432,
      "grad_norm": 1.6011443138122559,
      "learning_rate": 1.6035700535652465e-05,
      "loss": 1.1404,
      "step": 623
    },
    {
      "epoch": 0.6137201868699287,
      "grad_norm": 1.5918643474578857,
      "learning_rate": 1.562736968229992e-05,
      "loss": 1.0628,
      "step": 624
    },
    {
      "epoch": 0.6147037128104254,
      "grad_norm": 1.5177937746047974,
      "learning_rate": 1.5224093497742653e-05,
      "loss": 1.0399,
      "step": 625
    },
    {
      "epoch": 0.615687238750922,
      "grad_norm": 1.622794508934021,
      "learning_rate": 1.4825883038001054e-05,
      "loss": 1.1013,
      "step": 626
    },
    {
      "epoch": 0.6166707646914187,
      "grad_norm": 1.5753284692764282,
      "learning_rate": 1.4432749220216024e-05,
      "loss": 1.071,
      "step": 627
    },
    {
      "epoch": 0.6176542906319155,
      "grad_norm": 1.6217107772827148,
      "learning_rate": 1.4044702822349731e-05,
      "loss": 1.0002,
      "step": 628
    },
    {
      "epoch": 0.6186378165724121,
      "grad_norm": 1.5546847581863403,
      "learning_rate": 1.3661754482890222e-05,
      "loss": 1.0251,
      "step": 629
    },
    {
      "epoch": 0.6196213425129088,
      "grad_norm": 1.6065832376480103,
      "learning_rate": 1.3283914700559675e-05,
      "loss": 1.1277,
      "step": 630
    },
    {
      "epoch": 0.6206048684534055,
      "grad_norm": 1.582098364830017,
      "learning_rate": 1.2911193834026546e-05,
      "loss": 1.0606,
      "step": 631
    },
    {
      "epoch": 0.6215883943939021,
      "grad_norm": 1.5727086067199707,
      "learning_rate": 1.2543602101621709e-05,
      "loss": 1.0722,
      "step": 632
    },
    {
      "epoch": 0.6225719203343988,
      "grad_norm": 1.5591822862625122,
      "learning_rate": 1.2181149581058181e-05,
      "loss": 1.0718,
      "step": 633
    },
    {
      "epoch": 0.6235554462748955,
      "grad_norm": 1.5548235177993774,
      "learning_rate": 1.182384620915491e-05,
      "loss": 1.0455,
      "step": 634
    },
    {
      "epoch": 0.6245389722153922,
      "grad_norm": 1.5538259744644165,
      "learning_rate": 1.1471701781564314e-05,
      "loss": 1.0569,
      "step": 635
    },
    {
      "epoch": 0.6255224981558889,
      "grad_norm": 1.5374336242675781,
      "learning_rate": 1.1124725952503801e-05,
      "loss": 1.0752,
      "step": 636
    },
    {
      "epoch": 0.6265060240963856,
      "grad_norm": 1.590369462966919,
      "learning_rate": 1.078292823449094e-05,
      "loss": 1.0142,
      "step": 637
    },
    {
      "epoch": 0.6274895500368822,
      "grad_norm": 1.5979132652282715,
      "learning_rate": 1.0446317998082888e-05,
      "loss": 1.072,
      "step": 638
    },
    {
      "epoch": 0.6284730759773789,
      "grad_norm": 1.5855075120925903,
      "learning_rate": 1.0114904471619247e-05,
      "loss": 1.0809,
      "step": 639
    },
    {
      "epoch": 0.6294566019178756,
      "grad_norm": 1.6333812475204468,
      "learning_rate": 9.788696740969295e-06,
      "loss": 1.1021,
      "step": 640
    },
    {
      "epoch": 0.6304401278583722,
      "grad_norm": 1.646070957183838,
      "learning_rate": 9.467703749282764e-06,
      "loss": 1.0921,
      "step": 641
    },
    {
      "epoch": 0.631423653798869,
      "grad_norm": 1.5639617443084717,
      "learning_rate": 9.151934296744635e-06,
      "loss": 1.052,
      "step": 642
    },
