{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9975990396158463,
  "eval_steps": 500,
  "global_step": 312,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.006402561024409764,
      "grad_norm": 2.679521906145934,
      "learning_rate": 6.25e-08,
      "loss": 0.1981,
      "step": 1
    },
    {
      "epoch": 0.012805122048819529,
      "grad_norm": 2.5600704949176025,
      "learning_rate": 1.25e-07,
      "loss": 0.1782,
      "step": 2
    },
    {
      "epoch": 0.01920768307322929,
      "grad_norm": 2.8865232354148174,
      "learning_rate": 1.875e-07,
      "loss": 0.2347,
      "step": 3
    },
    {
      "epoch": 0.025610244097639057,
      "grad_norm": 2.76706815796981,
      "learning_rate": 2.5e-07,
      "loss": 0.2138,
      "step": 4
    },
    {
      "epoch": 0.03201280512204882,
      "grad_norm": 2.6835981054304963,
      "learning_rate": 3.1249999999999997e-07,
      "loss": 0.2395,
      "step": 5
    },
    {
      "epoch": 0.03841536614645858,
      "grad_norm": 2.6104328133293877,
      "learning_rate": 3.75e-07,
      "loss": 0.2417,
      "step": 6
    },
    {
      "epoch": 0.04481792717086835,
      "grad_norm": 2.551735870154201,
      "learning_rate": 4.375e-07,
      "loss": 0.175,
      "step": 7
    },
    {
      "epoch": 0.051220488195278115,
      "grad_norm": 3.4086729277405725,
      "learning_rate": 5e-07,
      "loss": 0.1706,
      "step": 8
    },
    {
      "epoch": 0.057623049219687875,
      "grad_norm": 2.7571882436426587,
      "learning_rate": 5.625e-07,
      "loss": 0.2184,
      "step": 9
    },
    {
      "epoch": 0.06402561024409764,
      "grad_norm": 2.7694984961484717,
      "learning_rate": 6.249999999999999e-07,
      "loss": 0.2017,
      "step": 10
    },
    {
      "epoch": 0.07042817126850741,
      "grad_norm": 2.547784892526759,
      "learning_rate": 6.875e-07,
      "loss": 0.1911,
      "step": 11
    },
    {
      "epoch": 0.07683073229291716,
      "grad_norm": 2.666671570148531,
      "learning_rate": 7.5e-07,
      "loss": 0.2143,
      "step": 12
    },
    {
      "epoch": 0.08323329331732693,
      "grad_norm": 2.7605278136164912,
      "learning_rate": 8.125e-07,
      "loss": 0.2027,
      "step": 13
    },
    {
      "epoch": 0.0896358543417367,
      "grad_norm": 2.4621112010955866,
      "learning_rate": 8.75e-07,
      "loss": 0.2068,
      "step": 14
    },
    {
      "epoch": 0.09603841536614646,
      "grad_norm": 2.803025975251869,
      "learning_rate": 9.374999999999999e-07,
      "loss": 0.213,
      "step": 15
    },
    {
      "epoch": 0.10244097639055623,
      "grad_norm": 2.901342575515257,
      "learning_rate": 1e-06,
      "loss": 0.2106,
      "step": 16
    },
    {
      "epoch": 0.10884353741496598,
      "grad_norm": 2.7337203386461644,
      "learning_rate": 1.0625e-06,
      "loss": 0.2281,
      "step": 17
    },
    {
      "epoch": 0.11524609843937575,
      "grad_norm": 2.6481197250458193,
      "learning_rate": 1.125e-06,
      "loss": 0.1827,
      "step": 18
    },
    {
      "epoch": 0.12164865946378552,
      "grad_norm": 2.7639376607457162,
      "learning_rate": 1.1874999999999999e-06,
      "loss": 0.2159,
      "step": 19
    },
    {
      "epoch": 0.12805122048819528,
      "grad_norm": 2.7175927412952467,
      "learning_rate": 1.2499999999999999e-06,
      "loss": 0.1745,
      "step": 20
    },
    {
      "epoch": 0.13445378151260504,
      "grad_norm": 3.1021924663248384,
      "learning_rate": 1.3125e-06,
      "loss": 0.1935,
      "step": 21
    },
    {
      "epoch": 0.14085634253701482,
      "grad_norm": 3.1941312523935563,
      "learning_rate": 1.375e-06,
      "loss": 0.2053,
      "step": 22
    },
    {
      "epoch": 0.14725890356142457,
      "grad_norm": 2.500688058457321,
      "learning_rate": 1.4375e-06,
      "loss": 0.1699,
      "step": 23
    },
    {
      "epoch": 0.15366146458583432,
      "grad_norm": 3.417365189000316,
      "learning_rate": 1.5e-06,
      "loss": 0.2222,
      "step": 24
    },
    {
      "epoch": 0.1600640256102441,
      "grad_norm": 2.7988997309084125,
      "learning_rate": 1.5624999999999999e-06,
      "loss": 0.1855,
      "step": 25
    },
    {
      "epoch": 0.16646658663465386,
      "grad_norm": 3.057142160060086,
      "learning_rate": 1.625e-06,
      "loss": 0.1698,
      "step": 26
    },
    {
      "epoch": 0.17286914765906364,
      "grad_norm": 2.8282556781813772,
      "learning_rate": 1.6875e-06,
      "loss": 0.2047,
      "step": 27
    },
    {
      "epoch": 0.1792717086834734,
      "grad_norm": 2.7193234045232413,
      "learning_rate": 1.75e-06,
      "loss": 0.1979,
      "step": 28
    },
    {
      "epoch": 0.18567426970788314,
      "grad_norm": 3.520125861739441,
      "learning_rate": 1.8125e-06,
      "loss": 0.2214,
      "step": 29
    },
    {
      "epoch": 0.19207683073229292,
      "grad_norm": 3.017660845656798,
      "learning_rate": 1.8749999999999998e-06,
      "loss": 0.2119,
      "step": 30
    },
    {
      "epoch": 0.19847939175670268,
      "grad_norm": 3.4420544526065515,
      "learning_rate": 1.9375e-06,
      "loss": 0.1649,
      "step": 31
    },
    {
      "epoch": 0.20488195278111246,
      "grad_norm": 2.858393701736414,
      "learning_rate": 2e-06,
      "loss": 0.2038,
      "step": 32
    },
    {
      "epoch": 0.2112845138055222,
      "grad_norm": 3.6069506971510723,
      "learning_rate": 1.9999370567547003e-06,
      "loss": 0.2194,
      "step": 33
    },
    {
      "epoch": 0.21768707482993196,
      "grad_norm": 3.1830415852739415,
      "learning_rate": 1.9997482349425066e-06,
      "loss": 0.143,
      "step": 34
    },
    {
      "epoch": 0.22408963585434175,
      "grad_norm": 3.4429216757147545,
      "learning_rate": 1.9994335583335335e-06,
      "loss": 0.2026,
      "step": 35
    },
    {
      "epoch": 0.2304921968787515,
      "grad_norm": 3.5752393511399463,
      "learning_rate": 1.9989930665413145e-06,
      "loss": 0.2087,
      "step": 36
    },
    {
      "epoch": 0.23689475790316125,
      "grad_norm": 4.08665235099645,
      "learning_rate": 1.9984268150178167e-06,
      "loss": 0.196,
      "step": 37
    },
    {
      "epoch": 0.24329731892757103,
      "grad_norm": 3.979136486495609,
      "learning_rate": 1.997734875046456e-06,
      "loss": 0.1824,
      "step": 38
    },
    {
      "epoch": 0.24969987995198079,
      "grad_norm": 3.866519745374768,
      "learning_rate": 1.996917333733128e-06,
      "loss": 0.1896,
      "step": 39
    },
    {
      "epoch": 0.25610244097639057,
      "grad_norm": 3.2575226153396115,
      "learning_rate": 1.995974293995239e-06,
      "loss": 0.194,
      "step": 40
    },
    {
