{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.008947845247016454,
  "eval_steps": 500,
  "global_step": 800,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.1184806558770566e-05,
      "grad_norm": 9.17889404296875,
      "learning_rate": 0.00019999999993826567,
      "loss": 4.6752,
      "step": 1
    },
    {
      "epoch": 2.236961311754113e-05,
      "grad_norm": 15.023734092712402,
      "learning_rate": 0.0001999999997530627,
      "loss": 4.8816,
      "step": 2
    },
    {
      "epoch": 3.35544196763117e-05,
      "grad_norm": 4.711775302886963,
      "learning_rate": 0.00019999999944439107,
      "loss": 4.3122,
      "step": 3
    },
    {
      "epoch": 4.473922623508226e-05,
      "grad_norm": 5.1041460037231445,
      "learning_rate": 0.0001999999990122508,
      "loss": 4.0207,
      "step": 4
    },
    {
      "epoch": 5.5924032793852833e-05,
      "grad_norm": 11.579492568969727,
      "learning_rate": 0.0001999999984566419,
      "loss": 3.656,
      "step": 5
    },
    {
      "epoch": 6.71088393526234e-05,
      "grad_norm": 4.573488235473633,
      "learning_rate": 0.00019999999777756431,
      "loss": 3.3136,
      "step": 6
    },
    {
      "epoch": 7.829364591139397e-05,
      "grad_norm": 3.6844234466552734,
      "learning_rate": 0.0001999999969750181,
      "loss": 3.0363,
      "step": 7
    },
    {
      "epoch": 8.947845247016453e-05,
      "grad_norm": 2.9362566471099854,
      "learning_rate": 0.00019999999604900323,
      "loss": 2.7911,
      "step": 8
    },
    {
      "epoch": 0.0001006632590289351,
      "grad_norm": 2.6654202938079834,
      "learning_rate": 0.0001999999949995197,
      "loss": 2.6176,
      "step": 9
    },
    {
      "epoch": 0.00011184806558770567,
      "grad_norm": 1.8987364768981934,
      "learning_rate": 0.00019999999382656758,
      "loss": 2.5172,
      "step": 10
    },
    {
      "epoch": 0.00012303287214647624,
      "grad_norm": 2.596072196960449,
      "learning_rate": 0.0001999999925301468,
      "loss": 2.3978,
      "step": 11
    },
    {
      "epoch": 0.0001342176787052468,
      "grad_norm": 1.6658835411071777,
      "learning_rate": 0.00019999999111025733,
      "loss": 2.2888,
      "step": 12
    },
    {
      "epoch": 0.00014540248526401735,
      "grad_norm": 1.5891242027282715,
      "learning_rate": 0.00019999998956689926,
      "loss": 2.1966,
      "step": 13
    },
    {
      "epoch": 0.00015658729182278794,
      "grad_norm": 1.7551047801971436,
      "learning_rate": 0.00019999998790007256,
      "loss": 2.1286,
      "step": 14
    },
    {
      "epoch": 0.0001677720983815585,
      "grad_norm": 1.515837550163269,
      "learning_rate": 0.0001999999861097772,
      "loss": 2.0329,
      "step": 15
    },
    {
      "epoch": 0.00017895690494032905,
      "grad_norm": 2.0875966548919678,
      "learning_rate": 0.0001999999841960132,
      "loss": 1.9836,
      "step": 16
    },
    {
      "epoch": 0.00019014171149909964,
      "grad_norm": 1.3364766836166382,
      "learning_rate": 0.0001999999821587806,
      "loss": 1.9412,
      "step": 17
    },
    {
      "epoch": 0.0002013265180578702,
      "grad_norm": 0.9586036205291748,
      "learning_rate": 0.00019999997999807934,
      "loss": 1.9021,
      "step": 18
    },
    {
      "epoch": 0.00021251132461664075,
      "grad_norm": 0.610419750213623,
      "learning_rate": 0.00019999997771390947,
      "loss": 1.9278,
      "step": 19
    },
    {
      "epoch": 0.00022369613117541133,
      "grad_norm": 0.6592239141464233,
      "learning_rate": 0.000199999975306271,
      "loss": 1.8956,
      "step": 20
    },
    {
      "epoch": 0.0002348809377341819,
      "grad_norm": 0.7091565132141113,
      "learning_rate": 0.00019999997277516388,
      "loss": 1.8629,
      "step": 21
    },
    {
      "epoch": 0.0002460657442929525,
      "grad_norm": 0.6687048077583313,
      "learning_rate": 0.00019999997012058819,
      "loss": 1.818,
      "step": 22
    },
    {
      "epoch": 0.000257250550851723,
      "grad_norm": 0.29321762919425964,
      "learning_rate": 0.00019999996734254382,
      "loss": 1.8024,
      "step": 23
    },
    {
      "epoch": 0.0002684353574104936,
      "grad_norm": 0.6186531186103821,
      "learning_rate": 0.00019999996444103086,
      "loss": 1.7958,
      "step": 24
    },
    {
      "epoch": 0.0002796201639692642,
      "grad_norm": 0.4960622489452362,
      "learning_rate": 0.0001999999614160493,
      "loss": 1.7714,
      "step": 25
    },
    {
      "epoch": 0.0002908049705280347,
      "grad_norm": 0.25318390130996704,
      "learning_rate": 0.00019999995826759916,
      "loss": 1.7419,
      "step": 26
    },
    {
      "epoch": 0.0003019897770868053,
      "grad_norm": 0.5521177649497986,
      "learning_rate": 0.0001999999549956804,
      "loss": 1.7336,
      "step": 27
    },
    {
      "epoch": 0.0003131745836455759,
      "grad_norm": 0.3085158169269562,
      "learning_rate": 0.00019999995160029305,
      "loss": 1.7304,
      "step": 28
    },
    {
      "epoch": 0.0003243593902043464,
      "grad_norm": 0.2978903353214264,
      "learning_rate": 0.0001999999480814371,
      "loss": 1.7283,
      "step": 29
    },
    {
      "epoch": 0.000335544196763117,
      "grad_norm": 0.40339481830596924,
      "learning_rate": 0.00019999994443911258,
      "loss": 1.7577,
      "step": 30
    },
    {
      "epoch": 0.00034672900332188757,
      "grad_norm": 0.13451404869556427,
      "learning_rate": 0.00019999994067331945,
      "loss": 1.7435,
      "step": 31
    },
    {
      "epoch": 0.0003579138098806581,
      "grad_norm": 0.3141914904117584,
      "learning_rate": 0.0001999999367840578,
      "loss": 1.7479,
      "step": 32
    },
    {
      "epoch": 0.0003690986164394287,
      "grad_norm": 0.12182258069515228,
      "learning_rate": 0.00019999993277132754,
      "loss": 1.7391,
      "step": 33
    },
    {
      "epoch": 0.00038028342299819927,
      "grad_norm": 0.3160305917263031,
      "learning_rate": 0.00019999992863512872,
      "loss": 1.7231,
      "step": 34
    },
    {
      "epoch": 0.0003914682295569698,
      "grad_norm": 0.18215563893318176,
      "learning_rate": 0.00019999992437546134,
      "loss": 1.7067,
      "step": 35
    },
    {
      "epoch": 0.0004026530361157404,
      "grad_norm": 0.24103401601314545,
      "learning_rate": 0.0001999999199923254,
      "loss": 1.6877,
      "step": 36
    },
    {
      "epoch": 0.00041383784267451097,
      "grad_norm": 0.17353500425815582,
      "learning_rate": 0.0001999999154857209,
      "loss": 1.6746,
      "step": 37
    },
    {
      "epoch": 0.0004250226492332815,
      "grad_norm": 0.19149154424667358,
      "learning_rate": 0.00019999991085564784,
      "loss": 1.6734,
      "step": 38
    },
    {
      "epoch": 0.0004362074557920521,
      "grad_norm": 0.15810631215572357,
      "learning_rate": 0.0001999999061021063,
      "loss": 1.6773,
      "step": 39
    },
    {
      "epoch": 0.00044739226235082267,
      "grad_norm": 0.14770038425922394,
      "learning_rate": 0.00019999990122509614,
      "loss": 1.6967,
      "step": 40
    },
    {
      "epoch": 0.0004585770689095932,
      "grad_norm": 0.15101520717144012,
      "learning_rate": 0.0001999998962246175,
      "loss": 1.6816,
      "step": 41
    },
    {
      "epoch": 0.0004697618754683638,
      "grad_norm": 0.1085171177983284,
      "learning_rate": 0.00019999989110067035,
      "loss": 1.6875,
      "step": 42
    },
    {
      "epoch": 0.00048094668202713437,
      "grad_norm": 0.13066136837005615,
      "learning_rate": 0.00019999988585325468,
      "loss": 1.6768,
      "step": 43
    },
    {
      "epoch": 0.000492131488585905,
      "grad_norm": 0.09783171862363815,
      "learning_rate": 0.0001999998804823705,
      "loss": 1.6559,
      "step": 44
    },
    {
      "epoch": 0.0005033162951446755,
      "grad_norm": 0.15001484751701355,
      "learning_rate": 0.00019999987498801777,
      "loss": 1.6662,
      "step": 45
    },
    {
      "epoch": 0.000514501101703446,
      "grad_norm": 0.06484243273735046,
      "learning_rate": 0.0001999998693701966,
      "loss": 1.6528,
      "step": 46
    },
    {
      "epoch": 0.0005256859082622166,
      "grad_norm": 0.12908180058002472,
      "learning_rate": 0.00019999986362890693,
      "loss": 1.67,
      "step": 47
    },
    {
      "epoch": 0.0005368707148209872,
      "grad_norm": 0.05700545758008957,
      "learning_rate": 0.00019999985776414877,
      "loss": 1.6643,
      "step": 48
    },
    {
      "epoch": 0.0005480555213797577,
      "grad_norm": 0.10586538165807724,
      "learning_rate": 0.00019999985177592211,
      "loss": 1.6725,
      "step": 49
    },
    {
      "epoch": 0.0005592403279385283,
      "grad_norm": 0.05411362275481224,
      "learning_rate": 0.00019999984566422703,
      "loss": 1.6495,
      "step": 50
    },
    {
      "epoch": 0.0005704251344972989,
      "grad_norm": 0.08841974288225174,
      "learning_rate": 0.00019999983942906347,
      "loss": 1.6397,
      "step": 51
    },
    {
      "epoch": 0.0005816099410560694,
      "grad_norm": 0.049202822148799896,
      "learning_rate": 0.00019999983307043145,
      "loss": 1.6601,
      "step": 52
    },
    {
      "epoch": 0.00059279474761484,
      "grad_norm": 0.06537283957004547,
      "learning_rate": 0.00019999982658833098,
      "loss": 1.6405,
      "step": 53
    },
    {
      "epoch": 0.0006039795541736106,
      "grad_norm": 0.04358899965882301,
      "learning_rate": 0.0001999998199827621,
      "loss": 1.6586,
      "step": 54
    },
    {
      "epoch": 0.0006151643607323811,
      "grad_norm": 0.05924156308174133,
      "learning_rate": 0.0001999998132537248,
      "loss": 1.6609,
      "step": 55
    },
    {
      "epoch": 0.0006263491672911517,
      "grad_norm": 0.047364529222249985,
      "learning_rate": 0.00019999980640121904,
      "loss": 1.6561,
      "step": 56
    },
    {
      "epoch": 0.0006375339738499223,
      "grad_norm": 0.05860909819602966,
      "learning_rate": 0.00019999979942524488,
      "loss": 1.644,
      "step": 57
    },
    {
      "epoch": 0.0006487187804086928,
      "grad_norm": 0.058639075607061386,
      "learning_rate": 0.00019999979232580235,
      "loss": 1.6582,
      "step": 58
    },
    {
      "epoch": 0.0006599035869674634,
      "grad_norm": 0.049288246780633926,
