{
  "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": 13.680483919438638,
      "learning_rate": 6.25e-08,
      "loss": 0.7419,
      "step": 1
    },
    {
      "epoch": 0.012805122048819529,
      "grad_norm": 12.895832423301675,
      "learning_rate": 1.25e-07,
      "loss": 0.6763,
      "step": 2
    },
    {
      "epoch": 0.01920768307322929,
      "grad_norm": 12.171002116468559,
      "learning_rate": 1.875e-07,
      "loss": 0.6351,
      "step": 3
    },
    {
      "epoch": 0.025610244097639057,
      "grad_norm": 12.464785861868467,
      "learning_rate": 2.5e-07,
      "loss": 0.7288,
      "step": 4
    },
    {
      "epoch": 0.03201280512204882,
      "grad_norm": 10.388017670547901,
      "learning_rate": 3.1249999999999997e-07,
      "loss": 0.6076,
      "step": 5
    },
    {
      "epoch": 0.03841536614645858,
      "grad_norm": 12.079460028084922,
      "learning_rate": 3.75e-07,
      "loss": 0.6902,
      "step": 6
    },
    {
      "epoch": 0.04481792717086835,
      "grad_norm": 12.19876157935689,
      "learning_rate": 4.375e-07,
      "loss": 0.6948,
      "step": 7
    },
    {
      "epoch": 0.051220488195278115,
      "grad_norm": 9.756977949266757,
      "learning_rate": 5e-07,
      "loss": 0.7175,
      "step": 8
    },
    {
      "epoch": 0.057623049219687875,
      "grad_norm": 10.46968505857669,
      "learning_rate": 5.625e-07,
      "loss": 0.6654,
      "step": 9
    },
    {
      "epoch": 0.06402561024409764,
      "grad_norm": 8.571146829037712,
      "learning_rate": 6.249999999999999e-07,
      "loss": 0.6365,
      "step": 10
    },
    {
      "epoch": 0.07042817126850741,
      "grad_norm": 7.915474952782504,
      "learning_rate": 6.875e-07,
      "loss": 0.6584,
      "step": 11
    },
    {
      "epoch": 0.07683073229291716,
      "grad_norm": 8.017393974111991,
      "learning_rate": 7.5e-07,
      "loss": 0.6789,
      "step": 12
    },
    {
      "epoch": 0.08323329331732693,
      "grad_norm": 7.146421473324447,
      "learning_rate": 8.125e-07,
      "loss": 0.6551,
      "step": 13
    },
    {
      "epoch": 0.0896358543417367,
      "grad_norm": 6.490398797125461,
      "learning_rate": 8.75e-07,
      "loss": 0.5721,
      "step": 14
    },
    {
      "epoch": 0.09603841536614646,
      "grad_norm": 6.813176303739041,
      "learning_rate": 9.374999999999999e-07,
      "loss": 0.5933,
      "step": 15
    },
    {
      "epoch": 0.10244097639055623,
      "grad_norm": 6.452956439451391,
      "learning_rate": 1e-06,
      "loss": 0.5332,
      "step": 16
    },
    {
      "epoch": 0.10884353741496598,
      "grad_norm": 6.639180709933743,
      "learning_rate": 1.0625e-06,
      "loss": 0.5836,
      "step": 17
    },
    {
      "epoch": 0.11524609843937575,
      "grad_norm": 5.461791194778153,
      "learning_rate": 1.125e-06,
      "loss": 0.5577,
      "step": 18
    },
    {
      "epoch": 0.12164865946378552,
      "grad_norm": 5.930832649263125,
      "learning_rate": 1.1874999999999999e-06,
      "loss": 0.5337,
      "step": 19
    },
    {
      "epoch": 0.12805122048819528,
      "grad_norm": 4.956904837601893,
      "learning_rate": 1.2499999999999999e-06,
      "loss": 0.4669,
      "step": 20
    },
    {
      "epoch": 0.13445378151260504,
      "grad_norm": 6.256208371824237,
      "learning_rate": 1.3125e-06,
      "loss": 0.4811,
      "step": 21
    },
    {
      "epoch": 0.14085634253701482,
      "grad_norm": 5.246513837804897,
      "learning_rate": 1.375e-06,
      "loss": 0.4414,
      "step": 22
    },
    {
      "epoch": 0.14725890356142457,
      "grad_norm": 5.528561343115844,
      "learning_rate": 1.4375e-06,
      "loss": 0.4948,
      "step": 23
    },
    {
      "epoch": 0.15366146458583432,
      "grad_norm": 5.333224339149711,
      "learning_rate": 1.5e-06,
      "loss": 0.4281,
      "step": 24
    },
    {
      "epoch": 0.1600640256102441,
      "grad_norm": 4.9864846595798324,
      "learning_rate": 1.5624999999999999e-06,
      "loss": 0.4362,
      "step": 25
    },
    {
      "epoch": 0.16646658663465386,
      "grad_norm": 5.269516171255013,
      "learning_rate": 1.625e-06,
      "loss": 0.4036,
      "step": 26
    },
    {
      "epoch": 0.17286914765906364,
      "grad_norm": 4.932131322685013,
      "learning_rate": 1.6875e-06,
      "loss": 0.4038,
      "step": 27
    },
    {
      "epoch": 0.1792717086834734,
      "grad_norm": 5.318953799775683,
      "learning_rate": 1.75e-06,
      "loss": 0.4283,
      "step": 28
    },
    {
      "epoch": 0.18567426970788314,
      "grad_norm": 5.190854642160747,
      "learning_rate": 1.8125e-06,
      "loss": 0.4044,
      "step": 29
    },
    {
      "epoch": 0.19207683073229292,
      "grad_norm": 4.807806644055786,
      "learning_rate": 1.8749999999999998e-06,
      "loss": 0.4117,
      "step": 30
    },
    {
      "epoch": 0.19847939175670268,
      "grad_norm": 4.292846106324363,
      "learning_rate": 1.9375e-06,
      "loss": 0.3416,
      "step": 31
    },
    {
      "epoch": 0.20488195278111246,
      "grad_norm": 4.03903415559866,
      "learning_rate": 2e-06,
      "loss": 0.3794,
      "step": 32
    },
    {
      "epoch": 0.2112845138055222,
      "grad_norm": 4.493745802649108,
      "learning_rate": 1.9999370567547003e-06,
      "loss": 0.3952,
      "step": 33
    },
    {
      "epoch": 0.21768707482993196,
      "grad_norm": 4.645817807816507,
      "learning_rate": 1.9997482349425066e-06,
      "loss": 0.3743,
      "step": 34
    },
    {
      "epoch": 0.22408963585434175,
      "grad_norm": 5.0435551125438,
      "learning_rate": 1.9994335583335335e-06,
      "loss": 0.3525,
      "step": 35
    },
    {
      "epoch": 0.2304921968787515,
      "grad_norm": 4.788576656798642,
      "learning_rate": 1.9989930665413145e-06,
      "loss": 0.3865,
      "step": 36
    },
    {
      "epoch": 0.23689475790316125,
      "grad_norm": 4.390755031941915,
      "learning_rate": 1.9984268150178167e-06,
      "loss": 0.3295,
      "step": 37
    },
    {
      "epoch": 0.24329731892757103,
      "grad_norm": 4.772214941084568,
      "learning_rate": 1.997734875046456e-06,
      "loss": 0.3334,
      "step": 38
    },
    {
      "epoch": 0.24969987995198079,
      "grad_norm": 5.273946498985277,
      "learning_rate": 1.996917333733128e-06,
      "loss": 0.3726,
      "step": 39
    },
    {
      "epoch": 0.25610244097639057,
      "grad_norm": 5.369010852231794,
      "learning_rate": 1.995974293995239e-06,
      "loss": 0.3707,
      "step": 40
    },
    {
      "epoch": 0.26250500200080035,
      "grad_norm": 5.767047861871442,
      "learning_rate": 1.994905874548752e-06,
      "loss": 0.3604,