      "epoch": 0.26250500200080035,
      "grad_norm": 3.7743686803486605,
      "learning_rate": 1.994905874548752e-06,
      "loss": 0.1882,
      "step": 41
    },
    {
      "epoch": 0.2689075630252101,
      "grad_norm": 4.287147445421594,
      "learning_rate": 1.9937122098932426e-06,
      "loss": 0.1772,
      "step": 42
    },
    {
      "epoch": 0.27531012404961985,
      "grad_norm": 3.832538953651167,
      "learning_rate": 1.9923934502949643e-06,
      "loss": 0.1904,
      "step": 43
    },
    {
      "epoch": 0.28171268507402963,
      "grad_norm": 5.1771162120228365,
      "learning_rate": 1.9909497617679347e-06,
      "loss": 0.1892,
      "step": 44
    },
    {
      "epoch": 0.28811524609843936,
      "grad_norm": 4.303644513637155,
      "learning_rate": 1.9893813260530367e-06,
      "loss": 0.2056,
      "step": 45
    },
    {
      "epoch": 0.29451780712284914,
      "grad_norm": 4.2352469784485445,
      "learning_rate": 1.9876883405951377e-06,
      "loss": 0.2316,
      "step": 46
    },
    {
      "epoch": 0.3009203681472589,
      "grad_norm": 4.5051673675815955,
      "learning_rate": 1.9858710185182355e-06,
      "loss": 0.1792,
      "step": 47
    },
    {
      "epoch": 0.30732292917166865,
      "grad_norm": 4.68950995491574,
      "learning_rate": 1.9839295885986295e-06,
      "loss": 0.1841,
      "step": 48
    },
    {
      "epoch": 0.3137254901960784,
      "grad_norm": 4.053751205533602,
      "learning_rate": 1.9818642952361183e-06,
      "loss": 0.1909,
      "step": 49
    },
    {
      "epoch": 0.3201280512204882,
      "grad_norm": 3.537770773869747,
      "learning_rate": 1.9796753984232355e-06,
      "loss": 0.1642,
      "step": 50
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 4.937399937551455,
      "learning_rate": 1.977363173712519e-06,
      "loss": 0.1628,
      "step": 51
    },
    {
      "epoch": 0.3329331732693077,
      "grad_norm": 4.458098457737047,
      "learning_rate": 1.9749279121818236e-06,
      "loss": 0.2108,
      "step": 52
    },
    {
      "epoch": 0.3393357342937175,
      "grad_norm": 4.692309558133837,
      "learning_rate": 1.9723699203976766e-06,
      "loss": 0.2023,
      "step": 53
    },
    {
      "epoch": 0.3457382953181273,
      "grad_norm": 4.277910169048945,
      "learning_rate": 1.9696895203766866e-06,
      "loss": 0.1885,
      "step": 54
    },
    {
      "epoch": 0.352140856342537,
      "grad_norm": 4.837495606765268,
      "learning_rate": 1.966887049545006e-06,
      "loss": 0.1899,
      "step": 55
    },
    {
      "epoch": 0.3585434173669468,
      "grad_norm": 4.723060685847474,
      "learning_rate": 1.9639628606958534e-06,
      "loss": 0.2454,
      "step": 56
    },
    {
      "epoch": 0.36494597839135656,
      "grad_norm": 4.472186733872948,
      "learning_rate": 1.9609173219450997e-06,
      "loss": 0.1902,
      "step": 57
    },
    {
      "epoch": 0.3713485394157663,
      "grad_norm": 4.27556686419538,
      "learning_rate": 1.9577508166849303e-06,
      "loss": 0.1962,
      "step": 58
    },
    {
      "epoch": 0.37775110044017607,
      "grad_norm": 3.8215311273874506,
      "learning_rate": 1.9544637435355806e-06,
      "loss": 0.1658,
      "step": 59
    },
    {
      "epoch": 0.38415366146458585,
      "grad_norm": 4.623141330416456,
      "learning_rate": 1.9510565162951534e-06,
      "loss": 0.2274,
      "step": 60
    },
    {
      "epoch": 0.3905562224889956,
      "grad_norm": 4.245526551915113,
      "learning_rate": 1.947529563887529e-06,
      "loss": 0.2097,
      "step": 61
    },
    {
      "epoch": 0.39695878351340536,
      "grad_norm": 3.8360308169316064,
      "learning_rate": 1.9438833303083674e-06,
      "loss": 0.1629,
      "step": 62
    },
    {
      "epoch": 0.40336134453781514,
      "grad_norm": 4.390197594681171,
      "learning_rate": 1.9401182745692187e-06,
      "loss": 0.1913,
      "step": 63
    },
    {
      "epoch": 0.4097639055622249,
      "grad_norm": 3.8567511370833594,
      "learning_rate": 1.936234870639737e-06,
      "loss": 0.204,
      "step": 64
    },
    {
      "epoch": 0.41616646658663464,
      "grad_norm": 4.3670167230444354,
      "learning_rate": 1.9322336073880143e-06,
      "loss": 0.1925,
      "step": 65
    },
    {
      "epoch": 0.4225690276110444,
      "grad_norm": 4.5446121221499505,
      "learning_rate": 1.928114988519039e-06,
      "loss": 0.2036,
      "step": 66
    },
    {
      "epoch": 0.4289715886354542,
      "grad_norm": 4.492339684008934,
      "learning_rate": 1.9238795325112867e-06,
      "loss": 0.1976,
      "step": 67
    },
    {
      "epoch": 0.43537414965986393,
      "grad_norm": 3.616344744906229,
      "learning_rate": 1.9195277725514506e-06,
      "loss": 0.2173,
      "step": 68
    },
    {
      "epoch": 0.4417767106842737,
      "grad_norm": 4.807993813843196,
      "learning_rate": 1.91506025646732e-06,
      "loss": 0.2297,
      "step": 69
    },
    {
      "epoch": 0.4481792717086835,
      "grad_norm": 3.9611147863165215,
      "learning_rate": 1.9104775466588157e-06,
      "loss": 0.1724,
      "step": 70
    },
    {
      "epoch": 0.4545818327330932,
      "grad_norm": 4.038448965820852,
      "learning_rate": 1.905780220027194e-06,
      "loss": 0.1817,
      "step": 71
    },
    {
      "epoch": 0.460984393757503,
      "grad_norm": 3.6234581360493308,
      "learning_rate": 1.9009688679024189e-06,
      "loss": 0.1898,
      "step": 72
    },
    {
      "epoch": 0.4673869547819128,
      "grad_norm": 4.412888570956389,
      "learning_rate": 1.8960440959687252e-06,
      "loss": 0.2056,
      "step": 73
    },
    {
      "epoch": 0.4737895158063225,
      "grad_norm": 4.438175504352378,
      "learning_rate": 1.8910065241883678e-06,
      "loss": 0.2206,
      "step": 74
    },
    {
      "epoch": 0.4801920768307323,
      "grad_norm": 4.188755548423674,
      "learning_rate": 1.8858567867235798e-06,
      "loss": 0.1955,
      "step": 75
    },
    {
      "epoch": 0.48659463785514206,
      "grad_norm": 4.043921178736117,
      "learning_rate": 1.8805955318567379e-06,
      "loss": 0.1948,
      "step": 76
    },
    {
      "epoch": 0.49299719887955185,
      "grad_norm": 3.3626388865946195,
      "learning_rate": 1.8752234219087537e-06,
      "loss": 0.2156,
      "step": 77
    },
    {
      "epoch": 0.49939975990396157,
      "grad_norm": 3.856090864925135,
      "learning_rate": 1.8697411331556953e-06,
      "loss": 0.1969,
      "step": 78
    },
    {
      "epoch": 0.5058023209283713,
      "grad_norm": 3.902527444697349,
      "learning_rate": 1.8641493557436548e-06,
      "loss": 0.2156,
      "step": 79
    },
    {
      "epoch": 0.5122048819527811,