      "learning_rate": 0.00019999978510289138,
      "loss": 1.6828,
      "step": 59
    },
    {
      "epoch": 0.000671088393526234,
      "grad_norm": 0.06293074041604996,
      "learning_rate": 0.00019999977775651207,
      "loss": 1.6852,
      "step": 60
    },
    {
      "epoch": 0.0006822732000850045,
      "grad_norm": 0.03848947212100029,
      "learning_rate": 0.00019999977028666436,
      "loss": 1.6792,
      "step": 61
    },
    {
      "epoch": 0.0006934580066437751,
      "grad_norm": 0.1217992901802063,
      "learning_rate": 0.00019999976269334828,
      "loss": 1.7142,
      "step": 62
    },
    {
      "epoch": 0.0007046428132025457,
      "grad_norm": 0.04006423428654671,
      "learning_rate": 0.00019999975497656384,
      "loss": 1.7108,
      "step": 63
    },
    {
      "epoch": 0.0007158276197613162,
      "grad_norm": 0.057927753776311874,
      "learning_rate": 0.0001999997471363111,
      "loss": 1.6902,
      "step": 64
    },
    {
      "epoch": 0.0007270124263200868,
      "grad_norm": 0.04537767171859741,
      "learning_rate": 0.00019999973917258997,
      "loss": 1.6627,
      "step": 65
    },
    {
      "epoch": 0.0007381972328788574,
      "grad_norm": 0.057307057082653046,
      "learning_rate": 0.00019999973108540052,
      "loss": 1.6561,
      "step": 66
    },
    {
      "epoch": 0.0007493820394376279,
      "grad_norm": 0.044002167880535126,
      "learning_rate": 0.00019999972287474272,
      "loss": 1.6595,
      "step": 67
    },
    {
      "epoch": 0.0007605668459963985,
      "grad_norm": 0.03991294279694557,
      "learning_rate": 0.00019999971454061666,
      "loss": 1.6563,
      "step": 68
    },
    {
      "epoch": 0.0007717516525551691,
      "grad_norm": 0.044279683381319046,
      "learning_rate": 0.0001999997060830223,
      "loss": 1.6352,
      "step": 69
    },
    {
      "epoch": 0.0007829364591139396,
      "grad_norm": 0.04319776967167854,
      "learning_rate": 0.00019999969750195967,
      "loss": 1.6319,
      "step": 70
    },
    {
      "epoch": 0.0007941212656727102,
      "grad_norm": 0.04436005651950836,
      "learning_rate": 0.00019999968879742873,
      "loss": 1.6556,
      "step": 71
    },
    {
      "epoch": 0.0008053060722314808,
      "grad_norm": 0.0416998416185379,
      "learning_rate": 0.00019999967996942952,
      "loss": 1.6646,
      "step": 72
    },
    {
      "epoch": 0.0008164908787902513,
      "grad_norm": 0.03601493313908577,
      "learning_rate": 0.00019999967101796208,
      "loss": 1.6605,
      "step": 73
    },
    {
      "epoch": 0.0008276756853490219,
      "grad_norm": 0.03957024961709976,
      "learning_rate": 0.00019999966194302637,
      "loss": 1.6714,
      "step": 74
    },
    {
      "epoch": 0.0008388604919077925,
      "grad_norm": 0.04401829466223717,
      "learning_rate": 0.00019999965274462245,
      "loss": 1.6497,
      "step": 75
    },
    {
      "epoch": 0.000850045298466563,
      "grad_norm": 0.04251580312848091,
      "learning_rate": 0.0001999996434227503,
      "loss": 1.6567,
      "step": 76
    },
    {
      "epoch": 0.0008612301050253336,
      "grad_norm": 0.038915786892175674,
      "learning_rate": 0.00019999963397740995,
      "loss": 1.6515,
      "step": 77
    },
    {
      "epoch": 0.0008724149115841042,
      "grad_norm": 0.04063812270760536,
      "learning_rate": 0.00019999962440860137,
      "loss": 1.6764,
      "step": 78
    },
    {
      "epoch": 0.0008835997181428747,
      "grad_norm": 0.04114954546093941,
      "learning_rate": 0.00019999961471632463,
      "loss": 1.6553,
      "step": 79
    },
    {
      "epoch": 0.0008947845247016453,
      "grad_norm": 0.039563097059726715,
      "learning_rate": 0.0001999996049005797,
      "loss": 1.6595,
      "step": 80
    },
    {
      "epoch": 0.0009059693312604159,
      "grad_norm": 0.03477632254362106,
      "learning_rate": 0.00019999959496136663,
      "loss": 1.6612,
      "step": 81
    },
    {
      "epoch": 0.0009171541378191864,
      "grad_norm": 0.04104992002248764,
      "learning_rate": 0.0001999995848986854,
      "loss": 1.6476,
      "step": 82
    },
    {
      "epoch": 0.000928338944377957,
      "grad_norm": 0.03909602388739586,
      "learning_rate": 0.00019999957471253602,
      "loss": 1.6445,
      "step": 83
    },
    {
      "epoch": 0.0009395237509367276,
      "grad_norm": 0.039677415043115616,
      "learning_rate": 0.00019999956440291855,
      "loss": 1.6368,
      "step": 84
    },
    {
      "epoch": 0.0009507085574954981,
      "grad_norm": 0.03493763506412506,
      "learning_rate": 0.00019999955396983292,
      "loss": 1.6343,
      "step": 85
    },
    {
      "epoch": 0.0009618933640542687,
      "grad_norm": 0.03530074283480644,
      "learning_rate": 0.0001999995434132792,
      "loss": 1.6843,
      "step": 86
    },
    {
      "epoch": 0.0009730781706130393,
      "grad_norm": 0.037156637758016586,
      "learning_rate": 0.00019999953273325743,
      "loss": 1.6721,
      "step": 87
    },
    {
      "epoch": 0.00098426297717181,
      "grad_norm": 0.04006032645702362,
      "learning_rate": 0.00019999952192976755,
      "loss": 1.6724,
      "step": 88
    },
    {
      "epoch": 0.0009954477837305804,
      "grad_norm": 0.03743763267993927,
      "learning_rate": 0.0001999995110028096,
      "loss": 1.6613,
      "step": 89
    },
    {
      "epoch": 0.001006632590289351,
      "grad_norm": 0.04100384563207626,
      "learning_rate": 0.00019999949995238369,
      "loss": 1.6444,
      "step": 90
    },
    {
      "epoch": 0.0010178173968481215,
      "grad_norm": 0.03799253702163696,
      "learning_rate": 0.00019999948877848965,
      "loss": 1.6641,
      "step": 91
    },
    {
      "epoch": 0.001029002203406892,
      "grad_norm": 0.040163811296224594,
      "learning_rate": 0.00019999947748112763,
      "loss": 1.6411,
      "step": 92
    },
    {
      "epoch": 0.0010401870099656626,
      "grad_norm": 0.03576591610908508,
      "learning_rate": 0.0001999994660602976,
      "loss": 1.6378,
      "step": 93
    },
    {
      "epoch": 0.0010513718165244333,
      "grad_norm": 0.03735070303082466,
      "learning_rate": 0.00019999945451599957,
      "loss": 1.644,
      "step": 94
    },
    {
      "epoch": 0.0010625566230832038,
      "grad_norm": 0.04157353192567825,
      "learning_rate": 0.00019999944284823358,
      "loss": 1.6532,
      "step": 95
    },
    {
      "epoch": 0.0010737414296419744,
      "grad_norm": 0.046478450298309326,
      "learning_rate": 0.0001999994310569996,
      "loss": 1.6744,
      "step": 96
    },
    {
      "epoch": 0.0010849262362007449,
      "grad_norm": 0.04813043400645256,
      "learning_rate": 0.0001999994191422977,
      "loss": 1.6716,
      "step": 97
    },
    {
      "epoch": 0.0010961110427595154,
      "grad_norm": 0.03780042380094528,
      "learning_rate": 0.00019999940710412788,
      "loss": 1.6502,
      "step": 98
    },
    {
      "epoch": 0.001107295849318286,
      "grad_norm": 0.03811471536755562,
      "learning_rate": 0.0001999993949424901,
      "loss": 1.6405,
      "step": 99
    },
    {
      "epoch": 0.0011184806558770567,
      "grad_norm": 0.03838631138205528,
      "learning_rate": 0.00019999938265738445,
      "loss": 1.6202,
      "step": 100
    },
    {
      "epoch": 0.0011296654624358272,
      "grad_norm": 0.04033266752958298,
      "learning_rate": 0.0001999993702488109,
      "loss": 1.6305,
      "step": 101
    },
    {
      "epoch": 0.0011408502689945978,
      "grad_norm": 0.038872625678777695,
      "learning_rate": 0.0001999993577167695,
      "loss": 1.6243,
      "step": 102
    },
    {
      "epoch": 0.0011520350755533683,
      "grad_norm": 0.047068770974874496,
      "learning_rate": 0.0001999993450612602,
      "loss": 1.6126,
      "step": 103
    },
    {
      "epoch": 0.0011632198821121388,
      "grad_norm": 0.038774486631155014,
      "learning_rate": 0.0001999993322822831,
      "loss": 1.6125,
      "step": 104
    },
    {
      "epoch": 0.0011744046886709093,
      "grad_norm": 0.046706534922122955,
      "learning_rate": 0.00019999931937983814,
      "loss": 1.6062,
      "step": 105
    },
    {
      "epoch": 0.00118558949522968,
      "grad_norm": 0.038454532623291016,
      "learning_rate": 0.00019999930635392538,
      "loss": 1.6145,
      "step": 106
    },
    {
      "epoch": 0.0011967743017884506,
      "grad_norm": 0.042938027530908585,
      "learning_rate": 0.0001999992932045448,
      "loss": 1.6064,
      "step": 107
    },
    {
      "epoch": 0.0012079591083472212,
      "grad_norm": 0.03728878125548363,
      "learning_rate": 0.00019999927993169652,
      "loss": 1.6203,
      "step": 108
    },
    {
      "epoch": 0.0012191439149059917,
      "grad_norm": 0.03831510245800018,
      "learning_rate": 0.00019999926653538043,
      "loss": 1.6214,
      "step": 109
    },
    {
      "epoch": 0.0012303287214647622,
      "grad_norm": 0.04032002389431,
      "learning_rate": 0.00019999925301559659,
      "loss": 1.6116,
      "step": 110
    },
    {
      "epoch": 0.0012415135280235327,
      "grad_norm": 0.04648670554161072,
      "learning_rate": 0.00019999923937234505,
      "loss": 1.6228,
      "step": 111
    },
    {
      "epoch": 0.0012526983345823035,
      "grad_norm": 0.03572435304522514,
      "learning_rate": 0.00019999922560562575,
      "loss": 1.6382,
      "step": 112
    },
    {
      "epoch": 0.001263883141141074,
      "grad_norm": 0.03520497307181358,
      "learning_rate": 0.0001999992117154388,
      "loss": 1.642,
      "step": 113
    },
    {
      "epoch": 0.0012750679476998446,
      "grad_norm": 0.037833768874406815,
      "learning_rate": 0.00019999919770178414,
      "loss": 1.6809,
      "step": 114
    },
    {
      "epoch": 0.001286252754258615,
      "grad_norm": 0.04116043448448181,
      "learning_rate": 0.00019999918356466186,
      "loss": 1.6669,
      "step": 115
    },
    {
      "epoch": 0.0012974375608173856,
      "grad_norm": 0.03686891868710518,
      "learning_rate": 0.00019999916930407192,
      "loss": 1.6336,
      "step": 116
    },
    {
      "epoch": 0.0013086223673761561,
      "grad_norm": 0.04540928825736046,
      "learning_rate": 0.00019999915492001434,
      "loss": 1.636,
      "step": 117
    },
    {
      "epoch": 0.0013198071739349269,
      "grad_norm": 0.03894014656543732,