      "step": 41
    },
    {
      "epoch": 0.2689075630252101,
      "grad_norm": 5.503870643763186,
      "learning_rate": 1.9937122098932426e-06,
      "loss": 0.3746,
      "step": 42
    },
    {
      "epoch": 0.27531012404961985,
      "grad_norm": 6.3642035735044935,
      "learning_rate": 1.9923934502949643e-06,
      "loss": 0.3524,
      "step": 43
    },
    {
      "epoch": 0.28171268507402963,
      "grad_norm": 5.870101299422478,
      "learning_rate": 1.9909497617679347e-06,
      "loss": 0.3672,
      "step": 44
    },
    {
      "epoch": 0.28811524609843936,
      "grad_norm": 4.907548990312214,
      "learning_rate": 1.9893813260530367e-06,
      "loss": 0.3748,
      "step": 45
    },
    {
      "epoch": 0.29451780712284914,
      "grad_norm": 4.106972565193302,
      "learning_rate": 1.9876883405951377e-06,
      "loss": 0.3265,
      "step": 46
    },
    {
      "epoch": 0.3009203681472589,
      "grad_norm": 4.649037094549095,
      "learning_rate": 1.9858710185182355e-06,
      "loss": 0.2802,
      "step": 47
    },
    {
      "epoch": 0.30732292917166865,
      "grad_norm": 4.578190605072973,
      "learning_rate": 1.9839295885986295e-06,
      "loss": 0.3414,
      "step": 48
    },
    {
      "epoch": 0.3137254901960784,
      "grad_norm": 5.338755744928473,
      "learning_rate": 1.9818642952361183e-06,
      "loss": 0.3513,
      "step": 49
    },
    {
      "epoch": 0.3201280512204882,
      "grad_norm": 5.16732801301707,
      "learning_rate": 1.9796753984232355e-06,
      "loss": 0.3385,
      "step": 50
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 4.308178011035292,
      "learning_rate": 1.977363173712519e-06,
      "loss": 0.3337,
      "step": 51
    },
    {
      "epoch": 0.3329331732693077,
      "grad_norm": 4.047304578891064,
      "learning_rate": 1.9749279121818236e-06,
      "loss": 0.3536,
      "step": 52
    },
    {
      "epoch": 0.3393357342937175,
      "grad_norm": 4.189108068341681,
      "learning_rate": 1.9723699203976766e-06,
      "loss": 0.3007,
      "step": 53
    },
    {
      "epoch": 0.3457382953181273,
      "grad_norm": 4.150519885409928,
      "learning_rate": 1.9696895203766866e-06,
      "loss": 0.2981,
      "step": 54
    },
    {
      "epoch": 0.352140856342537,
      "grad_norm": 4.59545667540724,
      "learning_rate": 1.966887049545006e-06,
      "loss": 0.3121,
      "step": 55
    },
    {
      "epoch": 0.3585434173669468,
      "grad_norm": 3.9736439160316683,
      "learning_rate": 1.9639628606958534e-06,
      "loss": 0.3049,
      "step": 56
    },
    {
      "epoch": 0.36494597839135656,
      "grad_norm": 4.3054703674323,
      "learning_rate": 1.9609173219450997e-06,
      "loss": 0.3124,
      "step": 57
    },
    {
      "epoch": 0.3713485394157663,
      "grad_norm": 3.4948815561353537,
      "learning_rate": 1.9577508166849303e-06,
      "loss": 0.3013,
      "step": 58
    },
    {
      "epoch": 0.37775110044017607,
      "grad_norm": 3.741965125562426,
      "learning_rate": 1.9544637435355806e-06,
      "loss": 0.2894,
      "step": 59
    },
    {
      "epoch": 0.38415366146458585,
      "grad_norm": 4.280653888850941,
      "learning_rate": 1.9510565162951534e-06,
      "loss": 0.3134,
      "step": 60
    },
    {
      "epoch": 0.3905562224889956,
      "grad_norm": 4.405700671682634,
      "learning_rate": 1.947529563887529e-06,
      "loss": 0.3177,
      "step": 61
    },
    {
      "epoch": 0.39695878351340536,
      "grad_norm": 3.9394485247143636,
      "learning_rate": 1.9438833303083674e-06,
      "loss": 0.2966,
      "step": 62
    },
    {
      "epoch": 0.40336134453781514,
      "grad_norm": 4.389702795617689,
      "learning_rate": 1.9401182745692187e-06,
      "loss": 0.2903,
      "step": 63
    },
    {
      "epoch": 0.4097639055622249,
      "grad_norm": 3.5979658724911077,
      "learning_rate": 1.936234870639737e-06,
      "loss": 0.2953,
      "step": 64
    },
    {
      "epoch": 0.41616646658663464,
      "grad_norm": 4.47441718112865,
      "learning_rate": 1.9322336073880143e-06,
      "loss": 0.3248,
      "step": 65
    },
    {
      "epoch": 0.4225690276110444,
      "grad_norm": 3.9355652185440637,
      "learning_rate": 1.928114988519039e-06,
      "loss": 0.3093,
      "step": 66
    },
    {
      "epoch": 0.4289715886354542,
      "grad_norm": 4.580138491679913,
      "learning_rate": 1.9238795325112867e-06,
      "loss": 0.3471,
      "step": 67
    },
    {
      "epoch": 0.43537414965986393,
      "grad_norm": 4.264136088165448,
      "learning_rate": 1.9195277725514506e-06,
      "loss": 0.3202,
      "step": 68
    },
    {
      "epoch": 0.4417767106842737,
      "grad_norm": 3.955882298206603,
      "learning_rate": 1.91506025646732e-06,
      "loss": 0.3297,
      "step": 69
    },
    {
      "epoch": 0.4481792717086835,
      "grad_norm": 3.5995979170148513,
      "learning_rate": 1.9104775466588157e-06,
      "loss": 0.2737,
      "step": 70
    },
    {
      "epoch": 0.4545818327330932,
      "grad_norm": 5.25272090784733,
      "learning_rate": 1.905780220027194e-06,
      "loss": 0.2706,
      "step": 71
    },
    {
      "epoch": 0.460984393757503,
      "grad_norm": 4.341561414420844,
      "learning_rate": 1.9009688679024189e-06,
      "loss": 0.3079,
      "step": 72
    },
    {
      "epoch": 0.4673869547819128,
      "grad_norm": 3.6758492701618226,
      "learning_rate": 1.8960440959687252e-06,
      "loss": 0.2924,
      "step": 73
    },
    {
      "epoch": 0.4737895158063225,
      "grad_norm": 3.9451817967595386,
      "learning_rate": 1.8910065241883678e-06,
      "loss": 0.2971,
      "step": 74
    },
    {
      "epoch": 0.4801920768307323,
      "grad_norm": 4.96315413342626,
      "learning_rate": 1.8858567867235798e-06,
      "loss": 0.2673,
      "step": 75
    },
    {
      "epoch": 0.48659463785514206,
      "grad_norm": 4.139530587876671,
      "learning_rate": 1.8805955318567379e-06,
      "loss": 0.27,
      "step": 76
    },
    {
      "epoch": 0.49299719887955185,
      "grad_norm": 4.036729827911165,
      "learning_rate": 1.8752234219087537e-06,
      "loss": 0.3079,
      "step": 77
    },
    {
      "epoch": 0.49939975990396157,
      "grad_norm": 4.267304261790974,
      "learning_rate": 1.8697411331556953e-06,
      "loss": 0.2777,
      "step": 78
    },
    {
      "epoch": 0.5058023209283713,
      "grad_norm": 4.140171639449446,
      "learning_rate": 1.8641493557436548e-06,
      "loss": 0.2965,
      "step": 79
    },
    {
      "epoch": 0.5122048819527811,
      "grad_norm": 4.42365589599126,
      "learning_rate": 1.858448793601866e-06,
      "loss": 0.3452,
      "step": 80
    },
    {