      "grad_norm": 3.8994885011285776,
      "learning_rate": 1.858448793601866e-06,
      "loss": 0.1978,
      "step": 80
    },
    {
      "epoch": 0.5186074429771909,
      "grad_norm": 3.842402356308699,
      "learning_rate": 1.852640164354092e-06,
      "loss": 0.2271,
      "step": 81
    },
    {
      "epoch": 0.5250100040016007,
      "grad_norm": 3.3752000095050376,
      "learning_rate": 1.8467241992282841e-06,
      "loss": 0.1707,
      "step": 82
    },
    {
      "epoch": 0.5314125650260104,
      "grad_norm": 4.043279107262682,
      "learning_rate": 1.8407016429645302e-06,
      "loss": 0.2163,
      "step": 83
    },
    {
      "epoch": 0.5378151260504201,
      "grad_norm": 4.403978560088287,
      "learning_rate": 1.8345732537213026e-06,
      "loss": 0.1739,
      "step": 84
    },
    {
      "epoch": 0.54421768707483,
      "grad_norm": 3.8981585910721552,
      "learning_rate": 1.8283398029800164e-06,
      "loss": 0.1696,
      "step": 85
    },
    {
      "epoch": 0.5506202480992397,
      "grad_norm": 4.932247696008996,
      "learning_rate": 1.82200207544791e-06,
      "loss": 0.1915,
      "step": 86
    },
    {
      "epoch": 0.5570228091236494,
      "grad_norm": 4.099692261040826,
      "learning_rate": 1.8155608689592601e-06,
      "loss": 0.1785,
      "step": 87
    },
    {
      "epoch": 0.5634253701480593,
      "grad_norm": 3.7071427170962585,
      "learning_rate": 1.8090169943749474e-06,
      "loss": 0.1643,
      "step": 88
    },
    {
      "epoch": 0.569827931172469,
      "grad_norm": 4.198912758644434,
      "learning_rate": 1.802371275480378e-06,
      "loss": 0.2079,
      "step": 89
    },
    {
      "epoch": 0.5762304921968787,
      "grad_norm": 4.08750592105122,
      "learning_rate": 1.795624548881781e-06,
      "loss": 0.1875,
      "step": 90
    },
    {
      "epoch": 0.5826330532212886,
      "grad_norm": 5.4368089983091945,
      "learning_rate": 1.7887776639008912e-06,
      "loss": 0.1842,
      "step": 91
    },
    {
      "epoch": 0.5890356142456983,
      "grad_norm": 3.6668016857716,
      "learning_rate": 1.7818314824680298e-06,
      "loss": 0.1927,
      "step": 92
    },
    {
      "epoch": 0.595438175270108,
      "grad_norm": 3.9325828012665482,
      "learning_rate": 1.774786879013601e-06,
      "loss": 0.1849,
      "step": 93
    },
    {
      "epoch": 0.6018407362945178,
      "grad_norm": 4.2551345503780125,
      "learning_rate": 1.767644740358011e-06,
      "loss": 0.2103,
      "step": 94
    },
    {
      "epoch": 0.6082432973189276,
      "grad_norm": 3.562360928332586,
      "learning_rate": 1.760405965600031e-06,
      "loss": 0.2292,
      "step": 95
    },
    {
      "epoch": 0.6146458583433373,
      "grad_norm": 4.074513845673071,
      "learning_rate": 1.753071466003611e-06,
      "loss": 0.164,
      "step": 96
    },
    {
      "epoch": 0.6210484193677471,
      "grad_norm": 4.264136629053828,
      "learning_rate": 1.7456421648831654e-06,
      "loss": 0.1919,
      "step": 97
    },
    {
      "epoch": 0.6274509803921569,
      "grad_norm": 4.101904029046064,
      "learning_rate": 1.7381189974873407e-06,
      "loss": 0.2492,
      "step": 98
    },
    {
      "epoch": 0.6338535414165666,
      "grad_norm": 4.129940049014058,
      "learning_rate": 1.7305029108812774e-06,
      "loss": 0.1899,
      "step": 99
    },
    {
      "epoch": 0.6402561024409764,
      "grad_norm": 4.2688607303525705,
      "learning_rate": 1.7227948638273915e-06,
      "loss": 0.1802,
      "step": 100
    },
    {
      "epoch": 0.6466586634653861,
      "grad_norm": 3.931916755773599,
      "learning_rate": 1.7149958266646754e-06,
      "loss": 0.1524,
      "step": 101
    },
    {
      "epoch": 0.6530612244897959,
      "grad_norm": 4.309265054679012,
      "learning_rate": 1.7071067811865474e-06,
      "loss": 0.2035,
      "step": 102
    },
    {
      "epoch": 0.6594637855142057,
      "grad_norm": 3.4899838789545847,
      "learning_rate": 1.6991287205172574e-06,
      "loss": 0.1885,
      "step": 103
    },
    {
      "epoch": 0.6658663465386154,
      "grad_norm": 4.603685696411418,
      "learning_rate": 1.6910626489868648e-06,
      "loss": 0.2008,
      "step": 104
    },
    {
      "epoch": 0.6722689075630253,
      "grad_norm": 3.682716771358225,
      "learning_rate": 1.682909582004807e-06,
      "loss": 0.2069,
      "step": 105
    },
    {
      "epoch": 0.678671468587435,
      "grad_norm": 5.35989531173202,
      "learning_rate": 1.6746705459320744e-06,
      "loss": 0.2185,
      "step": 106
    },
    {
      "epoch": 0.6850740296118447,
      "grad_norm": 4.4810176716864225,
      "learning_rate": 1.6663465779520037e-06,
      "loss": 0.2179,
      "step": 107
    },
    {
      "epoch": 0.6914765906362546,
      "grad_norm": 4.781878607782948,
      "learning_rate": 1.6579387259397126e-06,
      "loss": 0.2162,
      "step": 108
    },
    {
      "epoch": 0.6978791516606643,
      "grad_norm": 3.578367620297612,
      "learning_rate": 1.6494480483301835e-06,
      "loss": 0.1864,
      "step": 109
    },
    {
      "epoch": 0.704281712685074,
      "grad_norm": 4.938703350422557,
      "learning_rate": 1.640875613985024e-06,
      "loss": 0.1789,
      "step": 110
    },
    {
      "epoch": 0.7106842737094838,
      "grad_norm": 4.712146890436173,
      "learning_rate": 1.6322225020579096e-06,
      "loss": 0.1971,
      "step": 111
    },
    {
      "epoch": 0.7170868347338936,
      "grad_norm": 4.49092614751852,
      "learning_rate": 1.6234898018587336e-06,
      "loss": 0.1796,
      "step": 112
    },
    {
      "epoch": 0.7234893957583033,
      "grad_norm": 4.123243996580139,
      "learning_rate": 1.6146786127164771e-06,
      "loss": 0.1703,
      "step": 113
    },
    {
      "epoch": 0.7298919567827131,
      "grad_norm": 3.4610648898646623,
      "learning_rate": 1.6057900438408199e-06,
      "loss": 0.2069,
      "step": 114
    },
    {
      "epoch": 0.7362945178071229,
      "grad_norm": 3.701626323345168,
      "learning_rate": 1.5968252141825035e-06,
      "loss": 0.2037,
      "step": 115
    },
    {
      "epoch": 0.7426970788315326,
      "grad_norm": 4.5853683365418965,
      "learning_rate": 1.587785252292473e-06,
      "loss": 0.1867,
      "step": 116
    },
    {
      "epoch": 0.7490996398559424,
      "grad_norm": 3.813531824888892,
      "learning_rate": 1.578671296179806e-06,
      "loss": 0.1923,
      "step": 117
    },
    {
      "epoch": 0.7555022008803521,
      "grad_norm": 4.333755687391246,
      "learning_rate": 1.569484493168452e-06,
      "loss": 0.1917,
      "step": 118
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 3.5490254589371526,
      "learning_rate": 1.5602259997528027e-06,