      "learning_rate": 0.00019999914041248917,
      "loss": 1.642,
      "step": 118
    },
    {
      "epoch": 0.0013309919804936974,
      "grad_norm": 0.03868821635842323,
      "learning_rate": 0.0001999991257814964,
      "loss": 1.6459,
      "step": 119
    },
    {
      "epoch": 0.001342176787052468,
      "grad_norm": 0.05047065392136574,
      "learning_rate": 0.00019999911102703606,
      "loss": 1.6566,
      "step": 120
    },
    {
      "epoch": 0.0013533615936112385,
      "grad_norm": 0.037026163190603256,
      "learning_rate": 0.0001999990961491082,
      "loss": 1.644,
      "step": 121
    },
    {
      "epoch": 0.001364546400170009,
      "grad_norm": 0.05025867745280266,
      "learning_rate": 0.00019999908114771278,
      "loss": 1.6375,
      "step": 122
    },
    {
      "epoch": 0.0013757312067287795,
      "grad_norm": 0.05618799850344658,
      "learning_rate": 0.0001999990660228498,
      "loss": 1.6414,
      "step": 123
    },
    {
      "epoch": 0.0013869160132875503,
      "grad_norm": 0.04012198746204376,
      "learning_rate": 0.0001999990507745194,
      "loss": 1.6338,
      "step": 124
    },
    {
      "epoch": 0.0013981008198463208,
      "grad_norm": 0.03692522644996643,
      "learning_rate": 0.00019999903540272147,
      "loss": 1.6338,
      "step": 125
    },
    {
      "epoch": 0.0014092856264050913,
      "grad_norm": 0.04653295874595642,
      "learning_rate": 0.0001999990199074561,
      "loss": 1.6242,
      "step": 126
    },
    {
      "epoch": 0.0014204704329638619,
      "grad_norm": 0.038410138338804245,
      "learning_rate": 0.0001999990042887233,
      "loss": 1.6535,
      "step": 127
    },
    {
      "epoch": 0.0014316552395226324,
      "grad_norm": 0.03855321556329727,
      "learning_rate": 0.00019999898854652307,
      "loss": 1.6622,
      "step": 128
    },
    {
      "epoch": 0.001442840046081403,
      "grad_norm": 0.039161138236522675,
      "learning_rate": 0.00019999897268085543,
      "loss": 1.6596,
      "step": 129
    },
    {
      "epoch": 0.0014540248526401737,
      "grad_norm": 0.04229150339961052,
      "learning_rate": 0.00019999895669172042,
      "loss": 1.6044,
      "step": 130
    },
    {
      "epoch": 0.0014652096591989442,
      "grad_norm": 0.04695671424269676,
      "learning_rate": 0.00019999894057911804,
      "loss": 1.593,
      "step": 131
    },
    {
      "epoch": 0.0014763944657577147,
      "grad_norm": 0.05007043853402138,
      "learning_rate": 0.00019999892434304832,
      "loss": 1.6371,
      "step": 132
    },
    {
      "epoch": 0.0014875792723164853,
      "grad_norm": 0.04140834882855415,
      "learning_rate": 0.00019999890798351127,
      "loss": 1.6333,
      "step": 133
    },
    {
      "epoch": 0.0014987640788752558,
      "grad_norm": 0.03771176561713219,
      "learning_rate": 0.0001999988915005069,
      "loss": 1.6253,
      "step": 134
    },
    {
      "epoch": 0.0015099488854340263,
      "grad_norm": 0.03835882246494293,
      "learning_rate": 0.00019999887489403532,
      "loss": 1.6096,
      "step": 135
    },
    {
      "epoch": 0.001521133691992797,
      "grad_norm": 0.042071614414453506,
      "learning_rate": 0.00019999885816409643,
      "loss": 1.6158,
      "step": 136
    },
    {
      "epoch": 0.0015323184985515676,
      "grad_norm": 0.04736172780394554,
      "learning_rate": 0.0001999988413106903,
      "loss": 1.6076,
      "step": 137
    },
    {
      "epoch": 0.0015435033051103381,
      "grad_norm": 0.045473724603652954,
      "learning_rate": 0.00019999882433381695,
      "loss": 1.6002,
      "step": 138
    },
    {
      "epoch": 0.0015546881116691087,
      "grad_norm": 0.038239121437072754,
      "learning_rate": 0.0001999988072334764,
      "loss": 1.6093,
      "step": 139
    },
    {
      "epoch": 0.0015658729182278792,
      "grad_norm": 0.03806670382618904,
      "learning_rate": 0.0001999987900096687,
      "loss": 1.5958,
      "step": 140
    },
    {
      "epoch": 0.0015770577247866497,
      "grad_norm": 0.034363262355327606,
      "learning_rate": 0.00019999877266239382,
      "loss": 1.6182,
      "step": 141
    },
    {
      "epoch": 0.0015882425313454205,
      "grad_norm": 0.03781922906637192,
      "learning_rate": 0.0001999987551916518,
      "loss": 1.6041,
      "step": 142
    },
    {
      "epoch": 0.001599427337904191,
      "grad_norm": 0.037946417927742004,
      "learning_rate": 0.00019999873759744268,
      "loss": 1.6154,
      "step": 143
    },
    {
      "epoch": 0.0016106121444629615,
      "grad_norm": 0.03221859410405159,
      "learning_rate": 0.00019999871987976645,
      "loss": 1.6205,
      "step": 144
    },
    {
      "epoch": 0.001621796951021732,
      "grad_norm": 0.034399550408124924,
      "learning_rate": 0.00019999870203862318,
      "loss": 1.6077,
      "step": 145
    },
    {
      "epoch": 0.0016329817575805026,
      "grad_norm": 0.037965498864650726,
      "learning_rate": 0.00019999868407401285,
      "loss": 1.6226,
      "step": 146
    },
    {
      "epoch": 0.0016441665641392731,
      "grad_norm": 0.034158267080783844,
      "learning_rate": 0.00019999866598593549,
      "loss": 1.6364,
      "step": 147
    },
    {
      "epoch": 0.0016553513706980439,
      "grad_norm": 0.035179782658815384,
      "learning_rate": 0.00019999864777439113,
      "loss": 1.6015,
      "step": 148
    },
    {
      "epoch": 0.0016665361772568144,
      "grad_norm": 0.07380399852991104,
      "learning_rate": 0.00019999862943937977,
      "loss": 1.6036,
      "step": 149
    },
    {
      "epoch": 0.001677720983815585,
      "grad_norm": 0.055961351841688156,
      "learning_rate": 0.00019999861098090146,
      "loss": 1.6209,
      "step": 150
    },
    {
      "epoch": 0.0016889057903743555,
      "grad_norm": 0.06556452065706253,
      "learning_rate": 0.00019999859239895623,
      "loss": 1.6179,
      "step": 151
    },
    {
      "epoch": 0.001700090596933126,
      "grad_norm": 0.0406390018761158,
      "learning_rate": 0.0001999985736935441,
      "loss": 1.6251,
      "step": 152
    },
    {
      "epoch": 0.0017112754034918965,
      "grad_norm": 0.04337646812200546,
      "learning_rate": 0.00019999855486466504,
      "loss": 1.6229,
      "step": 153
    },
    {
      "epoch": 0.0017224602100506673,
      "grad_norm": 0.03773214668035507,
      "learning_rate": 0.00019999853591231914,
      "loss": 1.6163,
      "step": 154
    },
    {
      "epoch": 0.0017336450166094378,
      "grad_norm": 0.04170192405581474,
      "learning_rate": 0.0001999985168365064,
      "loss": 1.6187,
      "step": 155
    },
    {
      "epoch": 0.0017448298231682083,
      "grad_norm": 0.046784352511167526,
      "learning_rate": 0.00019999849763722684,
      "loss": 1.6313,
      "step": 156
    },
    {
      "epoch": 0.0017560146297269789,
      "grad_norm": 0.04641957953572273,
      "learning_rate": 0.00019999847831448048,
      "loss": 1.6353,
      "step": 157
    },
    {
      "epoch": 0.0017671994362857494,
      "grad_norm": 0.04125691205263138,
      "learning_rate": 0.00019999845886826736,
      "loss": 1.6136,
      "step": 158
    },
    {
      "epoch": 0.00177838424284452,
      "grad_norm": 0.03794560208916664,
      "learning_rate": 0.00019999843929858748,
      "loss": 1.6124,
      "step": 159
    },
    {
      "epoch": 0.0017895690494032907,
      "grad_norm": 0.04494895413517952,
      "learning_rate": 0.00019999841960544087,
      "loss": 1.6038,
      "step": 160
    },
    {
      "epoch": 0.0018007538559620612,
      "grad_norm": 0.0605354905128479,
      "learning_rate": 0.00019999839978882756,
      "loss": 1.6129,
      "step": 161
    },
    {
      "epoch": 0.0018119386625208317,
      "grad_norm": 0.03629598766565323,
      "learning_rate": 0.0001999983798487476,
      "loss": 1.6136,
      "step": 162
    },
    {
      "epoch": 0.0018231234690796023,
      "grad_norm": 0.051288772374391556,
      "learning_rate": 0.00019999835978520099,
      "loss": 1.6121,
      "step": 163
    },
    {
      "epoch": 0.0018343082756383728,
      "grad_norm": 0.0646372139453888,
      "learning_rate": 0.00019999833959818774,
      "loss": 1.6097,
      "step": 164
    },
    {
      "epoch": 0.0018454930821971433,
      "grad_norm": 0.05430466681718826,
      "learning_rate": 0.00019999831928770788,
      "loss": 1.6336,
      "step": 165
    },
    {
      "epoch": 0.001856677888755914,
      "grad_norm": 0.04446660354733467,
      "learning_rate": 0.00019999829885376146,
      "loss": 1.6446,
      "step": 166
    },
    {
      "epoch": 0.0018678626953146846,
      "grad_norm": 0.05089535191655159,
      "learning_rate": 0.0001999982782963485,
      "loss": 1.6295,
      "step": 167
    },
    {
      "epoch": 0.0018790475018734551,
      "grad_norm": 0.04738520458340645,
      "learning_rate": 0.000199998257615469,
      "loss": 1.6129,
      "step": 168
    },
    {
      "epoch": 0.0018902323084322257,
      "grad_norm": 0.06705950200557709,
      "learning_rate": 0.00019999823681112296,
      "loss": 1.6091,
      "step": 169
    },
    {
      "epoch": 0.0019014171149909962,
      "grad_norm": 0.08080725371837616,
      "learning_rate": 0.0001999982158833105,
      "loss": 1.6041,
      "step": 170
    },
    {
      "epoch": 0.0019126019215497667,
      "grad_norm": 0.08998562395572662,
      "learning_rate": 0.00019999819483203162,
      "loss": 1.61,
      "step": 171
    },
    {
      "epoch": 0.0019237867281085375,
      "grad_norm": 0.0561324767768383,
      "learning_rate": 0.00019999817365728626,
      "loss": 1.6019,
      "step": 172
    },
    {
      "epoch": 0.001934971534667308,
      "grad_norm": 0.06384976953268051,
      "learning_rate": 0.00019999815235907453,
      "loss": 1.6045,
      "step": 173
    },
    {
      "epoch": 0.0019461563412260785,
      "grad_norm": 0.08231502771377563,
      "learning_rate": 0.00019999813093739643,
      "loss": 1.5851,
      "step": 174
    },
    {
      "epoch": 0.001957341147784849,
      "grad_norm": 0.06544940173625946,
      "learning_rate": 0.00019999810939225196,
      "loss": 1.5902,
      "step": 175
    },
    {
      "epoch": 0.00196852595434362,
      "grad_norm": 0.05647379904985428,
      "learning_rate": 0.0001999980877236412,
      "loss": 1.5972,
      "step": 176
    },
    {
      "epoch": 0.00197971076090239,
      "grad_norm": 0.04889946058392525,