      "epoch": 0.5186074429771909,
      "grad_norm": 4.2607097203243125,
      "learning_rate": 1.852640164354092e-06,
      "loss": 0.2703,
      "step": 81
    },
    {
      "epoch": 0.5250100040016007,
      "grad_norm": 3.9797822473229814,
      "learning_rate": 1.8467241992282841e-06,
      "loss": 0.2994,
      "step": 82
    },
    {
      "epoch": 0.5314125650260104,
      "grad_norm": 3.7553052637845035,
      "learning_rate": 1.8407016429645302e-06,
      "loss": 0.296,
      "step": 83
    },
    {
      "epoch": 0.5378151260504201,
      "grad_norm": 3.7344036316964893,
      "learning_rate": 1.8345732537213026e-06,
      "loss": 0.3052,
      "step": 84
    },
    {
      "epoch": 0.54421768707483,
      "grad_norm": 4.104658693180337,
      "learning_rate": 1.8283398029800164e-06,
      "loss": 0.2794,
      "step": 85
    },
    {
      "epoch": 0.5506202480992397,
      "grad_norm": 3.86258232609451,
      "learning_rate": 1.82200207544791e-06,
      "loss": 0.2729,
      "step": 86
    },
    {
      "epoch": 0.5570228091236494,
      "grad_norm": 3.5583118098950064,
      "learning_rate": 1.8155608689592601e-06,
      "loss": 0.2882,
      "step": 87
    },
    {
      "epoch": 0.5634253701480593,
      "grad_norm": 4.521351652684801,
      "learning_rate": 1.8090169943749474e-06,
      "loss": 0.3041,
      "step": 88
    },
    {
      "epoch": 0.569827931172469,
      "grad_norm": 3.403629696215942,
      "learning_rate": 1.802371275480378e-06,
      "loss": 0.2579,
      "step": 89
    },
    {
      "epoch": 0.5762304921968787,
      "grad_norm": 3.6982154812345396,
      "learning_rate": 1.795624548881781e-06,
      "loss": 0.2464,
      "step": 90
    },
    {
      "epoch": 0.5826330532212886,
      "grad_norm": 4.352958033159996,
      "learning_rate": 1.7887776639008912e-06,
      "loss": 0.2641,
      "step": 91
    },
    {
      "epoch": 0.5890356142456983,
      "grad_norm": 4.54700815993712,
      "learning_rate": 1.7818314824680298e-06,
      "loss": 0.3019,
      "step": 92
    },
    {
      "epoch": 0.595438175270108,
      "grad_norm": 3.789128443890756,
      "learning_rate": 1.774786879013601e-06,
      "loss": 0.2883,
      "step": 93
    },
    {
      "epoch": 0.6018407362945178,
      "grad_norm": 4.340181858310212,
      "learning_rate": 1.767644740358011e-06,
      "loss": 0.3271,
      "step": 94
    },
    {
      "epoch": 0.6082432973189276,
      "grad_norm": 4.181233142586761,
      "learning_rate": 1.760405965600031e-06,
      "loss": 0.3036,
      "step": 95
    },
    {
      "epoch": 0.6146458583433373,
      "grad_norm": 4.468139724866329,
      "learning_rate": 1.753071466003611e-06,
      "loss": 0.2808,
      "step": 96
    },
    {
      "epoch": 0.6210484193677471,
      "grad_norm": 4.510323988503172,
      "learning_rate": 1.7456421648831654e-06,
      "loss": 0.26,
      "step": 97
    },
    {
      "epoch": 0.6274509803921569,
      "grad_norm": 3.8787104679210507,
      "learning_rate": 1.7381189974873407e-06,
      "loss": 0.2888,
      "step": 98
    },
    {
      "epoch": 0.6338535414165666,
      "grad_norm": 4.459830521867393,
      "learning_rate": 1.7305029108812774e-06,
      "loss": 0.2464,
      "step": 99
    },
    {
      "epoch": 0.6402561024409764,
      "grad_norm": 4.709563601825434,
      "learning_rate": 1.7227948638273915e-06,
      "loss": 0.3026,
      "step": 100
    },
    {
      "epoch": 0.6466586634653861,
      "grad_norm": 3.7651337955677255,
      "learning_rate": 1.7149958266646754e-06,
      "loss": 0.2786,
      "step": 101
    },
    {
      "epoch": 0.6530612244897959,
      "grad_norm": 5.5973077659411405,
      "learning_rate": 1.7071067811865474e-06,
      "loss": 0.288,
      "step": 102
    },
    {
      "epoch": 0.6594637855142057,
      "grad_norm": 3.8873700424843127,
      "learning_rate": 1.6991287205172574e-06,
      "loss": 0.2722,
      "step": 103
    },
    {
      "epoch": 0.6658663465386154,
      "grad_norm": 4.2130449731998425,
      "learning_rate": 1.6910626489868648e-06,
      "loss": 0.2955,
      "step": 104
    },
    {
      "epoch": 0.6722689075630253,
      "grad_norm": 4.14817042922031,
      "learning_rate": 1.682909582004807e-06,
      "loss": 0.3185,
      "step": 105
    },
    {
      "epoch": 0.678671468587435,
      "grad_norm": 4.776345876515052,
      "learning_rate": 1.6746705459320744e-06,
      "loss": 0.3116,
      "step": 106
    },
    {
      "epoch": 0.6850740296118447,
      "grad_norm": 3.857374024007739,
      "learning_rate": 1.6663465779520037e-06,
      "loss": 0.3189,
      "step": 107
    },
    {
      "epoch": 0.6914765906362546,
      "grad_norm": 4.133483454926766,
      "learning_rate": 1.6579387259397126e-06,
      "loss": 0.3082,
      "step": 108
    },
    {
      "epoch": 0.6978791516606643,
      "grad_norm": 4.020769377918411,
      "learning_rate": 1.6494480483301835e-06,
      "loss": 0.2825,
      "step": 109
    },
    {
      "epoch": 0.704281712685074,
      "grad_norm": 4.261022360359079,
      "learning_rate": 1.640875613985024e-06,
      "loss": 0.2907,
      "step": 110
    },
    {
      "epoch": 0.7106842737094838,
      "grad_norm": 3.712351214158889,
      "learning_rate": 1.6322225020579096e-06,
      "loss": 0.2822,
      "step": 111
    },
    {
      "epoch": 0.7170868347338936,
      "grad_norm": 3.440740014931854,
      "learning_rate": 1.6234898018587336e-06,
      "loss": 0.2381,
      "step": 112
    },
    {
      "epoch": 0.7234893957583033,
      "grad_norm": 3.775626420708748,
      "learning_rate": 1.6146786127164771e-06,
      "loss": 0.2937,
      "step": 113
    },
    {
      "epoch": 0.7298919567827131,
      "grad_norm": 3.8211915298740666,
      "learning_rate": 1.6057900438408199e-06,
      "loss": 0.2843,
      "step": 114
    },
    {
      "epoch": 0.7362945178071229,
      "grad_norm": 3.752355539307679,
      "learning_rate": 1.5968252141825035e-06,
      "loss": 0.2648,
      "step": 115
    },
    {
      "epoch": 0.7426970788315326,
      "grad_norm": 3.842629249054385,
      "learning_rate": 1.587785252292473e-06,
      "loss": 0.27,
      "step": 116
    },
    {
      "epoch": 0.7490996398559424,
      "grad_norm": 4.166939902099283,
      "learning_rate": 1.578671296179806e-06,
      "loss": 0.2521,
      "step": 117
    },
    {
      "epoch": 0.7555022008803521,
      "grad_norm": 3.8837453592291844,
      "learning_rate": 1.569484493168452e-06,
      "loss": 0.2926,
      "step": 118
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 3.9849389096374837,
      "learning_rate": 1.5602259997528027e-06,
      "loss": 0.2415,
      "step": 119
    },
    {
      "epoch": 0.7683073229291717,