      "loss": 0.21,
      "step": 119
    },
    {
      "epoch": 0.7683073229291717,
      "grad_norm": 3.8349133617246665,
      "learning_rate": 1.5508969814521024e-06,
      "loss": 0.1839,
      "step": 120
    },
    {
      "epoch": 0.7747098839535814,
      "grad_norm": 4.480906522239498,
      "learning_rate": 1.5414986126637257e-06,
      "loss": 0.1804,
      "step": 121
    },
    {
      "epoch": 0.7811124449779911,
      "grad_norm": 4.063797691144723,
      "learning_rate": 1.5320320765153365e-06,
      "loss": 0.1872,
      "step": 122
    },
    {
      "epoch": 0.787515006002401,
      "grad_norm": 3.837906572094775,
      "learning_rate": 1.5224985647159488e-06,
      "loss": 0.202,
      "step": 123
    },
    {
      "epoch": 0.7939175670268107,
      "grad_norm": 4.226911617716516,
      "learning_rate": 1.5128992774059062e-06,
      "loss": 0.2067,
      "step": 124
    },
    {
      "epoch": 0.8003201280512204,
      "grad_norm": 4.143910488552411,
      "learning_rate": 1.5032354230058002e-06,
      "loss": 0.1884,
      "step": 125
    },
    {
      "epoch": 0.8067226890756303,
      "grad_norm": 3.996874874040649,
      "learning_rate": 1.4935082180643467e-06,
      "loss": 0.1797,
      "step": 126
    },
    {
      "epoch": 0.81312525010004,
      "grad_norm": 4.241850731448795,
      "learning_rate": 1.4837188871052397e-06,
      "loss": 0.1973,
      "step": 127
    },
    {
      "epoch": 0.8195278111244498,
      "grad_norm": 4.8128807497544015,
      "learning_rate": 1.4738686624729987e-06,
      "loss": 0.1888,
      "step": 128
    },
    {
      "epoch": 0.8259303721488596,
      "grad_norm": 4.624440170471318,
      "learning_rate": 1.463958784177834e-06,
      "loss": 0.1855,
      "step": 129
    },
    {
      "epoch": 0.8323329331732693,
      "grad_norm": 3.760993103519322,
      "learning_rate": 1.4539904997395467e-06,
      "loss": 0.1662,
      "step": 130
    },
    {
      "epoch": 0.8387354941976791,
      "grad_norm": 4.333505468408259,
      "learning_rate": 1.4439650640304821e-06,
      "loss": 0.2089,
      "step": 131
    },
    {
      "epoch": 0.8451380552220888,
      "grad_norm": 4.278934798609152,
      "learning_rate": 1.433883739117558e-06,
      "loss": 0.1879,
      "step": 132
    },
    {
      "epoch": 0.8515406162464986,
      "grad_norm": 4.16072632298661,
      "learning_rate": 1.4237477941033886e-06,
      "loss": 0.1904,
      "step": 133
    },
    {
      "epoch": 0.8579431772709084,
      "grad_norm": 3.5517189505372784,
      "learning_rate": 1.4135585049665206e-06,
      "loss": 0.2498,
      "step": 134
    },
    {
      "epoch": 0.8643457382953181,
      "grad_norm": 3.9706094312160776,
      "learning_rate": 1.4033171544008051e-06,
      "loss": 0.1901,
      "step": 135
    },
    {
      "epoch": 0.8707482993197279,
      "grad_norm": 4.171292985573587,
      "learning_rate": 1.3930250316539235e-06,
      "loss": 0.1835,
      "step": 136
    },
    {
      "epoch": 0.8771508603441377,
      "grad_norm": 3.8333520124526896,
      "learning_rate": 1.3826834323650898e-06,
      "loss": 0.1747,
      "step": 137
    },
    {
      "epoch": 0.8835534213685474,
      "grad_norm": 4.281605413414975,
      "learning_rate": 1.3722936584019451e-06,
      "loss": 0.2296,
      "step": 138
    },
    {
      "epoch": 0.8899559823929571,
      "grad_norm": 4.001127332533811,
      "learning_rate": 1.3618570176966722e-06,
      "loss": 0.2248,
      "step": 139
    },
    {
      "epoch": 0.896358543417367,
      "grad_norm": 3.2545123553526585,
      "learning_rate": 1.3513748240813427e-06,
      "loss": 0.2172,
      "step": 140
    },
    {
      "epoch": 0.9027611044417767,
      "grad_norm": 3.6921559524457073,
      "learning_rate": 1.3408483971225249e-06,
      "loss": 0.1865,
      "step": 141
    },
    {
      "epoch": 0.9091636654661864,
      "grad_norm": 4.729238595946433,
      "learning_rate": 1.3302790619551672e-06,
      "loss": 0.1511,
      "step": 142
    },
    {
      "epoch": 0.9155662264905963,
      "grad_norm": 4.27641471239857,
      "learning_rate": 1.3196681491157816e-06,
      "loss": 0.1775,
      "step": 143
    },
    {
      "epoch": 0.921968787515006,
      "grad_norm": 4.195425093653204,
      "learning_rate": 1.3090169943749473e-06,
      "loss": 0.2102,
      "step": 144
    },
    {
      "epoch": 0.9283713485394157,
      "grad_norm": 4.4927187865344305,
      "learning_rate": 1.298326938569156e-06,
      "loss": 0.1794,
      "step": 145
    },
    {
      "epoch": 0.9347739095638256,
      "grad_norm": 3.6780116242547467,
      "learning_rate": 1.2875993274320173e-06,
      "loss": 0.1716,
      "step": 146
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 3.5052737613672926,
      "learning_rate": 1.2768355114248492e-06,
      "loss": 0.1909,
      "step": 147
    },
    {
      "epoch": 0.947579031612645,
      "grad_norm": 5.4003852624340976,
      "learning_rate": 1.266036845566675e-06,
      "loss": 0.1828,
      "step": 148
    },
    {
      "epoch": 0.9539815926370548,
      "grad_norm": 4.293449029987786,
      "learning_rate": 1.2552046892636426e-06,
      "loss": 0.1934,
      "step": 149
    },
    {
      "epoch": 0.9603841536614646,
      "grad_norm": 4.213735783478494,
      "learning_rate": 1.244340406137894e-06,
      "loss": 0.1635,
      "step": 150
    },
    {
      "epoch": 0.9667867146858744,
      "grad_norm": 5.682803899024664,
      "learning_rate": 1.2334453638559054e-06,
      "loss": 0.2282,
      "step": 151
    },
    {
      "epoch": 0.9731892757102841,
      "grad_norm": 5.125716971978602,
      "learning_rate": 1.2225209339563143e-06,
      "loss": 0.2124,
      "step": 152
    },
    {
      "epoch": 0.9795918367346939,
      "grad_norm": 3.6296567686595558,
      "learning_rate": 1.211568491677263e-06,
      "loss": 0.1686,
      "step": 153
    },
    {
      "epoch": 0.9859943977591037,
      "grad_norm": 4.032149136826441,
      "learning_rate": 1.2005894157832728e-06,
      "loss": 0.1801,
      "step": 154
    },
    {
      "epoch": 0.9923969587835134,
      "grad_norm": 4.328877120814621,
      "learning_rate": 1.1895850883916785e-06,
      "loss": 0.2116,
      "step": 155
    },
    {
      "epoch": 0.9987995198079231,
      "grad_norm": 4.42679277844411,
      "learning_rate": 1.1785568947986366e-06,
      "loss": 0.197,
      "step": 156
    },
    {
      "epoch": 1.005202080832333,
      "grad_norm": 2.995825299760741,
      "learning_rate": 1.1675062233047363e-06,
      "loss": 0.1754,
      "step": 157
    },
    {
      "epoch": 1.0116046418567426,
      "grad_norm": 3.314198590902652,
      "learning_rate": 1.156434465040231e-06,