      "learning_rate": 0.00019999806593156417,
      "loss": 1.6039,
      "step": 177
    },
    {
      "epoch": 0.001990895567461161,
      "grad_norm": 0.04563042148947716,
      "learning_rate": 0.00019999804401602087,
      "loss": 1.6137,
      "step": 178
    },
    {
      "epoch": 0.002002080374019931,
      "grad_norm": 0.057878173887729645,
      "learning_rate": 0.0001999980219770113,
      "loss": 1.603,
      "step": 179
    },
    {
      "epoch": 0.002013265180578702,
      "grad_norm": 0.04965892806649208,
      "learning_rate": 0.00019999799981453554,
      "loss": 1.6016,
      "step": 180
    },
    {
      "epoch": 0.0020244499871374727,
      "grad_norm": 0.042930275201797485,
      "learning_rate": 0.00019999797752859362,
      "loss": 1.5872,
      "step": 181
    },
    {
      "epoch": 0.002035634793696243,
      "grad_norm": 0.07251135259866714,
      "learning_rate": 0.00019999795511918553,
      "loss": 1.5973,
      "step": 182
    },
    {
      "epoch": 0.0020468196002550137,
      "grad_norm": 0.053799793124198914,
      "learning_rate": 0.00019999793258631133,
      "loss": 1.6177,
      "step": 183
    },
    {
      "epoch": 0.002058004406813784,
      "grad_norm": 0.05833510681986809,
      "learning_rate": 0.00019999790992997101,
      "loss": 1.5898,
      "step": 184
    },
    {
      "epoch": 0.002069189213372555,
      "grad_norm": 0.06137058511376381,
      "learning_rate": 0.0001999978871501646,
      "loss": 1.5836,
      "step": 185
    },
    {
      "epoch": 0.002080374019931325,
      "grad_norm": 0.04289720579981804,
      "learning_rate": 0.0001999978642468922,
      "loss": 1.5893,
      "step": 186
    },
    {
      "epoch": 0.002091558826490096,
      "grad_norm": 0.054850902408361435,
      "learning_rate": 0.00019999784122015375,
      "loss": 1.5894,
      "step": 187
    },
    {
      "epoch": 0.0021027436330488666,
      "grad_norm": 0.06448347866535187,
      "learning_rate": 0.00019999781806994934,
      "loss": 1.5862,
      "step": 188
    },
    {
      "epoch": 0.002113928439607637,
      "grad_norm": 0.06109186261892319,
      "learning_rate": 0.00019999779479627897,
      "loss": 1.5951,
      "step": 189
    },
    {
      "epoch": 0.0021251132461664077,
      "grad_norm": 0.05638093128800392,
      "learning_rate": 0.00019999777139914263,
      "loss": 1.5929,
      "step": 190
    },
    {
      "epoch": 0.002136298052725178,
      "grad_norm": 0.05871524661779404,
      "learning_rate": 0.00019999774787854047,
      "loss": 1.5979,
      "step": 191
    },
    {
      "epoch": 0.0021474828592839487,
      "grad_norm": 0.05536816641688347,
      "learning_rate": 0.00019999772423447238,
      "loss": 1.598,
      "step": 192
    },
    {
      "epoch": 0.0021586676658427195,
      "grad_norm": 0.04985928162932396,
      "learning_rate": 0.00019999770046693848,
      "loss": 1.6067,
      "step": 193
    },
    {
      "epoch": 0.0021698524724014898,
      "grad_norm": 0.03956759348511696,
      "learning_rate": 0.00019999767657593874,
      "loss": 1.6101,
      "step": 194
    },
    {
      "epoch": 0.0021810372789602605,
      "grad_norm": 0.043052803725004196,
      "learning_rate": 0.00019999765256147324,
      "loss": 1.5925,
      "step": 195
    },
    {
      "epoch": 0.002192222085519031,
      "grad_norm": 0.049073830246925354,
      "learning_rate": 0.000199997628423542,
      "loss": 1.5997,
      "step": 196
    },
    {
      "epoch": 0.0022034068920778016,
      "grad_norm": 0.0505862683057785,
      "learning_rate": 0.00019999760416214503,
      "loss": 1.5976,
      "step": 197
    },
    {
      "epoch": 0.002214591698636572,
      "grad_norm": 0.04482674226164818,
      "learning_rate": 0.00019999757977728235,
      "loss": 1.5762,
      "step": 198
    },
    {
      "epoch": 0.0022257765051953426,
      "grad_norm": 0.03707127273082733,
      "learning_rate": 0.00019999755526895402,
      "loss": 1.5805,
      "step": 199
    },
    {
      "epoch": 0.0022369613117541134,
      "grad_norm": 0.041122857481241226,
      "learning_rate": 0.00019999753063716002,
      "loss": 1.6041,
      "step": 200
    },
    {
      "epoch": 0.0022481461183128837,
      "grad_norm": 0.05945773050189018,
      "learning_rate": 0.00019999750588190046,
      "loss": 1.6068,
      "step": 201
    },
    {
      "epoch": 0.0022593309248716545,
      "grad_norm": 0.058612506836652756,
      "learning_rate": 0.00019999748100317532,
      "loss": 1.6,
      "step": 202
    },
    {
      "epoch": 0.0022705157314304248,
      "grad_norm": 0.05361739546060562,
      "learning_rate": 0.00019999745600098466,
      "loss": 1.5891,
      "step": 203
    },
    {
      "epoch": 0.0022817005379891955,
      "grad_norm": 0.045548051595687866,
      "learning_rate": 0.00019999743087532846,
      "loss": 1.6161,
      "step": 204
    },
    {
      "epoch": 0.002292885344547966,
      "grad_norm": 0.04524560272693634,
      "learning_rate": 0.00019999740562620682,
      "loss": 1.6092,
      "step": 205
    },
    {
      "epoch": 0.0023040701511067366,
      "grad_norm": 0.04346180334687233,
      "learning_rate": 0.0001999973802536197,
      "loss": 1.6076,
      "step": 206
    },
    {
      "epoch": 0.0023152549576655073,
      "grad_norm": 0.047505974769592285,
      "learning_rate": 0.00019999735475756717,
      "loss": 1.5825,
      "step": 207
    },
    {
      "epoch": 0.0023264397642242776,
      "grad_norm": 0.03851678594946861,
      "learning_rate": 0.00019999732913804927,
      "loss": 1.5991,
      "step": 208
    },
    {
      "epoch": 0.0023376245707830484,
      "grad_norm": 0.051913902163505554,
      "learning_rate": 0.000199997303395066,
      "loss": 1.6281,
      "step": 209
    },
    {
      "epoch": 0.0023488093773418187,
      "grad_norm": 0.06070960685610771,
      "learning_rate": 0.0001999972775286174,
      "loss": 1.641,
      "step": 210
    },
    {
      "epoch": 0.0023599941839005894,
      "grad_norm": 0.13532494008541107,
      "learning_rate": 0.00019999725153870354,
      "loss": 1.6009,
      "step": 211
    },
    {
      "epoch": 0.00237117899045936,
      "grad_norm": 0.30072659254074097,
      "learning_rate": 0.00019999722542532442,
      "loss": 1.6163,
      "step": 212
    },
    {
      "epoch": 0.0023823637970181305,
      "grad_norm": 0.3511681854724884,
      "learning_rate": 0.00019999719918848004,
      "loss": 1.6465,
      "step": 213
    },
    {
      "epoch": 0.0023935486035769012,
      "grad_norm": 0.2487097680568695,
      "learning_rate": 0.00019999717282817052,
      "loss": 1.6367,
      "step": 214
    },
    {
      "epoch": 0.0024047334101356716,
      "grad_norm": 0.23582719266414642,
      "learning_rate": 0.00019999714634439582,
      "loss": 1.6202,
      "step": 215
    },
    {
      "epoch": 0.0024159182166944423,
      "grad_norm": 0.17389710247516632,
      "learning_rate": 0.000199997119737156,
      "loss": 1.6308,
      "step": 216
    },
    {
      "epoch": 0.0024271030232532126,
      "grad_norm": 0.14822594821453094,
      "learning_rate": 0.00019999709300645105,
      "loss": 1.6175,
      "step": 217
    },
    {
      "epoch": 0.0024382878298119834,
      "grad_norm": 0.12898804247379303,
      "learning_rate": 0.00019999706615228107,
      "loss": 1.6069,
      "step": 218
    },
    {
      "epoch": 0.002449472636370754,
      "grad_norm": 0.16721111536026,
      "learning_rate": 0.00019999703917464605,
      "loss": 1.6206,
      "step": 219
    },
    {
      "epoch": 0.0024606574429295244,
      "grad_norm": 0.08022027462720871,
      "learning_rate": 0.00019999701207354606,
      "loss": 1.6082,
      "step": 220
    },
    {
      "epoch": 0.002471842249488295,
      "grad_norm": 0.1281793862581253,
      "learning_rate": 0.0001999969848489811,
      "loss": 1.5928,
      "step": 221
    },
    {
      "epoch": 0.0024830270560470655,
      "grad_norm": 0.10240975767374039,
      "learning_rate": 0.00019999695750095117,
      "loss": 1.6058,
      "step": 222
    },
    {
      "epoch": 0.0024942118626058362,
      "grad_norm": 0.08368588238954544,
      "learning_rate": 0.00019999693002945642,
      "loss": 1.6193,
      "step": 223
    },
    {
      "epoch": 0.002505396669164607,
      "grad_norm": 0.09998615831136703,
      "learning_rate": 0.00019999690243449676,
      "loss": 1.6096,
      "step": 224
    },
    {
      "epoch": 0.0025165814757233773,
      "grad_norm": 0.08170673996210098,
      "learning_rate": 0.00019999687471607228,
      "loss": 1.5874,
      "step": 225
    },
    {
      "epoch": 0.002527766282282148,
      "grad_norm": 0.09015469253063202,
      "learning_rate": 0.000199996846874183,
      "loss": 1.5903,
      "step": 226
    },
    {
      "epoch": 0.0025389510888409184,
      "grad_norm": 0.07382986694574356,
      "learning_rate": 0.000199996818908829,
      "loss": 1.5958,
      "step": 227
    },
    {
      "epoch": 0.002550135895399689,
      "grad_norm": 0.06669515371322632,
      "learning_rate": 0.00019999679082001023,
      "loss": 1.5881,
      "step": 228
    },
    {
      "epoch": 0.0025613207019584594,
      "grad_norm": 0.0616774745285511,
      "learning_rate": 0.0001999967626077268,
      "loss": 1.6038,
      "step": 229
    },
    {
      "epoch": 0.00257250550851723,
      "grad_norm": 0.05451146885752678,
      "learning_rate": 0.00019999673427197872,
      "loss": 1.6021,
      "step": 230
    },
    {
      "epoch": 0.002583690315076001,
      "grad_norm": 0.05640149116516113,
      "learning_rate": 0.000199996705812766,
      "loss": 1.6022,
      "step": 231
    },
    {
      "epoch": 0.0025948751216347712,
      "grad_norm": 0.06033660098910332,
      "learning_rate": 0.00019999667723008871,
      "loss": 1.6002,
      "step": 232
    },
    {
      "epoch": 0.002606059928193542,
      "grad_norm": 0.07876270264387131,
      "learning_rate": 0.0001999966485239469,
      "loss": 1.6043,
      "step": 233
    },
    {
      "epoch": 0.0026172447347523123,
      "grad_norm": 0.0829700380563736,
      "learning_rate": 0.00019999661969434055,
      "loss": 1.5915,
      "step": 234
    },
    {
      "epoch": 0.002628429541311083,
      "grad_norm": 0.05654975026845932,
      "learning_rate": 0.0001999965907412697,
      "loss": 1.5893,
      "step": 235
    },
    {
      "epoch": 0.0026396143478698538,
      "grad_norm": 0.06751634925603867,