      "grad_norm": 3.8214588045054407,
      "learning_rate": 1.5508969814521024e-06,
      "loss": 0.2896,
      "step": 120
    },
    {
      "epoch": 0.7747098839535814,
      "grad_norm": 3.596004790144505,
      "learning_rate": 1.5414986126637257e-06,
      "loss": 0.25,
      "step": 121
    },
    {
      "epoch": 0.7811124449779911,
      "grad_norm": 3.7868659894236734,
      "learning_rate": 1.5320320765153365e-06,
      "loss": 0.2742,
      "step": 122
    },
    {
      "epoch": 0.787515006002401,
      "grad_norm": 3.605103510806505,
      "learning_rate": 1.5224985647159488e-06,
      "loss": 0.2288,
      "step": 123
    },
    {
      "epoch": 0.7939175670268107,
      "grad_norm": 4.164526361851514,
      "learning_rate": 1.5128992774059062e-06,
      "loss": 0.257,
      "step": 124
    },
    {
      "epoch": 0.8003201280512204,
      "grad_norm": 4.2601493772240095,
      "learning_rate": 1.5032354230058002e-06,
      "loss": 0.2703,
      "step": 125
    },
    {
      "epoch": 0.8067226890756303,
      "grad_norm": 4.230695549396448,
      "learning_rate": 1.4935082180643467e-06,
      "loss": 0.315,
      "step": 126
    },
    {
      "epoch": 0.81312525010004,
      "grad_norm": 4.029734026287984,
      "learning_rate": 1.4837188871052397e-06,
      "loss": 0.2587,
      "step": 127
    },
    {
      "epoch": 0.8195278111244498,
      "grad_norm": 4.418674847039931,
      "learning_rate": 1.4738686624729987e-06,
      "loss": 0.3162,
      "step": 128
    },
    {
      "epoch": 0.8259303721488596,
      "grad_norm": 4.0453350041861205,
      "learning_rate": 1.463958784177834e-06,
      "loss": 0.2534,
      "step": 129
    },
    {
      "epoch": 0.8323329331732693,
      "grad_norm": 4.6877576825765335,
      "learning_rate": 1.4539904997395467e-06,
      "loss": 0.2631,
      "step": 130
    },
    {
      "epoch": 0.8387354941976791,
      "grad_norm": 3.780029409396397,
      "learning_rate": 1.4439650640304821e-06,
      "loss": 0.271,
      "step": 131
    },
    {
      "epoch": 0.8451380552220888,
      "grad_norm": 3.6785546417231756,
      "learning_rate": 1.433883739117558e-06,
      "loss": 0.2418,
      "step": 132
    },
    {
      "epoch": 0.8515406162464986,
      "grad_norm": 3.396354715623399,
      "learning_rate": 1.4237477941033886e-06,
      "loss": 0.2499,
      "step": 133
    },
    {
      "epoch": 0.8579431772709084,
      "grad_norm": 3.7280690326153265,
      "learning_rate": 1.4135585049665206e-06,
      "loss": 0.2846,
      "step": 134
    },
    {
      "epoch": 0.8643457382953181,
      "grad_norm": 4.004231717394044,
      "learning_rate": 1.4033171544008051e-06,
      "loss": 0.2938,
      "step": 135
    },
    {
      "epoch": 0.8707482993197279,
      "grad_norm": 3.342930768359313,
      "learning_rate": 1.3930250316539235e-06,
      "loss": 0.2693,
      "step": 136
    },
    {
      "epoch": 0.8771508603441377,
      "grad_norm": 3.9327560661895244,
      "learning_rate": 1.3826834323650898e-06,
      "loss": 0.3064,
      "step": 137
    },
    {
      "epoch": 0.8835534213685474,
      "grad_norm": 3.8159537564044954,
      "learning_rate": 1.3722936584019451e-06,
      "loss": 0.2529,
      "step": 138
    },
    {
      "epoch": 0.8899559823929571,
      "grad_norm": 3.453933647883322,
      "learning_rate": 1.3618570176966722e-06,
      "loss": 0.2837,
      "step": 139
    },
    {
      "epoch": 0.896358543417367,
      "grad_norm": 3.7776942360972035,
      "learning_rate": 1.3513748240813427e-06,
      "loss": 0.3414,
      "step": 140
    },
    {
      "epoch": 0.9027611044417767,
      "grad_norm": 3.701115924801193,
      "learning_rate": 1.3408483971225249e-06,
      "loss": 0.2845,
      "step": 141
    },
    {
      "epoch": 0.9091636654661864,
      "grad_norm": 3.642984026753104,
      "learning_rate": 1.3302790619551672e-06,
      "loss": 0.2472,
      "step": 142
    },
    {
      "epoch": 0.9155662264905963,
      "grad_norm": 3.875796995615745,
      "learning_rate": 1.3196681491157816e-06,
      "loss": 0.2481,
      "step": 143
    },
    {
      "epoch": 0.921968787515006,
      "grad_norm": 3.8424117933387256,
      "learning_rate": 1.3090169943749473e-06,
      "loss": 0.2819,
      "step": 144
    },
    {
      "epoch": 0.9283713485394157,
      "grad_norm": 4.037093475071882,
      "learning_rate": 1.298326938569156e-06,
      "loss": 0.324,
      "step": 145
    },
    {
      "epoch": 0.9347739095638256,
      "grad_norm": 3.5202146156430447,
      "learning_rate": 1.2875993274320173e-06,
      "loss": 0.2459,
      "step": 146
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 3.740795759086208,
      "learning_rate": 1.2768355114248492e-06,
      "loss": 0.2443,
      "step": 147
    },
    {
      "epoch": 0.947579031612645,
      "grad_norm": 4.3061782207302866,
      "learning_rate": 1.266036845566675e-06,
      "loss": 0.3083,
      "step": 148
    },
    {
      "epoch": 0.9539815926370548,
      "grad_norm": 3.9442811474090775,
      "learning_rate": 1.2552046892636426e-06,
      "loss": 0.244,
      "step": 149
    },
    {
      "epoch": 0.9603841536614646,
      "grad_norm": 3.9865071324515027,
      "learning_rate": 1.244340406137894e-06,
      "loss": 0.2277,
      "step": 150
    },
    {
      "epoch": 0.9667867146858744,
      "grad_norm": 4.016047319687184,
      "learning_rate": 1.2334453638559054e-06,
      "loss": 0.2541,
      "step": 151
    },
    {
      "epoch": 0.9731892757102841,
      "grad_norm": 4.032894518578795,
      "learning_rate": 1.2225209339563143e-06,
      "loss": 0.2575,
      "step": 152
    },
    {
      "epoch": 0.9795918367346939,
      "grad_norm": 4.065384007721365,
      "learning_rate": 1.211568491677263e-06,
      "loss": 0.2638,
      "step": 153
    },
    {
      "epoch": 0.9859943977591037,
      "grad_norm": 4.082382918746455,
      "learning_rate": 1.2005894157832728e-06,
      "loss": 0.3124,
      "step": 154
    },
    {
      "epoch": 0.9923969587835134,
      "grad_norm": 4.690622898413357,
      "learning_rate": 1.1895850883916785e-06,
      "loss": 0.2434,
      "step": 155
    },
    {
      "epoch": 0.9987995198079231,
      "grad_norm": 4.05650193960916,
      "learning_rate": 1.1785568947986366e-06,
      "loss": 0.3047,
      "step": 156
    },
    {
      "epoch": 1.005202080832333,
      "grad_norm": 3.2193287198369993,
      "learning_rate": 1.1675062233047363e-06,
      "loss": 0.22,
      "step": 157
    },
    {
      "epoch": 1.0116046418567426,
      "grad_norm": 2.8522266075564024,
      "learning_rate": 1.156434465040231e-06,
      "loss": 0.2104,
      "step": 158
    },
    {