    {
      "epoch": 0.6324071797393657,
      "grad_norm": 1.6541653871536255,
      "learning_rate": 8.841397040333976e-06,
      "loss": 1.0609,
      "step": 643
    },
    {
      "epoch": 0.6333907056798623,
      "grad_norm": 1.5831325054168701,
      "learning_rate": 8.536100493586552e-06,
      "loss": 1.1326,
      "step": 644
    },
    {
      "epoch": 0.634374231620359,
      "grad_norm": 1.496453046798706,
      "learning_rate": 8.23605302636139e-06,
      "loss": 1.0269,
      "step": 645
    },
    {
      "epoch": 0.6353577575608557,
      "grad_norm": 1.6001390218734741,
      "learning_rate": 7.941262864611387e-06,
      "loss": 1.0708,
      "step": 646
    },
    {
      "epoch": 0.6363412835013523,
      "grad_norm": 1.6304258108139038,
      "learning_rate": 7.651738090157733e-06,
      "loss": 1.0645,
      "step": 647
    },
    {
      "epoch": 0.637324809441849,
      "grad_norm": 1.5697929859161377,
      "learning_rate": 7.36748664046838e-06,
      "loss": 1.0812,
      "step": 648
    },
    {
      "epoch": 0.6383083353823457,
      "grad_norm": 1.6335376501083374,
      "learning_rate": 7.0885163084403846e-06,
      "loss": 1.0871,
      "step": 649
    },
    {
      "epoch": 0.6392918613228424,
      "grad_norm": 1.6851428747177124,
      "learning_rate": 6.81483474218636e-06,
      "loss": 1.066,
      "step": 650
    },
    {
      "epoch": 0.6402753872633391,
      "grad_norm": 1.5803459882736206,
      "learning_rate": 6.546449444824654e-06,
      "loss": 1.0518,
      "step": 651
    },
    {
      "epoch": 0.6412589132038358,
      "grad_norm": 1.6332086324691772,
      "learning_rate": 6.283367774273785e-06,
      "loss": 1.0243,
      "step": 652
    },
    {
      "epoch": 0.6422424391443324,
      "grad_norm": 1.5875962972640991,
      "learning_rate": 6.025596943050648e-06,
      "loss": 1.0768,
      "step": 653
    },
    {
      "epoch": 0.6432259650848291,
      "grad_norm": 1.648381233215332,
      "learning_rate": 5.773144018072807e-06,
      "loss": 1.1158,
      "step": 654
    },
    {
      "epoch": 0.6442094910253258,
      "grad_norm": 1.6049330234527588,
      "learning_rate": 5.5260159204646885e-06,
      "loss": 1.1665,
      "step": 655
    },
    {
      "epoch": 0.6451930169658224,
      "grad_norm": 1.5462381839752197,
      "learning_rate": 5.2842194253679424e-06,
      "loss": 1.0788,
      "step": 656
    },
    {
      "epoch": 0.6461765429063192,
      "grad_norm": 1.664047122001648,
      "learning_rate": 5.0477611617556485e-06,
      "loss": 1.156,
      "step": 657
    },
    {
      "epoch": 0.6471600688468159,
      "grad_norm": 1.7467290163040161,
      "learning_rate": 4.8166476122505135e-06,
      "loss": 1.1192,
      "step": 658
    },
    {
      "epoch": 0.6481435947873125,
      "grad_norm": 1.7353417873382568,
      "learning_rate": 4.590885112947274e-06,
      "loss": 1.0867,
      "step": 659
    },
    {
      "epoch": 0.6491271207278092,
      "grad_norm": 1.587186336517334,
      "learning_rate": 4.370479853238863e-06,
      "loss": 1.055,
      "step": 660
    },
    {
      "epoch": 0.6501106466683059,
      "grad_norm": 1.553141474723816,
      "learning_rate": 4.155437875646829e-06,
      "loss": 1.0706,
      "step": 661
    },
    {
      "epoch": 0.6510941726088025,
      "grad_norm": 1.5985965728759766,
      "learning_rate": 3.945765075655649e-06,
      "loss": 1.0794,