      "learning_rate": 0.00019999656166473444,
      "loss": 1.5757,
      "step": 236
    },
    {
      "epoch": 0.002650799154428624,
      "grad_norm": 0.06081743538379669,
      "learning_rate": 0.00019999653246473477,
      "loss": 1.5702,
      "step": 237
    },
    {
      "epoch": 0.002661983960987395,
      "grad_norm": 0.0666998103260994,
      "learning_rate": 0.00019999650314127075,
      "loss": 1.568,
      "step": 238
    },
    {
      "epoch": 0.002673168767546165,
      "grad_norm": 0.04934430122375488,
      "learning_rate": 0.00019999647369434235,
      "loss": 1.6017,
      "step": 239
    },
    {
      "epoch": 0.002684353574104936,
      "grad_norm": 0.0574209988117218,
      "learning_rate": 0.00019999644412394972,
      "loss": 1.5935,
      "step": 240
    },
    {
      "epoch": 0.002695538380663706,
      "grad_norm": 0.04870286583900452,
      "learning_rate": 0.00019999641443009278,
      "loss": 1.6035,
      "step": 241
    },
    {
      "epoch": 0.002706723187222477,
      "grad_norm": 0.04176439344882965,
      "learning_rate": 0.00019999638461277162,
      "loss": 1.598,
      "step": 242
    },
    {
      "epoch": 0.0027179079937812477,
      "grad_norm": 0.05534802004694939,
      "learning_rate": 0.0001999963546719863,
      "loss": 1.5945,
      "step": 243
    },
    {
      "epoch": 0.002729092800340018,
      "grad_norm": 0.04214160889387131,
      "learning_rate": 0.0001999963246077368,
      "loss": 1.6019,
      "step": 244
    },
    {
      "epoch": 0.0027402776068987888,
      "grad_norm": 0.04326852038502693,
      "learning_rate": 0.00019999629442002322,
      "loss": 1.604,
      "step": 245
    },
    {
      "epoch": 0.002751462413457559,
      "grad_norm": 0.04295732453465462,
      "learning_rate": 0.00019999626410884553,
      "loss": 1.6136,
      "step": 246
    },
    {
      "epoch": 0.00276264722001633,
      "grad_norm": 0.038508690893650055,
      "learning_rate": 0.00019999623367420385,
      "loss": 1.5904,
      "step": 247
    },
    {
      "epoch": 0.0027738320265751006,
      "grad_norm": 0.040281713008880615,
      "learning_rate": 0.0001999962031160981,
      "loss": 1.5955,
      "step": 248
    },
    {
      "epoch": 0.002785016833133871,
      "grad_norm": 0.041424721479415894,
      "learning_rate": 0.00019999617243452844,
      "loss": 1.608,
      "step": 249
    },
    {
      "epoch": 0.0027962016396926416,
      "grad_norm": 0.03804994374513626,
      "learning_rate": 0.00019999614162949484,
      "loss": 1.6125,
      "step": 250
    },
    {
      "epoch": 0.002807386446251412,
      "grad_norm": 0.04370785504579544,
      "learning_rate": 0.0001999961107009974,
      "loss": 1.6125,
      "step": 251
    },
    {
      "epoch": 0.0028185712528101827,
      "grad_norm": 0.047021038830280304,
      "learning_rate": 0.000199996079649036,
      "loss": 1.5931,
      "step": 252
    },
    {
      "epoch": 0.002829756059368953,
      "grad_norm": 0.036128588020801544,
      "learning_rate": 0.0001999960484736109,
      "loss": 1.5707,
      "step": 253
    },
    {
      "epoch": 0.0028409408659277238,
      "grad_norm": 0.04315561056137085,
      "learning_rate": 0.00019999601717472199,
      "loss": 1.5902,
      "step": 254
    },
    {
      "epoch": 0.0028521256724864945,
      "grad_norm": 0.04395722597837448,
      "learning_rate": 0.00019999598575236934,
      "loss": 1.5995,
      "step": 255
    },
    {
      "epoch": 0.002863310479045265,
      "grad_norm": 0.038929786533117294,
      "learning_rate": 0.000199995954206553,
      "loss": 1.5854,
      "step": 256
    },
    {
      "epoch": 0.0028744952856040356,
      "grad_norm": 0.041567280888557434,
      "learning_rate": 0.00019999592253727299,
      "loss": 1.5783,
      "step": 257
    },
    {
      "epoch": 0.002885680092162806,
      "grad_norm": 0.03894374892115593,
      "learning_rate": 0.0001999958907445294,
      "loss": 1.5846,
      "step": 258
    },
    {
      "epoch": 0.0028968648987215766,
      "grad_norm": 0.04269428178668022,
      "learning_rate": 0.00019999585882832222,
      "loss": 1.6023,
      "step": 259
    },
    {
      "epoch": 0.0029080497052803474,
      "grad_norm": 0.04121831804513931,
      "learning_rate": 0.00019999582678865147,
      "loss": 1.6051,
      "step": 260
    },
    {
      "epoch": 0.0029192345118391177,
      "grad_norm": 0.038076166063547134,
      "learning_rate": 0.00019999579462551728,
      "loss": 1.6136,
      "step": 261
    },
    {
      "epoch": 0.0029304193183978884,
      "grad_norm": 0.042008642107248306,
      "learning_rate": 0.0001999957623389196,
      "loss": 1.6063,
      "step": 262
    },
    {
      "epoch": 0.0029416041249566587,
      "grad_norm": 0.042438406497240067,
      "learning_rate": 0.0001999957299288585,
      "loss": 1.6033,
      "step": 263
    },
    {
      "epoch": 0.0029527889315154295,
      "grad_norm": 0.041119206696748734,
      "learning_rate": 0.000199995697395334,
      "loss": 1.6031,
      "step": 264
    },
    {
      "epoch": 0.0029639737380742,
      "grad_norm": 0.045258279889822006,
      "learning_rate": 0.00019999566473834622,
      "loss": 1.5853,
      "step": 265
    },
    {
      "epoch": 0.0029751585446329705,
      "grad_norm": 0.04734019562602043,
      "learning_rate": 0.0001999956319578951,
      "loss": 1.5921,
      "step": 266
    },
    {
      "epoch": 0.0029863433511917413,
      "grad_norm": 0.04389064759016037,
      "learning_rate": 0.00019999559905398072,
      "loss": 1.581,
      "step": 267
    },
    {
      "epoch": 0.0029975281577505116,
      "grad_norm": 0.04582642391324043,
      "learning_rate": 0.00019999556602660318,
      "loss": 1.5928,
      "step": 268
    },
    {
      "epoch": 0.0030087129643092824,
      "grad_norm": 0.04518941417336464,
      "learning_rate": 0.00019999553287576238,
      "loss": 1.5809,
      "step": 269
    },
    {
      "epoch": 0.0030198977708680527,
      "grad_norm": 0.04687381908297539,
      "learning_rate": 0.0001999954996014585,
      "loss": 1.5999,
      "step": 270
    },
    {
      "epoch": 0.0030310825774268234,
      "grad_norm": 0.04809357225894928,
      "learning_rate": 0.00019999546620369152,
      "loss": 1.5972,
      "step": 271
    },
    {
      "epoch": 0.003042267383985594,
      "grad_norm": 0.05907173454761505,
      "learning_rate": 0.0001999954326824615,
      "loss": 1.5862,
      "step": 272
    },
    {
      "epoch": 0.0030534521905443645,
      "grad_norm": 0.06583942472934723,
      "learning_rate": 0.00019999539903776842,
      "loss": 1.5846,
      "step": 273
    },
    {
      "epoch": 0.0030646369971031352,
      "grad_norm": 0.07557905465364456,
      "learning_rate": 0.0001999953652696124,
      "loss": 1.5962,
      "step": 274
    },
    {
      "epoch": 0.0030758218036619055,
      "grad_norm": 0.07007817178964615,
      "learning_rate": 0.00019999533137799347,
      "loss": 1.5951,
      "step": 275
    },
    {
      "epoch": 0.0030870066102206763,
      "grad_norm": 0.05711887776851654,
      "learning_rate": 0.00019999529736291162,
      "loss": 1.5932,
      "step": 276
    },
    {
      "epoch": 0.0030981914167794466,
      "grad_norm": 0.04450292885303497,
      "learning_rate": 0.00019999526322436696,
      "loss": 1.6071,
      "step": 277
    },
    {
      "epoch": 0.0031093762233382173,
      "grad_norm": 0.04260997474193573,
      "learning_rate": 0.00019999522896235947,
      "loss": 1.6038,
      "step": 278
    },
    {
      "epoch": 0.003120561029896988,
      "grad_norm": 0.05689796805381775,
      "learning_rate": 0.00019999519457688925,
      "loss": 1.5688,
      "step": 279
    },
    {
      "epoch": 0.0031317458364557584,
      "grad_norm": 0.06330379098653793,
      "learning_rate": 0.0001999951600679563,
      "loss": 1.5617,
      "step": 280
    },
    {
      "epoch": 0.003142930643014529,
      "grad_norm": 0.06195618584752083,
      "learning_rate": 0.00019999512543556066,
      "loss": 1.5602,
      "step": 281
    },
    {
      "epoch": 0.0031541154495732995,
      "grad_norm": 0.06677111238241196,
      "learning_rate": 0.0001999950906797024,
      "loss": 1.5659,
      "step": 282
    },
    {
      "epoch": 0.00316530025613207,
      "grad_norm": 0.05750421807169914,
      "learning_rate": 0.00019999505580038153,
      "loss": 1.5759,
      "step": 283
    },
    {
      "epoch": 0.003176485062690841,
      "grad_norm": 0.04907039552927017,
      "learning_rate": 0.00019999502079759817,
      "loss": 1.5833,
      "step": 284
    },
    {
      "epoch": 0.0031876698692496113,
      "grad_norm": 0.048877034336328506,
      "learning_rate": 0.00019999498567135223,
      "loss": 1.5836,
      "step": 285
    },
    {
      "epoch": 0.003198854675808382,
      "grad_norm": 0.05494236946105957,
      "learning_rate": 0.0001999949504216439,
      "loss": 1.5837,
      "step": 286
    },
    {
      "epoch": 0.0032100394823671523,
      "grad_norm": 0.04953937977552414,
      "learning_rate": 0.00019999491504847313,
      "loss": 1.5794,
      "step": 287
    },
    {
      "epoch": 0.003221224288925923,
      "grad_norm": 0.05240803211927414,
      "learning_rate": 0.00019999487955184,
      "loss": 1.5894,
      "step": 288
    },
    {
      "epoch": 0.0032324090954846934,
      "grad_norm": 0.05633338540792465,
      "learning_rate": 0.0001999948439317445,
      "loss": 1.5755,
      "step": 289
    },
    {
      "epoch": 0.003243593902043464,
      "grad_norm": 0.06563600897789001,
      "learning_rate": 0.00019999480818818675,
      "loss": 1.5912,
      "step": 290
    },
    {
      "epoch": 0.003254778708602235,
      "grad_norm": 0.05903002619743347,
      "learning_rate": 0.00019999477232116676,
      "loss": 1.5856,
      "step": 291
    },
    {
      "epoch": 0.003265963515161005,
      "grad_norm": 0.03582334890961647,
      "learning_rate": 0.00019999473633068457,
      "loss": 1.6079,
      "step": 292
    },
    {
      "epoch": 0.003277148321719776,
      "grad_norm": 0.05011364817619324,
      "learning_rate": 0.00019999470021674025,
      "loss": 1.5907,
      "step": 293
    },
    {
      "epoch": 0.0032883331282785463,
      "grad_norm": 0.0577118918299675,
      "learning_rate": 0.0001999946639793338,
      "loss": 1.5955,
      "step": 294
    },
    {
      "epoch": 0.003299517934837317,
      "grad_norm": 0.05170518904924393,