      "epoch": 1.0180072028811524,
      "grad_norm": 2.960293382016092,
      "learning_rate": 1.1453430137899128e-06,
      "loss": 0.224,
      "step": 159
    },
    {
      "epoch": 1.0244097639055623,
      "grad_norm": 3.1247052252152434,
      "learning_rate": 1.1342332658176555e-06,
      "loss": 0.2318,
      "step": 160
    },
    {
      "epoch": 1.0308123249299719,
      "grad_norm": 2.7511164775931163,
      "learning_rate": 1.123106619690643e-06,
      "loss": 0.2045,
      "step": 161
    },
    {
      "epoch": 1.0372148859543817,
      "grad_norm": 2.7289267777021027,
      "learning_rate": 1.1119644761033077e-06,
      "loss": 0.2038,
      "step": 162
    },
    {
      "epoch": 1.0436174469787916,
      "grad_norm": 2.713995483150959,
      "learning_rate": 1.1008082377010045e-06,
      "loss": 0.2129,
      "step": 163
    },
    {
      "epoch": 1.0500200080032012,
      "grad_norm": 2.78973076664929,
      "learning_rate": 1.0896393089034335e-06,
      "loss": 0.2182,
      "step": 164
    },
    {
      "epoch": 1.056422569027611,
      "grad_norm": 2.9643061675150038,
      "learning_rate": 1.078459095727845e-06,
      "loss": 0.2232,
      "step": 165
    },
    {
      "epoch": 1.0628251300520208,
      "grad_norm": 2.8380963314510113,
      "learning_rate": 1.0672690056120398e-06,
      "loss": 0.2227,
      "step": 166
    },
    {
      "epoch": 1.0692276910764307,
      "grad_norm": 2.687954696836688,
      "learning_rate": 1.0560704472371917e-06,
      "loss": 0.2114,
      "step": 167
    },
    {
      "epoch": 1.0756302521008403,
      "grad_norm": 2.9271042366476996,
      "learning_rate": 1.044864830350515e-06,
      "loss": 0.2059,
      "step": 168
    },
    {
      "epoch": 1.0820328131252501,
      "grad_norm": 3.0180594470719897,
      "learning_rate": 1.033653565587794e-06,
      "loss": 0.2217,
      "step": 169
    },
    {
      "epoch": 1.08843537414966,
      "grad_norm": 3.1673109290531154,
      "learning_rate": 1.022438064295805e-06,
      "loss": 0.2155,
      "step": 170
    },
    {
      "epoch": 1.0948379351740696,
      "grad_norm": 3.0766952480651164,
      "learning_rate": 1.0112197383546459e-06,
      "loss": 0.217,
      "step": 171
    },
    {
      "epoch": 1.1012404961984794,
      "grad_norm": 2.719327293458082,
      "learning_rate": 1e-06,
      "loss": 0.1822,
      "step": 172
    },
    {
      "epoch": 1.1076430572228892,
      "grad_norm": 2.770227862802081,
      "learning_rate": 9.88780261645354e-07,
      "loss": 0.1926,
      "step": 173
    },
    {
      "epoch": 1.1140456182472989,
      "grad_norm": 3.0009223093685704,
      "learning_rate": 9.77561935704195e-07,
      "loss": 0.1911,
      "step": 174
    },
    {
      "epoch": 1.1204481792717087,
      "grad_norm": 3.023185799043678,
      "learning_rate": 9.663464344122063e-07,
      "loss": 0.1903,
      "step": 175
    },
    {
      "epoch": 1.1268507402961185,
      "grad_norm": 3.457998950657755,
      "learning_rate": 9.551351696494853e-07,
      "loss": 0.251,
      "step": 176
    },
    {
      "epoch": 1.1332533013205282,
      "grad_norm": 2.9312099897929036,
      "learning_rate": 9.43929552762808e-07,
      "loss": 0.2176,
      "step": 177
    },
    {
      "epoch": 1.139655862344938,
      "grad_norm": 3.3046823347401175,
      "learning_rate": 9.327309943879603e-07,
      "loss": 0.2109,
      "step": 178
    },
    {
      "epoch": 1.1460584233693478,
      "grad_norm": 3.0921904822263078,
      "learning_rate": 9.215409042721551e-07,
      "loss": 0.22,
      "step": 179
    },
    {
      "epoch": 1.1524609843937574,
      "grad_norm": 2.991127231725485,
      "learning_rate": 9.103606910965665e-07,
      "loss": 0.188,
      "step": 180
    },
    {
      "epoch": 1.1588635454181673,
      "grad_norm": 3.281303486352065,
      "learning_rate": 8.991917622989955e-07,
      "loss": 0.2051,
      "step": 181
    },
    {
      "epoch": 1.165266106442577,
      "grad_norm": 3.2779897667339393,
      "learning_rate": 8.880355238966921e-07,
      "loss": 0.2128,
      "step": 182
    },
    {
      "epoch": 1.1716686674669867,
      "grad_norm": 2.908646079818215,
      "learning_rate": 8.768933803093572e-07,
      "loss": 0.211,
      "step": 183
    },
    {
      "epoch": 1.1780712284913966,
      "grad_norm": 3.511327611579083,
      "learning_rate": 8.657667341823448e-07,
      "loss": 0.2067,
      "step": 184
    },
    {
      "epoch": 1.1844737895158064,
      "grad_norm": 3.446334131319975,
      "learning_rate": 8.546569862100875e-07,
      "loss": 0.2126,
      "step": 185
    },
    {
      "epoch": 1.190876350540216,
      "grad_norm": 3.086417796365709,
      "learning_rate": 8.435655349597689e-07,
      "loss": 0.1961,
      "step": 186
    },
    {
      "epoch": 1.1972789115646258,
      "grad_norm": 3.486051214472594,
      "learning_rate": 8.324937766952636e-07,
      "loss": 0.1954,
      "step": 187
    },
    {
      "epoch": 1.2036814725890357,
      "grad_norm": 3.4725632158887625,
      "learning_rate": 8.214431052013634e-07,
      "loss": 0.2207,
      "step": 188
    },
    {
      "epoch": 1.2100840336134453,
      "grad_norm": 3.102428873715714,
      "learning_rate": 8.104149116083216e-07,
      "loss": 0.2167,
      "step": 189
    },
    {
      "epoch": 1.2164865946378551,
      "grad_norm": 3.2423731064102457,
      "learning_rate": 7.994105842167272e-07,
      "loss": 0.1924,
      "step": 190
    },
    {
      "epoch": 1.222889155662265,
      "grad_norm": 3.2545985252021152,
      "learning_rate": 7.884315083227372e-07,
      "loss": 0.2029,
      "step": 191
    },
    {
      "epoch": 1.2292917166866746,
      "grad_norm": 3.0971604028938517,
      "learning_rate": 7.774790660436857e-07,
      "loss": 0.1741,
      "step": 192
    },
    {
      "epoch": 1.2356942777110844,
      "grad_norm": 3.443986175823104,
      "learning_rate": 7.665546361440949e-07,
      "loss": 0.2451,
      "step": 193
    },
    {
      "epoch": 1.2420968387354943,
      "grad_norm": 3.3511781482482847,
      "learning_rate": 7.556595938621058e-07,
      "loss": 0.1843,
      "step": 194
    },
    {
      "epoch": 1.2484993997599039,
      "grad_norm": 3.4059727380620686,
      "learning_rate": 7.447953107363574e-07,
      "loss": 0.2354,
      "step": 195
    },
    {
      "epoch": 1.2549019607843137,
      "grad_norm": 3.2416330111232243,
      "learning_rate": 7.33963154433325e-07,
      "loss": 0.1955,
      "step": 196
    },
    {
      "epoch": 1.2613045218087235,
      "grad_norm": 4.2347778935869345,
      "learning_rate": 7.231644885751507e-07,
      "loss": 0.217,
      "step": 197
    },
    {
      "epoch": 1.2677070828331334,