      "loss": 0.1591,
      "step": 158
    },
    {
      "epoch": 1.0180072028811524,
      "grad_norm": 3.282342144727049,
      "learning_rate": 1.1453430137899128e-06,
      "loss": 0.1463,
      "step": 159
    },
    {
      "epoch": 1.0244097639055623,
      "grad_norm": 3.3852348105131327,
      "learning_rate": 1.1342332658176555e-06,
      "loss": 0.1417,
      "step": 160
    },
    {
      "epoch": 1.0308123249299719,
      "grad_norm": 2.5846365344808224,
      "learning_rate": 1.123106619690643e-06,
      "loss": 0.1579,
      "step": 161
    },
    {
      "epoch": 1.0372148859543817,
      "grad_norm": 3.4112665397765407,
      "learning_rate": 1.1119644761033077e-06,
      "loss": 0.1569,
      "step": 162
    },
    {
      "epoch": 1.0436174469787916,
      "grad_norm": 3.057068352071886,
      "learning_rate": 1.1008082377010045e-06,
      "loss": 0.1588,
      "step": 163
    },
    {
      "epoch": 1.0500200080032012,
      "grad_norm": 2.480458097820055,
      "learning_rate": 1.0896393089034335e-06,
      "loss": 0.1604,
      "step": 164
    },
    {
      "epoch": 1.056422569027611,
      "grad_norm": 2.5238894472575,
      "learning_rate": 1.078459095727845e-06,
      "loss": 0.1738,
      "step": 165
    },
    {
      "epoch": 1.0628251300520208,
      "grad_norm": 2.702306669134078,
      "learning_rate": 1.0672690056120398e-06,
      "loss": 0.1416,
      "step": 166
    },
    {
      "epoch": 1.0692276910764307,
      "grad_norm": 2.715141146658483,
      "learning_rate": 1.0560704472371917e-06,
      "loss": 0.127,
      "step": 167
    },
    {
      "epoch": 1.0756302521008403,
      "grad_norm": 2.698908112656233,
      "learning_rate": 1.044864830350515e-06,
      "loss": 0.1769,
      "step": 168
    },
    {
      "epoch": 1.0820328131252501,
      "grad_norm": 2.393553347708748,
      "learning_rate": 1.033653565587794e-06,
      "loss": 0.1595,
      "step": 169
    },
    {
      "epoch": 1.08843537414966,
      "grad_norm": 2.515931323925557,
      "learning_rate": 1.022438064295805e-06,
      "loss": 0.1509,
      "step": 170
    },
    {
      "epoch": 1.0948379351740696,
      "grad_norm": 3.122208532437112,
      "learning_rate": 1.0112197383546459e-06,
      "loss": 0.1362,
      "step": 171
    },
    {
      "epoch": 1.1012404961984794,
      "grad_norm": 2.9259499260174437,
      "learning_rate": 1e-06,
      "loss": 0.1393,
      "step": 172
    },
    {
      "epoch": 1.1076430572228892,
      "grad_norm": 2.6643868148585783,
      "learning_rate": 9.88780261645354e-07,
      "loss": 0.1602,
      "step": 173
    },
    {
      "epoch": 1.1140456182472989,
      "grad_norm": 2.529601808890674,
      "learning_rate": 9.77561935704195e-07,
      "loss": 0.16,
      "step": 174
    },
    {
      "epoch": 1.1204481792717087,
      "grad_norm": 2.4798313875614038,
      "learning_rate": 9.663464344122063e-07,
      "loss": 0.1667,
      "step": 175
    },
    {
      "epoch": 1.1268507402961185,
      "grad_norm": 2.528807676927116,
      "learning_rate": 9.551351696494853e-07,
      "loss": 0.142,
      "step": 176
    },
    {
      "epoch": 1.1332533013205282,
      "grad_norm": 2.222384060658359,
      "learning_rate": 9.43929552762808e-07,
      "loss": 0.1273,
      "step": 177
    },
    {
      "epoch": 1.139655862344938,
      "grad_norm": 2.6020312291887993,
      "learning_rate": 9.327309943879603e-07,
      "loss": 0.1209,
      "step": 178
    },
    {
      "epoch": 1.1460584233693478,
      "grad_norm": 2.5865819736528213,
      "learning_rate": 9.215409042721551e-07,
      "loss": 0.1147,
      "step": 179
    },
    {
      "epoch": 1.1524609843937574,
      "grad_norm": 2.6984970538621047,
      "learning_rate": 9.103606910965665e-07,
      "loss": 0.1267,
      "step": 180
    },
    {
      "epoch": 1.1588635454181673,
      "grad_norm": 3.064928156142861,
      "learning_rate": 8.991917622989955e-07,
      "loss": 0.1472,
      "step": 181
    },
    {
      "epoch": 1.165266106442577,
      "grad_norm": 2.6751065232729787,
      "learning_rate": 8.880355238966921e-07,
      "loss": 0.1481,
      "step": 182
    },
    {
      "epoch": 1.1716686674669867,
      "grad_norm": 2.4919821908211137,
      "learning_rate": 8.768933803093572e-07,
      "loss": 0.1582,
      "step": 183
    },
    {
      "epoch": 1.1780712284913966,
      "grad_norm": 2.7496916062023926,
      "learning_rate": 8.657667341823448e-07,
      "loss": 0.1132,
      "step": 184
    },
    {
      "epoch": 1.1844737895158064,
      "grad_norm": 2.8683066132542065,
      "learning_rate": 8.546569862100875e-07,
      "loss": 0.1537,
      "step": 185
    },
    {
      "epoch": 1.190876350540216,
      "grad_norm": 3.3497753630792504,
      "learning_rate": 8.435655349597689e-07,
      "loss": 0.1498,
      "step": 186
    },
    {
      "epoch": 1.1972789115646258,
      "grad_norm": 2.8175991760028776,
      "learning_rate": 8.324937766952636e-07,
      "loss": 0.1344,
      "step": 187
    },
    {
      "epoch": 1.2036814725890357,
      "grad_norm": 3.5078725354724978,
      "learning_rate": 8.214431052013634e-07,
      "loss": 0.1794,
      "step": 188
    },
    {
      "epoch": 1.2100840336134453,
      "grad_norm": 3.134230661450569,
      "learning_rate": 8.104149116083216e-07,
      "loss": 0.1636,
      "step": 189
    },
    {
      "epoch": 1.2164865946378551,
      "grad_norm": 2.789058239974061,
      "learning_rate": 7.994105842167272e-07,
      "loss": 0.1384,
      "step": 190
    },
    {
      "epoch": 1.222889155662265,
      "grad_norm": 3.7547713671098966,
      "learning_rate": 7.884315083227372e-07,
      "loss": 0.1774,
      "step": 191
    },
    {
      "epoch": 1.2292917166866746,
      "grad_norm": 3.664307191430093,
      "learning_rate": 7.774790660436857e-07,
      "loss": 0.1481,
      "step": 192
    },
    {
      "epoch": 1.2356942777110844,
      "grad_norm": 2.771171440914553,
      "learning_rate": 7.665546361440949e-07,
      "loss": 0.1996,
      "step": 193
    },
    {
      "epoch": 1.2420968387354943,
      "grad_norm": 2.6893659053745815,
      "learning_rate": 7.556595938621058e-07,
      "loss": 0.1411,
      "step": 194
    },
    {
      "epoch": 1.2484993997599039,
      "grad_norm": 3.678898327438179,
      "learning_rate": 7.447953107363574e-07,
      "loss": 0.1864,
      "step": 195
    },
    {
      "epoch": 1.2549019607843137,
      "grad_norm": 3.520031085298818,
      "learning_rate": 7.33963154433325e-07,
      "loss": 0.1681,
      "step": 196
    },
    {
      "epoch": 1.2613045218087235,
      "grad_norm": 3.1974510727727354,
      "learning_rate": 7.231644885751507e-07,
      "loss": 0.1197,
      "step": 197