      "learning_rate": 0.00019999462761846528,
      "loss": 1.5884,
      "step": 295
    },
    {
      "epoch": 0.0033107027413960878,
      "grad_norm": 0.05011725425720215,
      "learning_rate": 0.00019999459113413475,
      "loss": 1.6046,
      "step": 296
    },
    {
      "epoch": 0.003321887547954858,
      "grad_norm": 0.05645633116364479,
      "learning_rate": 0.00019999455452634224,
      "loss": 1.596,
      "step": 297
    },
    {
      "epoch": 0.003333072354513629,
      "grad_norm": 0.05705921724438667,
      "learning_rate": 0.0001999945177950878,
      "loss": 1.6136,
      "step": 298
    },
    {
      "epoch": 0.003344257161072399,
      "grad_norm": 0.05761184170842171,
      "learning_rate": 0.0001999944809403715,
      "loss": 1.6095,
      "step": 299
    },
    {
      "epoch": 0.00335544196763117,
      "grad_norm": 0.0613851472735405,
      "learning_rate": 0.00019999444396219337,
      "loss": 1.6141,
      "step": 300
    },
    {
      "epoch": 0.00336662677418994,
      "grad_norm": 0.06220489367842674,
      "learning_rate": 0.00019999440686055344,
      "loss": 1.5988,
      "step": 301
    },
    {
      "epoch": 0.003377811580748711,
      "grad_norm": 0.062393296509981155,
      "learning_rate": 0.00019999436963545177,
      "loss": 1.6075,
      "step": 302
    },
    {
      "epoch": 0.0033889963873074817,
      "grad_norm": 0.0625912994146347,
      "learning_rate": 0.00019999433228688838,
      "loss": 1.6102,
      "step": 303
    },
    {
      "epoch": 0.003400181193866252,
      "grad_norm": 0.06049802899360657,
      "learning_rate": 0.00019999429481486335,
      "loss": 1.5968,
      "step": 304
    },
    {
      "epoch": 0.0034113660004250227,
      "grad_norm": 0.05767315998673439,
      "learning_rate": 0.00019999425721937674,
      "loss": 1.5906,
      "step": 305
    },
    {
      "epoch": 0.003422550806983793,
      "grad_norm": 0.049920015037059784,
      "learning_rate": 0.00019999421950042854,
      "loss": 1.5694,
      "step": 306
    },
    {
      "epoch": 0.003433735613542564,
      "grad_norm": 0.04852724075317383,
      "learning_rate": 0.0001999941816580188,
      "loss": 1.5609,
      "step": 307
    },
    {
      "epoch": 0.0034449204201013345,
      "grad_norm": 0.05249037966132164,
      "learning_rate": 0.00019999414369214767,
      "loss": 1.5675,
      "step": 308
    },
    {
      "epoch": 0.003456105226660105,
      "grad_norm": 0.05167357623577118,
      "learning_rate": 0.00019999410560281506,
      "loss": 1.5645,
      "step": 309
    },
    {
      "epoch": 0.0034672900332188756,
      "grad_norm": 0.05197747051715851,
      "learning_rate": 0.00019999406739002108,
      "loss": 1.5707,
      "step": 310
    },
    {
      "epoch": 0.003478474839777646,
      "grad_norm": 0.05140923336148262,
      "learning_rate": 0.00019999402905376582,
      "loss": 1.5745,
      "step": 311
    },
    {
      "epoch": 0.0034896596463364167,
      "grad_norm": 0.05357779935002327,
      "learning_rate": 0.00019999399059404923,
      "loss": 1.5824,
      "step": 312
    },
    {
      "epoch": 0.003500844452895187,
      "grad_norm": 0.04196302220225334,
      "learning_rate": 0.00019999395201087143,
      "loss": 1.5685,
      "step": 313
    },
    {
      "epoch": 0.0035120292594539577,
      "grad_norm": 0.04698769748210907,
      "learning_rate": 0.00019999391330423246,
      "loss": 1.5645,
      "step": 314
    },
    {
      "epoch": 0.0035232140660127285,
      "grad_norm": 0.055174414068460464,
      "learning_rate": 0.00019999387447413236,
      "loss": 1.5702,
      "step": 315
    },
    {
      "epoch": 0.003534398872571499,
      "grad_norm": 0.05560048297047615,
      "learning_rate": 0.00019999383552057114,
      "loss": 1.5746,
      "step": 316
    },
    {
      "epoch": 0.0035455836791302695,
      "grad_norm": 0.059730809181928635,
      "learning_rate": 0.0001999937964435489,
      "loss": 1.5698,
      "step": 317
    },
    {
      "epoch": 0.00355676848568904,
      "grad_norm": 0.06850636750459671,
      "learning_rate": 0.00019999375724306568,
      "loss": 1.5837,
      "step": 318
    },
    {
      "epoch": 0.0035679532922478106,
      "grad_norm": 0.0658111497759819,
      "learning_rate": 0.00019999371791912148,
      "loss": 1.5759,
      "step": 319
    },
    {
      "epoch": 0.0035791380988065813,
      "grad_norm": 0.05440279841423035,
      "learning_rate": 0.00019999367847171643,
      "loss": 1.5873,
      "step": 320
    },
    {
      "epoch": 0.0035903229053653517,
      "grad_norm": 0.06169600412249565,
      "learning_rate": 0.0001999936389008505,
      "loss": 1.5791,
      "step": 321
    },
    {
      "epoch": 0.0036015077119241224,
      "grad_norm": 0.06897033751010895,
      "learning_rate": 0.0001999935992065238,
      "loss": 1.5815,
      "step": 322
    },
    {
      "epoch": 0.0036126925184828927,
      "grad_norm": 0.06641620397567749,
      "learning_rate": 0.00019999355938873635,
      "loss": 1.5826,
      "step": 323
    },
    {
      "epoch": 0.0036238773250416635,
      "grad_norm": 0.057002220302820206,
      "learning_rate": 0.00019999351944748818,
      "loss": 1.5716,
      "step": 324
    },
    {
      "epoch": 0.0036350621316004338,
      "grad_norm": 0.06431427597999573,
      "learning_rate": 0.00019999347938277938,
      "loss": 1.5626,
      "step": 325
    },
    {
      "epoch": 0.0036462469381592045,
      "grad_norm": 0.06504250317811966,
      "learning_rate": 0.00019999343919460997,
      "loss": 1.574,
      "step": 326
    },
    {
      "epoch": 0.0036574317447179753,
      "grad_norm": 0.06940289586782455,
      "learning_rate": 0.00019999339888298004,
      "loss": 1.5676,
      "step": 327
    },
    {
      "epoch": 0.0036686165512767456,
      "grad_norm": 0.06492604315280914,
      "learning_rate": 0.00019999335844788957,
      "loss": 1.5676,
      "step": 328
    },
    {
      "epoch": 0.0036798013578355163,
      "grad_norm": 0.07069146633148193,
      "learning_rate": 0.0001999933178893387,
      "loss": 1.5601,
      "step": 329
    },
    {
      "epoch": 0.0036909861643942866,
      "grad_norm": 0.07502440363168716,
      "learning_rate": 0.00019999327720732736,
      "loss": 1.5651,
      "step": 330
    },
    {
      "epoch": 0.0037021709709530574,
      "grad_norm": 0.06407099217176437,
      "learning_rate": 0.00019999323640185573,
      "loss": 1.5818,
      "step": 331
    },
    {
      "epoch": 0.003713355777511828,
      "grad_norm": 0.05621904134750366,
      "learning_rate": 0.00019999319547292377,
      "loss": 1.5975,
      "step": 332
    },
    {
      "epoch": 0.0037245405840705985,
      "grad_norm": 0.053219810128211975,
      "learning_rate": 0.00019999315442053157,
      "loss": 1.5925,
      "step": 333
    },
    {
      "epoch": 0.003735725390629369,
      "grad_norm": 0.0538603812456131,
      "learning_rate": 0.00019999311324467919,
      "loss": 1.5784,
      "step": 334
    },
    {
      "epoch": 0.0037469101971881395,
      "grad_norm": 0.05239463597536087,
      "learning_rate": 0.00019999307194536664,
      "loss": 1.5782,
      "step": 335
    },
    {
      "epoch": 0.0037580950037469103,
      "grad_norm": 0.053746242076158524,
      "learning_rate": 0.00019999303052259398,
      "loss": 1.5802,
      "step": 336
    },
    {
      "epoch": 0.0037692798103056806,
      "grad_norm": 0.04721551761031151,
      "learning_rate": 0.0001999929889763613,
      "loss": 1.5739,
      "step": 337
    },
    {
      "epoch": 0.0037804646168644513,
      "grad_norm": 0.04483070224523544,
      "learning_rate": 0.00019999294730666862,
      "loss": 1.5823,
      "step": 338
    },
    {
      "epoch": 0.003791649423423222,
      "grad_norm": 0.05224015936255455,
      "learning_rate": 0.000199992905513516,
      "loss": 1.5559,
      "step": 339
    },
    {
      "epoch": 0.0038028342299819924,
      "grad_norm": 0.05772995948791504,
      "learning_rate": 0.0001999928635969035,
      "loss": 1.5574,
      "step": 340
    },
    {
      "epoch": 0.003814019036540763,
      "grad_norm": 0.059287503361701965,
      "learning_rate": 0.00019999282155683116,
      "loss": 1.5694,
      "step": 341
    },
    {
      "epoch": 0.0038252038430995334,
      "grad_norm": 0.050815433263778687,
      "learning_rate": 0.00019999277939329902,
      "loss": 1.5655,
      "step": 342
    },
    {
      "epoch": 0.003836388649658304,
      "grad_norm": 0.047384679317474365,
      "learning_rate": 0.00019999273710630714,
      "loss": 1.5704,
      "step": 343
    },
    {
      "epoch": 0.003847573456217075,
      "grad_norm": 0.04918666183948517,
      "learning_rate": 0.00019999269469585555,
      "loss": 1.5694,
      "step": 344
    },
    {
      "epoch": 0.0038587582627758452,
      "grad_norm": 0.058182474225759506,
      "learning_rate": 0.0001999926521619444,
      "loss": 1.5971,
      "step": 345
    },
    {
      "epoch": 0.003869943069334616,
      "grad_norm": 0.07194909453392029,
      "learning_rate": 0.00019999260950457362,
      "loss": 1.6033,
      "step": 346
    },
    {
      "epoch": 0.0038811278758933863,
      "grad_norm": 0.08051355183124542,
      "learning_rate": 0.0001999925667237433,
      "loss": 1.5969,
      "step": 347
    },
    {
      "epoch": 0.003892312682452157,
      "grad_norm": 0.09640984982252121,
      "learning_rate": 0.00019999252381945357,
      "loss": 1.6414,
      "step": 348
    },
    {
      "epoch": 0.0039034974890109274,
      "grad_norm": 0.11357203125953674,
      "learning_rate": 0.00019999248079170437,
      "loss": 1.6522,
      "step": 349
    },
    {
      "epoch": 0.003914682295569698,
      "grad_norm": 0.14623580873012543,
      "learning_rate": 0.00019999243764049586,
      "loss": 1.6482,
      "step": 350
    },
    {
      "epoch": 0.003925867102128468,
      "grad_norm": 0.158810093998909,
      "learning_rate": 0.00019999239436582796,
      "loss": 1.5814,
      "step": 351
    },
    {
      "epoch": 0.00393705190868724,
      "grad_norm": 0.11811669170856476,
      "learning_rate": 0.00019999235096770086,
      "loss": 1.5782,
      "step": 352
    },
    {
      "epoch": 0.00394823671524601,
      "grad_norm": 0.09518411755561829,
      "learning_rate": 0.0001999923074461145,
      "loss": 1.5852,
      "step": 353
    },
    {
      "epoch": 0.00395942152180478,
      "grad_norm": 0.1165471076965332,