      "grad_norm": 3.567306963297163,
      "learning_rate": 7.124006725679828e-07,
      "loss": 0.2212,
      "step": 198
    },
    {
      "epoch": 1.274109643857543,
      "grad_norm": 3.00005386027494,
      "learning_rate": 7.016730614308439e-07,
      "loss": 0.1926,
      "step": 199
    },
    {
      "epoch": 1.2805122048819528,
      "grad_norm": 3.414969994137627,
      "learning_rate": 6.909830056250526e-07,
      "loss": 0.2044,
      "step": 200
    },
    {
      "epoch": 1.2869147659063627,
      "grad_norm": 3.5101423542049868,
      "learning_rate": 6.803318508842186e-07,
      "loss": 0.1958,
      "step": 201
    },
    {
      "epoch": 1.2933173269307723,
      "grad_norm": 3.08086263335831,
      "learning_rate": 6.697209380448332e-07,
      "loss": 0.2239,
      "step": 202
    },
    {
      "epoch": 1.2997198879551821,
      "grad_norm": 3.6991398316157835,
      "learning_rate": 6.59151602877475e-07,
      "loss": 0.2019,
      "step": 203
    },
    {
      "epoch": 1.306122448979592,
      "grad_norm": 3.1898949896135953,
      "learning_rate": 6.486251759186572e-07,
      "loss": 0.2276,
      "step": 204
    },
    {
      "epoch": 1.3125250100040016,
      "grad_norm": 3.15209044029482,
      "learning_rate": 6.381429823033279e-07,
      "loss": 0.1788,
      "step": 205
    },
    {
      "epoch": 1.3189275710284114,
      "grad_norm": 3.2074734897831756,
      "learning_rate": 6.277063415980548e-07,
      "loss": 0.2434,
      "step": 206
    },
    {
      "epoch": 1.3253301320528212,
      "grad_norm": 2.875848443017429,
      "learning_rate": 6.173165676349102e-07,
      "loss": 0.1809,
      "step": 207
    },
    {
      "epoch": 1.3317326930772309,
      "grad_norm": 3.273739860976,
      "learning_rate": 6.069749683460764e-07,
      "loss": 0.2,
      "step": 208
    },
    {
      "epoch": 1.3381352541016407,
      "grad_norm": 3.4156670470458494,
      "learning_rate": 5.96682845599195e-07,
      "loss": 0.2044,
      "step": 209
    },
    {
      "epoch": 1.3445378151260505,
      "grad_norm": 4.24358157090511,
      "learning_rate": 5.864414950334795e-07,
      "loss": 0.194,
      "step": 210
    },
    {
      "epoch": 1.3509403761504601,
      "grad_norm": 3.116102100075619,
      "learning_rate": 5.762522058966113e-07,
      "loss": 0.2089,
      "step": 211
    },
    {
      "epoch": 1.35734293717487,
      "grad_norm": 2.8625746825319034,
      "learning_rate": 5.661162608824419e-07,
      "loss": 0.2083,
      "step": 212
    },
    {
      "epoch": 1.3637454981992798,
      "grad_norm": 3.2057903093169555,
      "learning_rate": 5.56034935969518e-07,
      "loss": 0.1952,
      "step": 213
    },
    {
      "epoch": 1.3701480592236894,
      "grad_norm": 2.669239875130929,
      "learning_rate": 5.460095002604532e-07,
      "loss": 0.2056,
      "step": 214
    },
    {
      "epoch": 1.3765506202480993,
      "grad_norm": 3.2441222068201605,
      "learning_rate": 5.36041215822166e-07,
      "loss": 0.209,
      "step": 215
    },
    {
      "epoch": 1.382953181272509,
      "grad_norm": 3.0334649559489724,
      "learning_rate": 5.261313375270013e-07,
      "loss": 0.1821,
      "step": 216
    },
    {
      "epoch": 1.3893557422969187,
      "grad_norm": 3.005021545247239,
      "learning_rate": 5.162811128947602e-07,
      "loss": 0.1919,
      "step": 217
    },
    {
      "epoch": 1.3957583033213286,
      "grad_norm": 3.0185970628882184,
      "learning_rate": 5.064917819356531e-07,
      "loss": 0.2124,
      "step": 218
    },
    {
      "epoch": 1.4021608643457384,
      "grad_norm": 2.9841034532353645,
      "learning_rate": 4.967645769941999e-07,
      "loss": 0.1751,
      "step": 219
    },
    {
      "epoch": 1.408563425370148,
      "grad_norm": 3.1214597510428783,
      "learning_rate": 4.871007225940939e-07,
      "loss": 0.2245,
      "step": 220
    },
    {
      "epoch": 1.4149659863945578,
      "grad_norm": 3.118933296947675,
      "learning_rate": 4.775014352840512e-07,
      "loss": 0.2171,
      "step": 221
    },
    {
      "epoch": 1.4213685474189677,
      "grad_norm": 3.1742771966166567,
      "learning_rate": 4.6796792348466353e-07,
      "loss": 0.2132,
      "step": 222
    },
    {
      "epoch": 1.4277711084433773,
      "grad_norm": 2.602593575781098,
      "learning_rate": 4.585013873362743e-07,
      "loss": 0.1799,
      "step": 223
    },
    {
      "epoch": 1.4341736694677871,
      "grad_norm": 2.94290229867875,
      "learning_rate": 4.4910301854789755e-07,
      "loss": 0.1815,
      "step": 224
    },
    {
      "epoch": 1.440576230492197,
      "grad_norm": 3.2300005719414204,
      "learning_rate": 4.397740002471972e-07,
      "loss": 0.2143,
      "step": 225
    },
    {
      "epoch": 1.4469787915166066,
      "grad_norm": 2.8425860620393943,
      "learning_rate": 4.3051550683154804e-07,
      "loss": 0.2251,
      "step": 226
    },
    {
      "epoch": 1.4533813525410164,
      "grad_norm": 2.7744967195870136,
      "learning_rate": 4.2132870382019427e-07,
      "loss": 0.2023,
      "step": 227
    },
    {
      "epoch": 1.4597839135654262,
      "grad_norm": 2.8299937805822384,
      "learning_rate": 4.1221474770752696e-07,
      "loss": 0.1825,
      "step": 228
    },
    {
      "epoch": 1.4661864745898359,
      "grad_norm": 2.842358054849413,
      "learning_rate": 4.031747858174964e-07,
      "loss": 0.1815,
      "step": 229
    },
    {
      "epoch": 1.4725890356142457,
      "grad_norm": 3.0046695988860064,
      "learning_rate": 3.942099561591802e-07,
      "loss": 0.1876,
      "step": 230
    },
    {
      "epoch": 1.4789915966386555,
      "grad_norm": 3.049335410632264,
      "learning_rate": 3.853213872835228e-07,
      "loss": 0.1965,
      "step": 231
    },
    {
      "epoch": 1.4853941576630652,
      "grad_norm": 2.9908629914418663,
      "learning_rate": 3.765101981412665e-07,
      "loss": 0.206,
      "step": 232
    },
    {
      "epoch": 1.491796718687475,
      "grad_norm": 3.6142447406768,
      "learning_rate": 3.677774979420903e-07,
      "loss": 0.1962,
      "step": 233
    },
    {
      "epoch": 1.4981992797118848,
      "grad_norm": 3.1083828790336265,
      "learning_rate": 3.5912438601497584e-07,
      "loss": 0.1907,
      "step": 234
    },
    {
      "epoch": 1.5046018407362944,
      "grad_norm": 3.8490048708212017,
      "learning_rate": 3.5055195166981646e-07,
      "loss": 0.2483,
      "step": 235
    },
    {
      "epoch": 1.5110044017607043,
      "grad_norm": 3.5135746675092574,
      "learning_rate": 3.420612740602874e-07,
      "loss": 0.2197,
      "step": 236
    },
    {
      "epoch": 1.517406962785114,