    },
    {
      "epoch": 1.2677070828331334,
      "grad_norm": 3.0861033592576703,
      "learning_rate": 7.124006725679828e-07,
      "loss": 0.1852,
      "step": 198
    },
    {
      "epoch": 1.274109643857543,
      "grad_norm": 2.392211457828515,
      "learning_rate": 7.016730614308439e-07,
      "loss": 0.1478,
      "step": 199
    },
    {
      "epoch": 1.2805122048819528,
      "grad_norm": 2.6382593348580636,
      "learning_rate": 6.909830056250526e-07,
      "loss": 0.1499,
      "step": 200
    },
    {
      "epoch": 1.2869147659063627,
      "grad_norm": 3.310685530214126,
      "learning_rate": 6.803318508842186e-07,
      "loss": 0.1892,
      "step": 201
    },
    {
      "epoch": 1.2933173269307723,
      "grad_norm": 2.9330195000820054,
      "learning_rate": 6.697209380448332e-07,
      "loss": 0.1853,
      "step": 202
    },
    {
      "epoch": 1.2997198879551821,
      "grad_norm": 2.8596301823674115,
      "learning_rate": 6.59151602877475e-07,
      "loss": 0.1328,
      "step": 203
    },
    {
      "epoch": 1.306122448979592,
      "grad_norm": 2.794306696992077,
      "learning_rate": 6.486251759186572e-07,
      "loss": 0.1807,
      "step": 204
    },
    {
      "epoch": 1.3125250100040016,
      "grad_norm": 4.390407508756913,
      "learning_rate": 6.381429823033279e-07,
      "loss": 0.1536,
      "step": 205
    },
    {
      "epoch": 1.3189275710284114,
      "grad_norm": 2.6300011319364915,
      "learning_rate": 6.277063415980548e-07,
      "loss": 0.1431,
      "step": 206
    },
    {
      "epoch": 1.3253301320528212,
      "grad_norm": 2.62332174376127,
      "learning_rate": 6.173165676349102e-07,
      "loss": 0.1482,
      "step": 207
    },
    {
      "epoch": 1.3317326930772309,
      "grad_norm": 2.733993586202797,
      "learning_rate": 6.069749683460764e-07,
      "loss": 0.1228,
      "step": 208
    },
    {
      "epoch": 1.3381352541016407,
      "grad_norm": 3.315910975674784,
      "learning_rate": 5.96682845599195e-07,
      "loss": 0.2054,
      "step": 209
    },
    {
      "epoch": 1.3445378151260505,
      "grad_norm": 2.930927185763075,
      "learning_rate": 5.864414950334795e-07,
      "loss": 0.159,
      "step": 210
    },
    {
      "epoch": 1.3509403761504601,
      "grad_norm": 3.1317606911937568,
      "learning_rate": 5.762522058966113e-07,
      "loss": 0.1586,
      "step": 211
    },
    {
      "epoch": 1.35734293717487,
      "grad_norm": 4.040118089874495,
      "learning_rate": 5.661162608824419e-07,
      "loss": 0.1505,
      "step": 212
    },
    {
      "epoch": 1.3637454981992798,
      "grad_norm": 2.746598755946971,
      "learning_rate": 5.56034935969518e-07,
      "loss": 0.1761,
      "step": 213
    },
    {
      "epoch": 1.3701480592236894,
      "grad_norm": 2.9141623052335137,
      "learning_rate": 5.460095002604532e-07,
      "loss": 0.19,
      "step": 214
    },
    {
      "epoch": 1.3765506202480993,
      "grad_norm": 4.001391421545072,
      "learning_rate": 5.36041215822166e-07,
      "loss": 0.1321,
      "step": 215
    },
    {
      "epoch": 1.382953181272509,
      "grad_norm": 3.6428702220535736,
      "learning_rate": 5.261313375270013e-07,
      "loss": 0.1905,
      "step": 216
    },
    {
      "epoch": 1.3893557422969187,
      "grad_norm": 3.324099015382822,
      "learning_rate": 5.162811128947602e-07,
      "loss": 0.1669,
      "step": 217
    },
    {
      "epoch": 1.3957583033213286,
      "grad_norm": 2.6236595277867294,
      "learning_rate": 5.064917819356531e-07,
      "loss": 0.1235,
      "step": 218
    },
    {
      "epoch": 1.4021608643457384,
      "grad_norm": 3.259881675873255,
      "learning_rate": 4.967645769941999e-07,
      "loss": 0.1814,
      "step": 219
    },
    {
      "epoch": 1.408563425370148,
      "grad_norm": 2.8694964354069117,
      "learning_rate": 4.871007225940939e-07,
      "loss": 0.1495,
      "step": 220
    },
    {
      "epoch": 1.4149659863945578,
      "grad_norm": 2.901840928815238,
      "learning_rate": 4.775014352840512e-07,
      "loss": 0.1951,
      "step": 221
    },
    {
      "epoch": 1.4213685474189677,
      "grad_norm": 3.418089743568751,
      "learning_rate": 4.6796792348466353e-07,
      "loss": 0.1547,
      "step": 222
    },
    {
      "epoch": 1.4277711084433773,
      "grad_norm": 2.830949019249368,
      "learning_rate": 4.585013873362743e-07,
      "loss": 0.1046,
      "step": 223
    },
    {
      "epoch": 1.4341736694677871,
      "grad_norm": 3.5740356720869206,
      "learning_rate": 4.4910301854789755e-07,
      "loss": 0.1482,
      "step": 224
    },
    {
      "epoch": 1.440576230492197,
      "grad_norm": 2.7706265692202217,
      "learning_rate": 4.397740002471972e-07,
      "loss": 0.1842,
      "step": 225
    },
    {
      "epoch": 1.4469787915166066,
      "grad_norm": 2.705953973970954,
      "learning_rate": 4.3051550683154804e-07,
      "loss": 0.173,
      "step": 226
    },
    {
      "epoch": 1.4533813525410164,
      "grad_norm": 3.5200476993463976,
      "learning_rate": 4.2132870382019427e-07,
      "loss": 0.1268,
      "step": 227
    },
    {
      "epoch": 1.4597839135654262,
      "grad_norm": 2.980462619255481,
      "learning_rate": 4.1221474770752696e-07,
      "loss": 0.1454,
      "step": 228
    },
    {
      "epoch": 1.4661864745898359,
      "grad_norm": 3.0488417880599274,
      "learning_rate": 4.031747858174964e-07,
      "loss": 0.1573,
      "step": 229
    },
    {
      "epoch": 1.4725890356142457,
      "grad_norm": 2.866558859217372,
      "learning_rate": 3.942099561591802e-07,
      "loss": 0.1354,
      "step": 230
    },
    {
      "epoch": 1.4789915966386555,
      "grad_norm": 3.3643880310904017,
      "learning_rate": 3.853213872835228e-07,
      "loss": 0.1527,
      "step": 231
    },
    {
      "epoch": 1.4853941576630652,
      "grad_norm": 2.3493487438896463,
      "learning_rate": 3.765101981412665e-07,
      "loss": 0.1483,
      "step": 232
    },
    {
      "epoch": 1.491796718687475,
      "grad_norm": 2.482947067390714,
      "learning_rate": 3.677774979420903e-07,
      "loss": 0.1302,
      "step": 233
    },
    {
      "epoch": 1.4981992797118848,
      "grad_norm": 2.599751282623492,
      "learning_rate": 3.5912438601497584e-07,
      "loss": 0.1522,
      "step": 234
    },
    {
      "epoch": 1.5046018407362944,
      "grad_norm": 3.191973535201136,
      "learning_rate": 3.5055195166981646e-07,
      "loss": 0.1615,
      "step": 235
    },
    {
      "epoch": 1.5110044017607043,
      "grad_norm": 3.15562179961526,
      "learning_rate": 3.420612740602874e-07,
      "loss": 0.1478,
      "step": 236
    },
    {
      "epoch": 1.517406962785114,