      "learning_rate": 0.00019999226380106906,
      "loss": 1.5766,
      "step": 354
    },
    {
      "epoch": 0.0039706063283635505,
      "grad_norm": 0.09517768025398254,
      "learning_rate": 0.00019999222003256448,
      "loss": 1.5829,
      "step": 355
    },
    {
      "epoch": 0.003981791134922322,
      "grad_norm": 0.06006886065006256,
      "learning_rate": 0.00019999217614060085,
      "loss": 1.5708,
      "step": 356
    },
    {
      "epoch": 0.003992975941481092,
      "grad_norm": 0.08933499455451965,
      "learning_rate": 0.00019999213212517825,
      "loss": 1.5718,
      "step": 357
    },
    {
      "epoch": 0.004004160748039862,
      "grad_norm": 0.0867113471031189,
      "learning_rate": 0.0001999920879862967,
      "loss": 1.5711,
      "step": 358
    },
    {
      "epoch": 0.0040153455545986335,
      "grad_norm": 0.06603030860424042,
      "learning_rate": 0.00019999204372395628,
      "loss": 1.5778,
      "step": 359
    },
    {
      "epoch": 0.004026530361157404,
      "grad_norm": 0.07514684647321701,
      "learning_rate": 0.00019999199933815702,
      "loss": 1.5738,
      "step": 360
    },
    {
      "epoch": 0.004037715167716174,
      "grad_norm": 0.060728590935468674,
      "learning_rate": 0.00019999195482889897,
      "loss": 1.5694,
      "step": 361
    },
    {
      "epoch": 0.004048899974274945,
      "grad_norm": 0.06927715986967087,
      "learning_rate": 0.00019999191019618224,
      "loss": 1.5664,
      "step": 362
    },
    {
      "epoch": 0.004060084780833716,
      "grad_norm": 0.0576176755130291,
      "learning_rate": 0.00019999186544000685,
      "loss": 1.5704,
      "step": 363
    },
    {
      "epoch": 0.004071269587392486,
      "grad_norm": 0.047579534351825714,
      "learning_rate": 0.00019999182056037285,
      "loss": 1.5732,
      "step": 364
    },
    {
      "epoch": 0.004082454393951256,
      "grad_norm": 0.06114533543586731,
      "learning_rate": 0.00019999177555728027,
      "loss": 1.5776,
      "step": 365
    },
    {
      "epoch": 0.0040936392005100275,
      "grad_norm": 0.05183887854218483,
      "learning_rate": 0.0001999917304307292,
      "loss": 1.5697,
      "step": 366
    },
    {
      "epoch": 0.004104824007068798,
      "grad_norm": 0.05595005676150322,
      "learning_rate": 0.0001999916851807197,
      "loss": 1.5611,
      "step": 367
    },
    {
      "epoch": 0.004116008813627568,
      "grad_norm": 0.04884869232773781,
      "learning_rate": 0.00019999163980725183,
      "loss": 1.5754,
      "step": 368
    },
    {
      "epoch": 0.004127193620186339,
      "grad_norm": 0.050160013139247894,
      "learning_rate": 0.00019999159431032562,
      "loss": 1.5818,
      "step": 369
    },
    {
      "epoch": 0.00413837842674511,
      "grad_norm": 0.0458194725215435,
      "learning_rate": 0.00019999154868994111,
      "loss": 1.5821,
      "step": 370
    },
    {
      "epoch": 0.00414956323330388,
      "grad_norm": 0.04877634719014168,
      "learning_rate": 0.00019999150294609845,
      "loss": 1.5828,
      "step": 371
    },
    {
      "epoch": 0.00416074803986265,
      "grad_norm": 0.045320361852645874,
      "learning_rate": 0.00019999145707879758,
      "loss": 1.5946,
      "step": 372
    },
    {
      "epoch": 0.004171932846421421,
      "grad_norm": 0.05053291842341423,
      "learning_rate": 0.00019999141108803864,
      "loss": 1.6002,
      "step": 373
    },
    {
      "epoch": 0.004183117652980192,
      "grad_norm": 0.053211431950330734,
      "learning_rate": 0.0001999913649738216,
      "loss": 1.5914,
      "step": 374
    },
    {
      "epoch": 0.004194302459538962,
      "grad_norm": 0.045671332627534866,
      "learning_rate": 0.00019999131873614664,
      "loss": 1.5806,
      "step": 375
    },
    {
      "epoch": 0.004205487266097733,
      "grad_norm": 0.051272232085466385,
      "learning_rate": 0.0001999912723750137,
      "loss": 1.5898,
      "step": 376
    },
    {
      "epoch": 0.0042166720726565035,
      "grad_norm": 0.05297670140862465,
      "learning_rate": 0.0001999912258904229,
      "loss": 1.6003,
      "step": 377
    },
    {
      "epoch": 0.004227856879215274,
      "grad_norm": 0.044414643198251724,
      "learning_rate": 0.00019999117928237427,
      "loss": 1.6069,
      "step": 378
    },
    {
      "epoch": 0.004239041685774044,
      "grad_norm": 0.04553841054439545,
      "learning_rate": 0.0001999911325508679,
      "loss": 1.5798,
      "step": 379
    },
    {
      "epoch": 0.004250226492332815,
      "grad_norm": 0.05364730581641197,
      "learning_rate": 0.00019999108569590383,
      "loss": 1.5856,
      "step": 380
    },
    {
      "epoch": 0.004261411298891586,
      "grad_norm": 0.05173739790916443,
      "learning_rate": 0.0001999910387174821,
      "loss": 1.5764,
      "step": 381
    },
    {
      "epoch": 0.004272596105450356,
      "grad_norm": 0.05577515438199043,
      "learning_rate": 0.00019999099161560282,
      "loss": 1.5855,
      "step": 382
    },
    {
      "epoch": 0.004283780912009127,
      "grad_norm": 0.057436104863882065,
      "learning_rate": 0.00019999094439026598,
      "loss": 1.5785,
      "step": 383
    },
    {
      "epoch": 0.0042949657185678974,
      "grad_norm": 0.03927776962518692,
      "learning_rate": 0.00019999089704147166,
      "loss": 1.5735,
      "step": 384
    },
    {
      "epoch": 0.004306150525126668,
      "grad_norm": 0.04739474132657051,
      "learning_rate": 0.00019999084956921997,
      "loss": 1.5805,
      "step": 385
    },
    {
      "epoch": 0.004317335331685439,
      "grad_norm": 0.04832485690712929,
      "learning_rate": 0.0001999908019735109,
      "loss": 1.6164,
      "step": 386
    },
    {
      "epoch": 0.004328520138244209,
      "grad_norm": 0.049625612795352936,
      "learning_rate": 0.00019999075425434452,
      "loss": 1.6468,
      "step": 387
    },
    {
      "epoch": 0.0043397049448029796,
      "grad_norm": 0.04835371673107147,
      "learning_rate": 0.00019999070641172094,
      "loss": 1.6438,
      "step": 388
    },
    {
      "epoch": 0.00435088975136175,
      "grad_norm": 0.05029625818133354,
      "learning_rate": 0.00019999065844564018,
      "loss": 1.6103,
      "step": 389
    },
    {
      "epoch": 0.004362074557920521,
      "grad_norm": 0.055567558854818344,
      "learning_rate": 0.0001999906103561023,
      "loss": 1.602,
      "step": 390
    },
    {
      "epoch": 0.004373259364479291,
      "grad_norm": 0.06654093414545059,
      "learning_rate": 0.00019999056214310733,
      "loss": 1.5815,
      "step": 391
    },
    {
      "epoch": 0.004384444171038062,
      "grad_norm": 0.06477198004722595,
      "learning_rate": 0.0001999905138066554,
      "loss": 1.5883,
      "step": 392
    },
    {
      "epoch": 0.004395628977596833,
      "grad_norm": 0.0623844638466835,
      "learning_rate": 0.00019999046534674656,
      "loss": 1.5877,
      "step": 393
    },
    {
      "epoch": 0.004406813784155603,
      "grad_norm": 0.059734586626291275,
      "learning_rate": 0.00019999041676338077,
      "loss": 1.5855,
      "step": 394
    },
    {
      "epoch": 0.0044179985907143735,
      "grad_norm": 0.05187408998608589,
      "learning_rate": 0.0001999903680565582,
      "loss": 1.5827,
      "step": 395
    },
    {
      "epoch": 0.004429183397273144,
      "grad_norm": 0.05175703763961792,
      "learning_rate": 0.00019999031922627886,
      "loss": 1.5758,
      "step": 396
    },
    {
      "epoch": 0.004440368203831915,
      "grad_norm": 0.05059249326586723,
      "learning_rate": 0.00019999027027254286,
      "loss": 1.5758,
      "step": 397
    },
    {
      "epoch": 0.004451553010390685,
      "grad_norm": 0.050220511853694916,
      "learning_rate": 0.0001999902211953502,
      "loss": 1.5771,
      "step": 398
    },
    {
      "epoch": 0.004462737816949456,
      "grad_norm": 0.057194001972675323,
      "learning_rate": 0.00019999017199470094,
      "loss": 1.5828,
      "step": 399
    },
    {
      "epoch": 0.004473922623508227,
      "grad_norm": 0.07026943564414978,
      "learning_rate": 0.00019999012267059519,
      "loss": 1.5728,
      "step": 400
    },
    {
      "epoch": 0.004485107430066997,
      "grad_norm": 0.08094791322946548,
      "learning_rate": 0.00019999007322303296,
      "loss": 1.5634,
      "step": 401
    },
    {
      "epoch": 0.004496292236625767,
      "grad_norm": 0.08370808511972427,
      "learning_rate": 0.0001999900236520144,
      "loss": 1.5747,
      "step": 402
    },
    {
      "epoch": 0.004507477043184538,
      "grad_norm": 0.09409447014331818,
      "learning_rate": 0.00019998997395753945,
      "loss": 1.5703,
      "step": 403
    },
    {
      "epoch": 0.004518661849743309,
      "grad_norm": 0.09207552671432495,
      "learning_rate": 0.0001999899241396082,
      "loss": 1.5628,
      "step": 404
    },
    {
      "epoch": 0.004529846656302079,
      "grad_norm": 0.07077619433403015,
      "learning_rate": 0.0001999898741982208,
      "loss": 1.5826,
      "step": 405
    },
    {
      "epoch": 0.0045410314628608495,
      "grad_norm": 0.06743451207876205,
      "learning_rate": 0.00019998982413337724,
      "loss": 1.6047,
      "step": 406
    },
    {
      "epoch": 0.004552216269419621,
      "grad_norm": 0.10414531826972961,
      "learning_rate": 0.0001999897739450776,
      "loss": 1.6005,
      "step": 407
    },
    {
      "epoch": 0.004563401075978391,
      "grad_norm": 0.10947459191083908,
      "learning_rate": 0.0001999897236333219,
      "loss": 1.6025,
      "step": 408
    },
    {
      "epoch": 0.004574585882537161,
      "grad_norm": 0.06659277528524399,
      "learning_rate": 0.00019998967319811027,
      "loss": 1.6059,
      "step": 409
    },
    {
      "epoch": 0.004585770689095932,
      "grad_norm": 0.06038579344749451,
      "learning_rate": 0.00019998962263944274,
      "loss": 1.5826,
      "step": 410
    },
    {
      "epoch": 0.004596955495654703,
      "grad_norm": 0.08695773035287857,
      "learning_rate": 0.00019998957195731934,
      "loss": 1.5779,
      "step": 411
    },
    {
      "epoch": 0.004608140302213473,
      "grad_norm": 0.07446157187223434,
      "learning_rate": 0.0001999895211517402,
      "loss": 1.5812,
      "step": 412
    },
    {
      "epoch": 0.0046193251087722435,
      "grad_norm": 0.04878260940313339,