      "grad_norm": 2.9605758005174083,
      "learning_rate": 3.3365342204799606e-07,
      "loss": 0.2048,
      "step": 237
    },
    {
      "epoch": 1.5238095238095237,
      "grad_norm": 3.102724567495504,
      "learning_rate": 3.253294540679257e-07,
      "loss": 0.2163,
      "step": 238
    },
    {
      "epoch": 1.5302120848339336,
      "grad_norm": 3.338826460184625,
      "learning_rate": 3.170904179951931e-07,
      "loss": 0.1892,
      "step": 239
    },
    {
      "epoch": 1.5366146458583434,
      "grad_norm": 3.234763671406949,
      "learning_rate": 3.0893735101313535e-07,
      "loss": 0.2762,
      "step": 240
    },
    {
      "epoch": 1.543017206882753,
      "grad_norm": 2.812979283705818,
      "learning_rate": 3.008712794827426e-07,
      "loss": 0.1717,
      "step": 241
    },
    {
      "epoch": 1.5494197679071628,
      "grad_norm": 2.934584165405633,
      "learning_rate": 2.9289321881345254e-07,
      "loss": 0.2,
      "step": 242
    },
    {
      "epoch": 1.5558223289315727,
      "grad_norm": 3.017927185569003,
      "learning_rate": 2.850041733353247e-07,
      "loss": 0.1887,
      "step": 243
    },
    {
      "epoch": 1.5622248899559823,
      "grad_norm": 2.821836163783163,
      "learning_rate": 2.7720513617260855e-07,
      "loss": 0.1685,
      "step": 244
    },
    {
      "epoch": 1.5686274509803921,
      "grad_norm": 2.9956132520307692,
      "learning_rate": 2.6949708911872247e-07,
      "loss": 0.1944,
      "step": 245
    },
    {
      "epoch": 1.575030012004802,
      "grad_norm": 3.012268102569264,
      "learning_rate": 2.6188100251265943e-07,
      "loss": 0.2168,
      "step": 246
    },
    {
      "epoch": 1.5814325730292116,
      "grad_norm": 2.6551981273651646,
      "learning_rate": 2.543578351168344e-07,
      "loss": 0.1587,
      "step": 247
    },
    {
      "epoch": 1.5878351340536214,
      "grad_norm": 2.6439156715272913,
      "learning_rate": 2.4692853399638913e-07,
      "loss": 0.1889,
      "step": 248
    },
    {
      "epoch": 1.5942376950780313,
      "grad_norm": 2.9350723750826266,
      "learning_rate": 2.395940343999691e-07,
      "loss": 0.2073,
      "step": 249
    },
    {
      "epoch": 1.6006402561024409,
      "grad_norm": 3.0827120888341013,
      "learning_rate": 2.3235525964198888e-07,
      "loss": 0.1895,
      "step": 250
    },
    {
      "epoch": 1.6070428171268507,
      "grad_norm": 3.1686080831269483,
      "learning_rate": 2.252131209863991e-07,
      "loss": 0.2029,
      "step": 251
    },
    {
      "epoch": 1.6134453781512605,
      "grad_norm": 2.851390750231077,
      "learning_rate": 2.181685175319702e-07,
      "loss": 0.2061,
      "step": 252
    },
    {
      "epoch": 1.6198479391756702,
      "grad_norm": 3.094325341444059,
      "learning_rate": 2.11222336099109e-07,
      "loss": 0.192,
      "step": 253
    },
    {
      "epoch": 1.62625050020008,
      "grad_norm": 3.1464174144330026,
      "learning_rate": 2.043754511182191e-07,
      "loss": 0.2103,
      "step": 254
    },
    {
      "epoch": 1.6326530612244898,
      "grad_norm": 3.120376340739763,
      "learning_rate": 1.9762872451962208e-07,
      "loss": 0.1705,
      "step": 255
    },
    {
      "epoch": 1.6390556222488994,
      "grad_norm": 3.2179511708389037,
      "learning_rate": 1.9098300562505264e-07,
      "loss": 0.1948,
      "step": 256
    },
    {
      "epoch": 1.6454581832733093,
      "grad_norm": 2.8898679416961692,
      "learning_rate": 1.8443913104073982e-07,
      "loss": 0.1823,
      "step": 257
    },
    {
      "epoch": 1.6518607442977191,
      "grad_norm": 2.993303933297039,
      "learning_rate": 1.7799792455209016e-07,
      "loss": 0.2126,
      "step": 258
    },
    {
      "epoch": 1.6582633053221287,
      "grad_norm": 3.2103383308153925,
      "learning_rate": 1.716601970199836e-07,
      "loss": 0.1961,
      "step": 259
    },
    {
      "epoch": 1.6646658663465386,
      "grad_norm": 3.235948737487849,
      "learning_rate": 1.6542674627869734e-07,
      "loss": 0.2057,
      "step": 260
    },
    {
      "epoch": 1.6710684273709484,
      "grad_norm": 3.103836779762489,
      "learning_rate": 1.592983570354699e-07,
      "loss": 0.2186,
      "step": 261
    },
    {
      "epoch": 1.677470988395358,
      "grad_norm": 3.014947796395499,
      "learning_rate": 1.5327580077171588e-07,
      "loss": 0.1912,
      "step": 262
    },
    {
      "epoch": 1.6838735494197679,
      "grad_norm": 3.2700341051751245,
      "learning_rate": 1.473598356459078e-07,
      "loss": 0.2456,
      "step": 263
    },
    {
      "epoch": 1.6902761104441777,
      "grad_norm": 3.2091848048221308,
      "learning_rate": 1.415512063981339e-07,
      "loss": 0.2099,
      "step": 264
    },
    {
      "epoch": 1.6966786714685873,
      "grad_norm": 2.750474336934699,
      "learning_rate": 1.358506442563454e-07,
      "loss": 0.2159,
      "step": 265
    },
    {
      "epoch": 1.7030812324929971,
      "grad_norm": 4.614673424699257,
      "learning_rate": 1.3025886684430465e-07,
      "loss": 0.2067,
      "step": 266
    },
    {
      "epoch": 1.709483793517407,
      "grad_norm": 2.9058686083863514,
      "learning_rate": 1.2477657809124632e-07,
      "loss": 0.1992,
      "step": 267
    },
    {
      "epoch": 1.7158863545418166,
      "grad_norm": 2.862146073630623,
      "learning_rate": 1.19404468143262e-07,
      "loss": 0.2142,
      "step": 268
    },
    {
      "epoch": 1.7222889155662267,
      "grad_norm": 3.287005616915961,
      "learning_rate": 1.1414321327642019e-07,
      "loss": 0.2051,
      "step": 269
    },
    {
      "epoch": 1.7286914765906363,
      "grad_norm": 2.9357679152099383,
      "learning_rate": 1.089934758116322e-07,
      "loss": 0.2113,
      "step": 270
    },
    {
      "epoch": 1.7350940376150459,
      "grad_norm": 2.9564341258962683,
      "learning_rate": 1.0395590403127486e-07,
      "loss": 0.1711,
      "step": 271
    },
    {
      "epoch": 1.741496598639456,
      "grad_norm": 3.1982558699085484,
      "learning_rate": 9.903113209758096e-08,
      "loss": 0.2086,
      "step": 272
    },
    {
      "epoch": 1.7478991596638656,
      "grad_norm": 2.9531309947941566,
      "learning_rate": 9.421977997280594e-08,
      "loss": 0.1959,
      "step": 273
    },
    {
      "epoch": 1.7543017206882752,
      "grad_norm": 3.070946141827486,
      "learning_rate": 8.952245334118413e-08,
      "loss": 0.1975,
      "step": 274
    },
    {
      "epoch": 1.7607042817126852,
      "grad_norm": 2.758150815701359,
      "learning_rate": 8.493974353268019e-08,
      "loss": 0.1922,
      "step": 275
    },
    {
      "epoch": 1.7671068427370948,