      "grad_norm": 2.8880563361860823,
      "learning_rate": 3.3365342204799606e-07,
      "loss": 0.1397,
      "step": 237
    },
    {
      "epoch": 1.5238095238095237,
      "grad_norm": 3.688725379316976,
      "learning_rate": 3.253294540679257e-07,
      "loss": 0.1875,
      "step": 238
    },
    {
      "epoch": 1.5302120848339336,
      "grad_norm": 2.817507541571056,
      "learning_rate": 3.170904179951931e-07,
      "loss": 0.1216,
      "step": 239
    },
    {
      "epoch": 1.5366146458583434,
      "grad_norm": 2.9412647550371975,
      "learning_rate": 3.0893735101313535e-07,
      "loss": 0.1398,
      "step": 240
    },
    {
      "epoch": 1.543017206882753,
      "grad_norm": 2.762627387485447,
      "learning_rate": 3.008712794827426e-07,
      "loss": 0.1537,
      "step": 241
    },
    {
      "epoch": 1.5494197679071628,
      "grad_norm": 2.8082508660384633,
      "learning_rate": 2.9289321881345254e-07,
      "loss": 0.1367,
      "step": 242
    },
    {
      "epoch": 1.5558223289315727,
      "grad_norm": 3.1843132713343314,
      "learning_rate": 2.850041733353247e-07,
      "loss": 0.1451,
      "step": 243
    },
    {
      "epoch": 1.5622248899559823,
      "grad_norm": 2.549326946513764,
      "learning_rate": 2.7720513617260855e-07,
      "loss": 0.1578,
      "step": 244
    },
    {
      "epoch": 1.5686274509803921,
      "grad_norm": 2.035464821513073,
      "learning_rate": 2.6949708911872247e-07,
      "loss": 0.1271,
      "step": 245
    },
    {
      "epoch": 1.575030012004802,
      "grad_norm": 2.494783934906446,
      "learning_rate": 2.6188100251265943e-07,
      "loss": 0.1446,
      "step": 246
    },
    {
      "epoch": 1.5814325730292116,
      "grad_norm": 4.226886500584243,
      "learning_rate": 2.543578351168344e-07,
      "loss": 0.1417,
      "step": 247
    },
    {
      "epoch": 1.5878351340536214,
      "grad_norm": 2.495677422705392,
      "learning_rate": 2.4692853399638913e-07,
      "loss": 0.1364,
      "step": 248
    },
    {
      "epoch": 1.5942376950780313,
      "grad_norm": 3.1917365323900446,
      "learning_rate": 2.395940343999691e-07,
      "loss": 0.1626,
      "step": 249
    },
    {
      "epoch": 1.6006402561024409,
      "grad_norm": 3.0344619777547868,
      "learning_rate": 2.3235525964198888e-07,
      "loss": 0.1623,
      "step": 250
    },
    {
      "epoch": 1.6070428171268507,
      "grad_norm": 3.688983622501882,
      "learning_rate": 2.252131209863991e-07,
      "loss": 0.1654,
      "step": 251
    },
    {
      "epoch": 1.6134453781512605,
      "grad_norm": 2.760266999319,
      "learning_rate": 2.181685175319702e-07,
      "loss": 0.1757,
      "step": 252
    },
    {
      "epoch": 1.6198479391756702,
      "grad_norm": 3.3493718404784816,
      "learning_rate": 2.11222336099109e-07,
      "loss": 0.1877,
      "step": 253
    },
    {
      "epoch": 1.62625050020008,
      "grad_norm": 2.7059015648474767,
      "learning_rate": 2.043754511182191e-07,
      "loss": 0.1302,
      "step": 254
    },
    {
      "epoch": 1.6326530612244898,
      "grad_norm": 2.6710609583138636,
      "learning_rate": 1.9762872451962208e-07,
      "loss": 0.1376,
      "step": 255
    },
    {
      "epoch": 1.6390556222488994,
      "grad_norm": 2.636600973542151,
      "learning_rate": 1.9098300562505264e-07,
      "loss": 0.1667,
      "step": 256
    },
    {
      "epoch": 1.6454581832733093,
      "grad_norm": 2.7163444391021647,
      "learning_rate": 1.8443913104073982e-07,
      "loss": 0.121,
      "step": 257
    },
    {
      "epoch": 1.6518607442977191,
      "grad_norm": 5.493549571701205,
      "learning_rate": 1.7799792455209016e-07,
      "loss": 0.1702,
      "step": 258
    },
    {
      "epoch": 1.6582633053221287,
      "grad_norm": 2.795867026435785,
      "learning_rate": 1.716601970199836e-07,
      "loss": 0.1657,
      "step": 259
    },
    {
      "epoch": 1.6646658663465386,
      "grad_norm": 2.697456578390499,
      "learning_rate": 1.6542674627869734e-07,
      "loss": 0.1108,
      "step": 260
    },
    {
      "epoch": 1.6710684273709484,
      "grad_norm": 3.700105341120021,
      "learning_rate": 1.592983570354699e-07,
      "loss": 0.1526,
      "step": 261
    },
    {
      "epoch": 1.677470988395358,
      "grad_norm": 2.8767911269966957,
      "learning_rate": 1.5327580077171588e-07,
      "loss": 0.1581,
      "step": 262
    },
    {
      "epoch": 1.6838735494197679,
      "grad_norm": 2.8392553114328183,
      "learning_rate": 1.473598356459078e-07,
      "loss": 0.1387,
      "step": 263
    },
    {
      "epoch": 1.6902761104441777,
      "grad_norm": 2.848243119062683,
      "learning_rate": 1.415512063981339e-07,
      "loss": 0.1352,
      "step": 264
    },
    {
      "epoch": 1.6966786714685873,
      "grad_norm": 4.008263998101646,
      "learning_rate": 1.358506442563454e-07,
      "loss": 0.1717,
      "step": 265
    },
    {
      "epoch": 1.7030812324929971,
      "grad_norm": 3.3644795533763747,
      "learning_rate": 1.3025886684430465e-07,
      "loss": 0.1851,
      "step": 266
    },
    {
      "epoch": 1.709483793517407,
      "grad_norm": 2.1927113979713457,
      "learning_rate": 1.2477657809124632e-07,
      "loss": 0.1529,
      "step": 267
    },
    {
      "epoch": 1.7158863545418166,
      "grad_norm": 4.20748377167738,
      "learning_rate": 1.19404468143262e-07,
      "loss": 0.1506,
      "step": 268
    },
    {
      "epoch": 1.7222889155662267,
      "grad_norm": 2.6119161837028897,
      "learning_rate": 1.1414321327642019e-07,
      "loss": 0.1692,
      "step": 269
    },
    {
      "epoch": 1.7286914765906363,
      "grad_norm": 3.25899049531815,
      "learning_rate": 1.089934758116322e-07,
      "loss": 0.1552,
      "step": 270
    },
    {
      "epoch": 1.7350940376150459,
      "grad_norm": 2.4457520889715085,
      "learning_rate": 1.0395590403127486e-07,
      "loss": 0.1072,
      "step": 271
    },
    {
      "epoch": 1.741496598639456,
      "grad_norm": 3.725879684312294,
      "learning_rate": 9.903113209758096e-08,
      "loss": 0.1297,
      "step": 272
    },
    {
      "epoch": 1.7478991596638656,
      "grad_norm": 4.086760833693879,
      "learning_rate": 9.421977997280594e-08,
      "loss": 0.134,
      "step": 273
    },
    {
      "epoch": 1.7543017206882752,
      "grad_norm": 2.8228633836720944,
      "learning_rate": 8.952245334118413e-08,
      "loss": 0.1429,
      "step": 274
    },
    {
      "epoch": 1.7607042817126852,
      "grad_norm": 2.5768217353103906,
      "learning_rate": 8.493974353268019e-08,
      "loss": 0.138,
      "step": 275
    },
    {
      "epoch": 1.7671068427370948,