      "step": 662
    },
    {
      "epoch": 0.6520776985492992,
      "grad_norm": 1.505215048789978,
      "learning_rate": 3.7414672015509746e-06,
      "loss": 1.0067,
      "step": 663
    },
    {
      "epoch": 0.6530612244897959,
      "grad_norm": 1.575181484222412,
      "learning_rate": 3.542549854262278e-06,
      "loss": 1.0166,
      "step": 664
    },
    {
      "epoch": 0.6540447504302926,
      "grad_norm": 1.6103475093841553,
      "learning_rate": 3.349018487209099e-06,
      "loss": 1.1203,
      "step": 665
    },
    {
      "epoch": 0.6550282763707893,
      "grad_norm": 1.6004595756530762,
      "learning_rate": 3.160878406151624e-06,
      "loss": 1.0892,
      "step": 666
    },
    {
      "epoch": 0.656011802311286,
      "grad_norm": 1.5086390972137451,
      "learning_rate": 2.9781347690452266e-06,
      "loss": 1.051,
      "step": 667
    },
    {
      "epoch": 0.6569953282517826,
      "grad_norm": 1.570785403251648,
      "learning_rate": 2.800792585899026e-06,
      "loss": 0.9648,
      "step": 668
    },
    {
      "epoch": 0.6579788541922793,
      "grad_norm": 1.54836905002594,
      "learning_rate": 2.6288567186385505e-06,
      "loss": 1.0549,
      "step": 669
    },
    {
      "epoch": 0.658962380132776,
      "grad_norm": 1.660688042640686,
      "learning_rate": 2.462331880972468e-06,
      "loss": 1.0269,
      "step": 670
    },
    {
      "epoch": 0.6599459060732726,
      "grad_norm": 1.5790220499038696,
      "learning_rate": 2.3012226382632894e-06,
      "loss": 1.0473,
      "step": 671
    },
    {
      "epoch": 0.6609294320137694,
      "grad_norm": 1.6120561361312866,
      "learning_rate": 2.1455334074023337e-06,
      "loss": 1.1179,
      "step": 672
    },
    {
      "epoch": 0.6619129579542661,
      "grad_norm": 1.5404844284057617,
      "learning_rate": 1.9952684566884926e-06,
      "loss": 1.0333,
      "step": 673
    },
    {
      "epoch": 0.6628964838947627,
      "grad_norm": 1.5952411890029907,
      "learning_rate": 1.8504319057112806e-06,
      "loss": 1.0023,
      "step": 674
    },
    {
      "epoch": 0.6638800098352594,
      "grad_norm": 1.549829363822937,
      "learning_rate": 1.7110277252379236e-06,
      "loss": 1.0127,
      "step": 675
    },
    {
      "epoch": 0.6648635357757561,
      "grad_norm": 1.5770440101623535,
      "learning_rate": 1.577059737104447e-06,
      "loss": 1.0096,
      "step": 676
    },
    {
      "epoch": 0.6658470617162527,
      "grad_norm": 1.625446081161499,
      "learning_rate": 1.4485316141108928e-06,
      "loss": 1.0445,
      "step": 677
    },
    {
      "epoch": 0.6668305876567494,
      "grad_norm": 1.6090573072433472,
      "learning_rate": 1.325446879920711e-06,
      "loss": 1.0989,
      "step": 678
    },
    {
      "epoch": 0.6678141135972461,
      "grad_norm": 1.5554827451705933,
      "learning_rate": 1.2078089089640809e-06,
      "loss": 1.022,
      "step": 679
    },
    {
      "epoch": 0.6687976395377428,
      "grad_norm": 1.5727711915969849,
      "learning_rate": 1.0956209263453421e-06,
      "loss": 1.0687,
      "step": 680
    },
    {
      "epoch": 0.6697811654782395,
      "grad_norm": 1.590895175933838,
      "learning_rate": 9.888860077547524e-07,
      "loss": 1.1218,
      "step": 681
    },
    {
      "epoch": 0.6707646914187362,
      "grad_norm": 1.6273709535598755,
      "learning_rate": 8.876070793840008e-07,