      "grad_norm": 2.9185225875350858,
      "learning_rate": 8.047222744854942e-08,
      "loss": 0.1825,
      "step": 276
    },
    {
      "epoch": 1.7735094037615045,
      "grad_norm": 3.110148299633103,
      "learning_rate": 7.612046748871326e-08,
      "loss": 0.2096,
      "step": 277
    },
    {
      "epoch": 1.7799119647859145,
      "grad_norm": 3.1130566100683246,
      "learning_rate": 7.188501148096116e-08,
      "loss": 0.1733,
      "step": 278
    },
    {
      "epoch": 1.7863145258103241,
      "grad_norm": 3.1415180042361444,
      "learning_rate": 6.77663926119858e-08,
      "loss": 0.2043,
      "step": 279
    },
    {
      "epoch": 1.7927170868347337,
      "grad_norm": 3.3828521625639554,
      "learning_rate": 6.376512936026279e-08,
      "loss": 0.2248,
      "step": 280
    },
    {
      "epoch": 1.7991196478591438,
      "grad_norm": 3.0831187019400916,
      "learning_rate": 5.988172543078096e-08,
      "loss": 0.2238,
      "step": 281
    },
    {
      "epoch": 1.8055222088835534,
      "grad_norm": 3.583225680392213,
      "learning_rate": 5.611666969163242e-08,
      "loss": 0.2025,
      "step": 282
    },
    {
      "epoch": 1.811924769907963,
      "grad_norm": 2.7749576641359055,
      "learning_rate": 5.2470436112471264e-08,
      "loss": 0.1918,
      "step": 283
    },
    {
      "epoch": 1.818327330932373,
      "grad_norm": 3.491814950661689,
      "learning_rate": 4.8943483704846465e-08,
      "loss": 0.2017,
      "step": 284
    },
    {
      "epoch": 1.8247298919567827,
      "grad_norm": 3.3544823516715656,
      "learning_rate": 4.553625646441928e-08,
      "loss": 0.2182,
      "step": 285
    },
    {
      "epoch": 1.8311324529811923,
      "grad_norm": 3.03556242043361,
      "learning_rate": 4.224918331506955e-08,
      "loss": 0.2172,
      "step": 286
    },
    {
      "epoch": 1.8375350140056024,
      "grad_norm": 3.1512946254390815,
      "learning_rate": 3.908267805490051e-08,
      "loss": 0.2375,
      "step": 287
    },
    {
      "epoch": 1.843937575030012,
      "grad_norm": 3.1583349292530687,
      "learning_rate": 3.6037139304146756e-08,
      "loss": 0.2164,
      "step": 288
    },
    {
      "epoch": 1.8503401360544216,
      "grad_norm": 2.6885025755750007,
      "learning_rate": 3.3112950454993625e-08,
      "loss": 0.2124,
      "step": 289
    },
    {
      "epoch": 1.8567426970788317,
      "grad_norm": 2.899693820698532,
      "learning_rate": 3.0310479623313125e-08,
      "loss": 0.2094,
      "step": 290
    },
    {
      "epoch": 1.8631452581032413,
      "grad_norm": 2.9718624867551746,
      "learning_rate": 2.7630079602323443e-08,
      "loss": 0.1863,
      "step": 291
    },
    {
      "epoch": 1.8695478191276511,
      "grad_norm": 3.1641777461916396,
      "learning_rate": 2.507208781817638e-08,
      "loss": 0.253,
      "step": 292
    },
    {
      "epoch": 1.875950380152061,
      "grad_norm": 2.979215203416192,
      "learning_rate": 2.263682628748087e-08,
      "loss": 0.218,
      "step": 293
    },
    {
      "epoch": 1.8823529411764706,
      "grad_norm": 2.935051364990218,
      "learning_rate": 2.032460157676452e-08,
      "loss": 0.1981,
      "step": 294
    },
    {
      "epoch": 1.8887555022008804,
      "grad_norm": 2.818283326905153,
      "learning_rate": 1.8135704763881598e-08,
      "loss": 0.2028,
      "step": 295
    },
    {
      "epoch": 1.8951580632252902,
      "grad_norm": 3.2868954384023823,
      "learning_rate": 1.607041140137033e-08,
      "loss": 0.2051,
      "step": 296
    },
    {
      "epoch": 1.9015606242496998,
      "grad_norm": 2.948811609921979,
      "learning_rate": 1.4128981481764113e-08,
      "loss": 0.2096,
      "step": 297
    },
    {
      "epoch": 1.9079631852741097,
      "grad_norm": 2.6648158990098314,
      "learning_rate": 1.231165940486234e-08,
      "loss": 0.1701,
      "step": 298
    },
    {
      "epoch": 1.9143657462985195,
      "grad_norm": 2.828105773518058,
      "learning_rate": 1.0618673946963364e-08,
      "loss": 0.2012,
      "step": 299
    },
    {
      "epoch": 1.9207683073229291,
      "grad_norm": 3.20360356272725,
      "learning_rate": 9.050238232065299e-09,
      "loss": 0.1798,
      "step": 300
    },
    {
      "epoch": 1.927170868347339,
      "grad_norm": 3.1211562291161457,
      "learning_rate": 7.606549705035935e-09,
      "loss": 0.2063,
      "step": 301
    },
    {
      "epoch": 1.9335734293717488,
      "grad_norm": 3.514934313160573,
      "learning_rate": 6.2877901067573955e-09,
      "loss": 0.1871,
      "step": 302
    },
    {
      "epoch": 1.9399759903961584,
      "grad_norm": 3.080937668829297,
      "learning_rate": 5.094125451247655e-09,
      "loss": 0.2097,
      "step": 303
    },
    {
      "epoch": 1.9463785514205683,
      "grad_norm": 3.035817857990733,
      "learning_rate": 4.025706004760931e-09,
      "loss": 0.2057,
      "step": 304
    },
    {
      "epoch": 1.952781112444978,
      "grad_norm": 3.3455920299884734,
      "learning_rate": 3.082666266872036e-09,
      "loss": 0.2198,
      "step": 305
    },
    {
      "epoch": 1.9591836734693877,
      "grad_norm": 3.2826612214440845,
      "learning_rate": 2.2651249535439177e-09,
      "loss": 0.2028,
      "step": 306
    },
    {
      "epoch": 1.9655862344937975,
      "grad_norm": 3.3305034767663546,
      "learning_rate": 1.5731849821833953e-09,
      "loss": 0.2134,
      "step": 307
    },
    {
      "epoch": 1.9719887955182074,
      "grad_norm": 3.5471720967981573,
      "learning_rate": 1.0069334586854105e-09,
      "loss": 0.1908,
      "step": 308
    },
    {
      "epoch": 1.978391356542617,
      "grad_norm": 3.0004004418384542,
      "learning_rate": 5.664416664666882e-10,
      "loss": 0.2138,
      "step": 309
    },
    {
      "epoch": 1.9847939175670268,
      "grad_norm": 2.8720482666550136,
      "learning_rate": 2.517650574934693e-10,
      "loss": 0.2277,
      "step": 310
    },
    {
      "epoch": 1.9911964785914367,
      "grad_norm": 2.7009139828058832,
      "learning_rate": 6.29432452994294e-11,
      "loss": 0.215,
      "step": 311
    },
    {
      "epoch": 1.9975990396158463,
      "grad_norm": 3.2699084579180306,
      "learning_rate": 0.0,
      "loss": 0.2042,
      "step": 312
    },
    {
      "epoch": 1.9975990396158463,
      "step": 312,
      "total_flos": 212071482654720.0,
      "train_loss": 0.2767890989780426,
      "train_runtime": 7611.443,
      "train_samples_per_second": 10.506,
      "train_steps_per_second": 0.041
    }
  ],
  "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": 212071482654720.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}