      "grad_norm": 3.920988018065823,
      "learning_rate": 8.047222744854942e-08,
      "loss": 0.168,
      "step": 276
    },
    {
      "epoch": 1.7735094037615045,
      "grad_norm": 2.665420920196723,
      "learning_rate": 7.612046748871326e-08,
      "loss": 0.1198,
      "step": 277
    },
    {
      "epoch": 1.7799119647859145,
      "grad_norm": 2.84264633693513,
      "learning_rate": 7.188501148096116e-08,
      "loss": 0.1724,
      "step": 278
    },
    {
      "epoch": 1.7863145258103241,
      "grad_norm": 2.7132283377188386,
      "learning_rate": 6.77663926119858e-08,
      "loss": 0.1369,
      "step": 279
    },
    {
      "epoch": 1.7927170868347337,
      "grad_norm": 2.8632960787364494,
      "learning_rate": 6.376512936026279e-08,
      "loss": 0.1606,
      "step": 280
    },
    {
      "epoch": 1.7991196478591438,
      "grad_norm": 2.738598278709088,
      "learning_rate": 5.988172543078096e-08,
      "loss": 0.1679,
      "step": 281
    },
    {
      "epoch": 1.8055222088835534,
      "grad_norm": 2.311034901841601,
      "learning_rate": 5.611666969163242e-08,
      "loss": 0.1644,
      "step": 282
    },
    {
      "epoch": 1.811924769907963,
      "grad_norm": 2.8470945573238837,
      "learning_rate": 5.2470436112471264e-08,
      "loss": 0.1817,
      "step": 283
    },
    {
      "epoch": 1.818327330932373,
      "grad_norm": 2.9917534547915476,
      "learning_rate": 4.8943483704846465e-08,
      "loss": 0.1472,
      "step": 284
    },
    {
      "epoch": 1.8247298919567827,
      "grad_norm": 2.6305242577097685,
      "learning_rate": 4.553625646441928e-08,
      "loss": 0.1289,
      "step": 285
    },
    {
      "epoch": 1.8311324529811923,
      "grad_norm": 2.5380081332282014,
      "learning_rate": 4.224918331506955e-08,
      "loss": 0.1379,
      "step": 286
    },
    {
      "epoch": 1.8375350140056024,
      "grad_norm": 3.110480752833179,
      "learning_rate": 3.908267805490051e-08,
      "loss": 0.1623,
      "step": 287
    },
    {
      "epoch": 1.843937575030012,
      "grad_norm": 3.1789215828737074,
      "learning_rate": 3.6037139304146756e-08,
      "loss": 0.1441,
      "step": 288
    },
    {
      "epoch": 1.8503401360544216,
      "grad_norm": 3.2716298643823927,
      "learning_rate": 3.3112950454993625e-08,
      "loss": 0.166,
      "step": 289
    },
    {
      "epoch": 1.8567426970788317,
      "grad_norm": 4.5463007787733725,
      "learning_rate": 3.0310479623313125e-08,
      "loss": 0.1391,
      "step": 290
    },
    {
      "epoch": 1.8631452581032413,
      "grad_norm": 3.0644978198504966,
      "learning_rate": 2.7630079602323443e-08,
      "loss": 0.1359,
      "step": 291
    },
    {
      "epoch": 1.8695478191276511,
      "grad_norm": 2.915540706565416,
      "learning_rate": 2.507208781817638e-08,
      "loss": 0.1854,
      "step": 292
    },
    {
      "epoch": 1.875950380152061,
      "grad_norm": 2.7471867583483958,
      "learning_rate": 2.263682628748087e-08,
      "loss": 0.1593,
      "step": 293
    },
    {
      "epoch": 1.8823529411764706,
      "grad_norm": 3.3996282117269323,
      "learning_rate": 2.032460157676452e-08,
      "loss": 0.1655,
      "step": 294
    },
    {
      "epoch": 1.8887555022008804,
      "grad_norm": 2.782949881418279,
      "learning_rate": 1.8135704763881598e-08,
      "loss": 0.1411,
      "step": 295
    },
    {
      "epoch": 1.8951580632252902,
      "grad_norm": 2.817009087504296,
      "learning_rate": 1.607041140137033e-08,
      "loss": 0.1741,
      "step": 296
    },
    {
      "epoch": 1.9015606242496998,
      "grad_norm": 2.514339605431768,
      "learning_rate": 1.4128981481764113e-08,
      "loss": 0.1639,
      "step": 297
    },
    {
      "epoch": 1.9079631852741097,
      "grad_norm": 2.502723430413198,
      "learning_rate": 1.231165940486234e-08,
      "loss": 0.1458,
      "step": 298
    },
    {
      "epoch": 1.9143657462985195,
      "grad_norm": 2.9756989992004,
      "learning_rate": 1.0618673946963364e-08,
      "loss": 0.1899,
      "step": 299
    },
    {
      "epoch": 1.9207683073229291,
      "grad_norm": 3.029713811900019,
      "learning_rate": 9.050238232065299e-09,
      "loss": 0.1194,
      "step": 300
    },
    {
      "epoch": 1.927170868347339,
      "grad_norm": 2.8129240654015555,
      "learning_rate": 7.606549705035935e-09,
      "loss": 0.1564,
      "step": 301
    },
    {
      "epoch": 1.9335734293717488,
      "grad_norm": 3.5641124301233904,
      "learning_rate": 6.2877901067573955e-09,
      "loss": 0.1272,
      "step": 302
    },
    {
      "epoch": 1.9399759903961584,
      "grad_norm": 2.5825170270283024,
      "learning_rate": 5.094125451247655e-09,
      "loss": 0.1864,
      "step": 303
    },
    {
      "epoch": 1.9463785514205683,
      "grad_norm": 2.75200131179709,
      "learning_rate": 4.025706004760931e-09,
      "loss": 0.1568,
      "step": 304
    },
    {
      "epoch": 1.952781112444978,
      "grad_norm": 3.1581337599381527,
      "learning_rate": 3.082666266872036e-09,
      "loss": 0.1256,
      "step": 305
    },
    {
      "epoch": 1.9591836734693877,
      "grad_norm": 3.4365108386154786,
      "learning_rate": 2.2651249535439177e-09,
      "loss": 0.1516,
      "step": 306
    },
    {
      "epoch": 1.9655862344937975,
      "grad_norm": 2.552383623057842,
      "learning_rate": 1.5731849821833953e-09,
      "loss": 0.1362,
      "step": 307
    },
    {
      "epoch": 1.9719887955182074,
      "grad_norm": 2.5409532995249657,
      "learning_rate": 1.0069334586854105e-09,
      "loss": 0.1238,
      "step": 308
    },
    {
      "epoch": 1.978391356542617,
      "grad_norm": 4.158643521191336,
      "learning_rate": 5.664416664666882e-10,
      "loss": 0.1615,
      "step": 309
    },
    {
      "epoch": 1.9847939175670268,
      "grad_norm": 3.0805263918986983,
      "learning_rate": 2.517650574934693e-10,
      "loss": 0.1513,
      "step": 310
    },
    {
      "epoch": 1.9911964785914367,
      "grad_norm": 3.1343235054789926,
      "learning_rate": 6.29432452994294e-11,
      "loss": 0.1493,
      "step": 311
    },
    {
      "epoch": 1.9975990396158463,
      "grad_norm": 4.792054679066387,
      "learning_rate": 0.0,
      "loss": 0.167,
      "step": 312
    },
    {
      "epoch": 1.9975990396158463,
      "step": 312,
      "total_flos": 211327818203136.0,
      "train_loss": 0.17435487460058469,
      "train_runtime": 7374.5307,
      "train_samples_per_second": 10.842,
      "train_steps_per_second": 0.042
    }
  ],
  "logging_steps": 1,
  "max_steps": 312,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 211327818203136.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}