      "loss": 1.0526,
      "step": 682
    },
    {
      "epoch": 0.6717482173592328,
      "grad_norm": 1.5715453624725342,
      "learning_rate": 7.917869178460934e-07,
      "loss": 1.0828,
      "step": 683
    },
    {
      "epoch": 0.6727317432997295,
      "grad_norm": 1.5626118183135986,
      "learning_rate": 7.01428150099126e-07,
      "loss": 1.1077,
      "step": 684
    },
    {
      "epoch": 0.6737152692402262,
      "grad_norm": 1.5551321506500244,
      "learning_rate": 6.165332533744073e-07,
      "loss": 1.0527,
      "step": 685
    },
    {
      "epoch": 0.6746987951807228,
      "grad_norm": 1.6907010078430176,
      "learning_rate": 5.371045551083808e-07,
      "loss": 1.0607,
      "step": 686
    },
    {
      "epoch": 0.6756823211212196,
      "grad_norm": 1.5568333864212036,
      "learning_rate": 4.631442328789426e-07,
      "loss": 0.9964,
      "step": 687
    },
    {
      "epoch": 0.6766658470617163,
      "grad_norm": 1.6289827823638916,
      "learning_rate": 3.946543143456882e-07,
      "loss": 1.0931,
      "step": 688
    },
    {
      "epoch": 0.677649373002213,
      "grad_norm": 1.5730174779891968,
      "learning_rate": 3.3163667719433576e-07,
      "loss": 1.0916,
      "step": 689
    },
    {
      "epoch": 0.6786328989427096,
      "grad_norm": 1.5842548608779907,
      "learning_rate": 2.7409304908523336e-07,
      "loss": 1.051,
      "step": 690
    },
    {
      "epoch": 0.6796164248832063,
      "grad_norm": 1.617651343345642,
      "learning_rate": 2.220250076060193e-07,
      "loss": 1.0332,
      "step": 691
    },
    {
      "epoch": 0.680599950823703,
      "grad_norm": 1.6278129816055298,
      "learning_rate": 1.7543398022832336e-07,
      "loss": 1.1475,
      "step": 692
    },
    {
      "epoch": 0.6815834767641996,
      "grad_norm": 1.5694608688354492,
      "learning_rate": 1.343212442687536e-07,
      "loss": 1.0138,
      "step": 693
    },
    {
      "epoch": 0.6825670027046964,
      "grad_norm": 1.5981290340423584,
      "learning_rate": 9.868792685368e-08,
      "loss": 0.9832,
      "step": 694
    },
    {
      "epoch": 0.6835505286451931,
      "grad_norm": 1.5327750444412231,
      "learning_rate": 6.853500488854803e-08,
      "loss": 1.0314,
      "step": 695
    },
    {
      "epoch": 0.6845340545856897,
      "grad_norm": 1.5427277088165283,
      "learning_rate": 4.386330503090008e-08,
      "loss": 1.0459,
      "step": 696
    },
    {
      "epoch": 0.6855175805261864,
      "grad_norm": 1.5548511743545532,
      "learning_rate": 2.4673503667882458e-08,
      "loss": 1.0574,
      "step": 697
    },
    {
      "epoch": 0.686501106466683,
      "grad_norm": 1.5951749086380005,
      "learning_rate": 1.0966126897571372e-08,
      "loss": 1.1301,
      "step": 698
    },
    {
      "epoch": 0.6874846324071797,
      "grad_norm": 1.5737613439559937,
      "learning_rate": 2.741550514651081e-09,
      "loss": 1.0478,
      "step": 699
    },
    {
      "epoch": 0.6884681583476764,
      "grad_norm": 1.5692044496536255,
      "learning_rate": 0.0,
      "loss": 1.0708,
      "step": 700
    }
  ],
  "logging_steps": 1,
  "max_steps": 700,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 250,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 1,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.438718091440947e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}