      "learning_rate": 0.00019998947022270534,
      "loss": 1.6033,
      "step": 413
    },
    {
      "epoch": 0.004630509915331015,
      "grad_norm": 0.06283459812402725,
      "learning_rate": 0.00019998941917021484,
      "loss": 1.5879,
      "step": 414
    },
    {
      "epoch": 0.004641694721889785,
      "grad_norm": 0.05891675129532814,
      "learning_rate": 0.00019998936799426874,
      "loss": 1.59,
      "step": 415
    },
    {
      "epoch": 0.004652879528448555,
      "grad_norm": 0.04745139181613922,
      "learning_rate": 0.0001999893166948671,
      "loss": 1.5981,
      "step": 416
    },
    {
      "epoch": 0.0046640643350073265,
      "grad_norm": 0.05297010764479637,
      "learning_rate": 0.00019998926527201003,
      "loss": 1.5937,
      "step": 417
    },
    {
      "epoch": 0.004675249141566097,
      "grad_norm": 0.05115678906440735,
      "learning_rate": 0.00019998921372569757,
      "loss": 1.5941,
      "step": 418
    },
    {
      "epoch": 0.004686433948124867,
      "grad_norm": 0.05678752437233925,
      "learning_rate": 0.00019998916205592974,
      "loss": 1.5631,
      "step": 419
    },
    {
      "epoch": 0.004697618754683637,
      "grad_norm": 0.05227034166455269,
      "learning_rate": 0.00019998911026270668,
      "loss": 1.5755,
      "step": 420
    },
    {
      "epoch": 0.004708803561242409,
      "grad_norm": 0.05871938541531563,
      "learning_rate": 0.0001999890583460284,
      "loss": 1.5843,
      "step": 421
    },
    {
      "epoch": 0.004719988367801179,
      "grad_norm": 0.06751812249422073,
      "learning_rate": 0.00019998900630589493,
      "loss": 1.5922,
      "step": 422
    },
    {
      "epoch": 0.004731173174359949,
      "grad_norm": 0.061179131269454956,
      "learning_rate": 0.00019998895414230646,
      "loss": 1.5939,
      "step": 423
    },
    {
      "epoch": 0.00474235798091872,
      "grad_norm": 0.06404510140419006,
      "learning_rate": 0.00019998890185526292,
      "loss": 1.5857,
      "step": 424
    },
    {
      "epoch": 0.004753542787477491,
      "grad_norm": 0.07117751985788345,
      "learning_rate": 0.0001999888494447645,
      "loss": 1.5676,
      "step": 425
    },
    {
      "epoch": 0.004764727594036261,
      "grad_norm": 0.06996233761310577,
      "learning_rate": 0.00019998879691081114,
      "loss": 1.5769,
      "step": 426
    },
    {
      "epoch": 0.004775912400595031,
      "grad_norm": 0.0711674690246582,
      "learning_rate": 0.00019998874425340298,
      "loss": 1.5622,
      "step": 427
    },
    {
      "epoch": 0.0047870972071538025,
      "grad_norm": 0.079228475689888,
      "learning_rate": 0.0001999886914725401,
      "loss": 1.571,
      "step": 428
    },
    {
      "epoch": 0.004798282013712573,
      "grad_norm": 0.08016793429851532,
      "learning_rate": 0.00019998863856822248,
      "loss": 1.6029,
      "step": 429
    },
    {
      "epoch": 0.004809466820271343,
      "grad_norm": 0.0795593336224556,
      "learning_rate": 0.00019998858554045026,
      "loss": 1.6062,
      "step": 430
    },
    {
      "epoch": 0.004820651626830114,
      "grad_norm": 0.06341350823640823,
      "learning_rate": 0.00019998853238922348,
      "loss": 1.5884,
      "step": 431
    },
    {
      "epoch": 0.004831836433388885,
      "grad_norm": 0.05454142764210701,
      "learning_rate": 0.00019998847911454219,
      "loss": 1.5921,
      "step": 432
    },
    {
      "epoch": 0.004843021239947655,
      "grad_norm": 0.0625983327627182,
      "learning_rate": 0.0001999884257164065,
      "loss": 1.582,
      "step": 433
    },
    {
      "epoch": 0.004854206046506425,
      "grad_norm": 0.06116988882422447,
      "learning_rate": 0.00019998837219481645,
      "loss": 1.5804,
      "step": 434
    },
    {
      "epoch": 0.004865390853065196,
      "grad_norm": 0.057872429490089417,
      "learning_rate": 0.0001999883185497721,
      "loss": 1.5721,
      "step": 435
    },
    {
      "epoch": 0.004876575659623967,
      "grad_norm": 0.04977230727672577,
      "learning_rate": 0.00019998826478127352,
      "loss": 1.5539,
      "step": 436
    },
    {
      "epoch": 0.004887760466182737,
      "grad_norm": 0.06137122958898544,
      "learning_rate": 0.00019998821088932077,
      "loss": 1.5715,
      "step": 437
    },
    {
      "epoch": 0.004898945272741508,
      "grad_norm": 0.05595165491104126,
      "learning_rate": 0.00019998815687391396,
      "loss": 1.5841,
      "step": 438
    },
    {
      "epoch": 0.0049101300793002785,
      "grad_norm": 0.049001295119524,
      "learning_rate": 0.00019998810273505311,
      "loss": 1.571,
      "step": 439
    },
    {
      "epoch": 0.004921314885859049,
      "grad_norm": 0.04792420566082001,
      "learning_rate": 0.00019998804847273828,
      "loss": 1.5695,
      "step": 440
    },
    {
      "epoch": 0.00493249969241782,
      "grad_norm": 0.052222106605768204,
      "learning_rate": 0.00019998799408696956,
      "loss": 1.5671,
      "step": 441
    },
    {
      "epoch": 0.00494368449897659,
      "grad_norm": 0.05545085668563843,
      "learning_rate": 0.00019998793957774703,
      "loss": 1.5679,
      "step": 442
    },
    {
      "epoch": 0.004954869305535361,
      "grad_norm": 0.06147260218858719,
      "learning_rate": 0.00019998788494507075,
      "loss": 1.5746,
      "step": 443
    },
    {
      "epoch": 0.004966054112094131,
      "grad_norm": 0.06299655884504318,
      "learning_rate": 0.00019998783018894073,
      "loss": 1.5719,
      "step": 444
    },
    {
      "epoch": 0.004977238918652902,
      "grad_norm": 0.05477641522884369,
      "learning_rate": 0.00019998777530935713,
      "loss": 1.5933,
      "step": 445
    },
    {
      "epoch": 0.0049884237252116725,
      "grad_norm": 0.054924603551626205,
      "learning_rate": 0.00019998772030631993,
      "loss": 1.6381,
      "step": 446
    },
    {
      "epoch": 0.004999608531770443,
      "grad_norm": 0.05982334539294243,
      "learning_rate": 0.0001999876651798293,
      "loss": 1.6049,
      "step": 447
    },
    {
      "epoch": 0.005010793338329214,
      "grad_norm": 0.07177302241325378,
      "learning_rate": 0.0001999876099298852,
      "loss": 1.5885,
      "step": 448
    },
    {
      "epoch": 0.005021978144887984,
      "grad_norm": 0.06020566448569298,
      "learning_rate": 0.00019998755455648778,
      "loss": 1.5918,
      "step": 449
    },
    {
      "epoch": 0.005033162951446755,
      "grad_norm": 0.0725252702832222,
      "learning_rate": 0.00019998749905963706,
      "loss": 1.5948,
      "step": 450
    },
    {
      "epoch": 0.005044347758005525,
      "grad_norm": 0.07799220085144043,
      "learning_rate": 0.00019998744343933313,
      "loss": 1.5903,
      "step": 451
    },
    {
      "epoch": 0.005055532564564296,
      "grad_norm": 0.06732252240180969,
      "learning_rate": 0.00019998738769557605,
      "loss": 1.6123,
      "step": 452
    },
    {
      "epoch": 0.005066717371123066,
      "grad_norm": 0.056653380393981934,
      "learning_rate": 0.0001999873318283659,
      "loss": 1.5976,
      "step": 453
    },
    {
      "epoch": 0.005077902177681837,
      "grad_norm": 0.06148442253470421,
      "learning_rate": 0.00019998727583770274,
      "loss": 1.5826,
      "step": 454
    },
    {
      "epoch": 0.005089086984240608,
      "grad_norm": 0.0657142624258995,
      "learning_rate": 0.00019998721972358662,
      "loss": 1.5504,
      "step": 455
    },
    {
      "epoch": 0.005100271790799378,
      "grad_norm": 0.06259225308895111,
      "learning_rate": 0.00019998716348601766,
      "loss": 1.552,
      "step": 456
    },
    {
      "epoch": 0.0051114565973581485,
      "grad_norm": 0.08781653642654419,
      "learning_rate": 0.00019998710712499585,
      "loss": 1.5523,
      "step": 457
    },
    {
      "epoch": 0.005122641403916919,
      "grad_norm": 0.0888068750500679,
      "learning_rate": 0.00019998705064052137,
      "loss": 1.5581,
      "step": 458
    },
    {
      "epoch": 0.00513382621047569,
      "grad_norm": 0.11727220565080643,
      "learning_rate": 0.0001999869940325942,
      "loss": 1.5714,
      "step": 459
    },
    {
      "epoch": 0.00514501101703446,
      "grad_norm": 0.10440998524427414,
      "learning_rate": 0.00019998693730121443,
      "loss": 1.5567,
      "step": 460
    },
    {
      "epoch": 0.005156195823593231,
      "grad_norm": 0.0791282206773758,
      "learning_rate": 0.00019998688044638215,
      "loss": 1.5605,
      "step": 461
    },
    {
      "epoch": 0.005167380630152002,
      "grad_norm": 0.07670507580041885,
      "learning_rate": 0.0001999868234680974,
      "loss": 1.5676,
      "step": 462
    },
    {
      "epoch": 0.005178565436710772,
      "grad_norm": 0.09401030838489532,
      "learning_rate": 0.0001999867663663603,
      "loss": 1.555,
      "step": 463
    },
    {
      "epoch": 0.0051897502432695424,
      "grad_norm": 0.08279041945934296,
      "learning_rate": 0.00019998670914117087,
      "loss": 1.5436,
      "step": 464
    },
    {
      "epoch": 0.005200935049828314,
      "grad_norm": 0.17070412635803223,
      "learning_rate": 0.0001999866517925292,
      "loss": 1.5575,
      "step": 465
    },
    {
      "epoch": 0.005212119856387084,
      "grad_norm": 0.09067776054143906,
      "learning_rate": 0.00019998659432043537,
      "loss": 1.5665,
      "step": 466
    },
    {
      "epoch": 0.005223304662945854,
      "grad_norm": 0.08899036049842834,
      "learning_rate": 0.00019998653672488942,
      "loss": 1.5678,
      "step": 467
    },
    {
      "epoch": 0.005234489469504625,
      "grad_norm": 0.06300584226846695,
      "learning_rate": 0.00019998647900589144,
      "loss": 1.5826,
      "step": 468
    },
    {
      "epoch": 0.005245674276063396,
      "grad_norm": 0.07066696137189865,
      "learning_rate": 0.00019998642116344156,
      "loss": 1.5696,
      "step": 469
    },
    {
      "epoch": 0.005256859082622166,
      "grad_norm": 0.0663958340883255,
      "learning_rate": 0.00019998636319753973,
      "loss": 1.5878,
      "step": 470
    },
    {
      "epoch": 0.005268043889180936,
      "grad_norm": 0.06905809789896011,
      "learning_rate": 0.00019998630510818612,
      "loss": 1.5667,
      "step": 471
    },
    {
      "epoch": 0.0052792286957397076,
      "grad_norm": 0.05643589049577713,
      "learning_rate": 0.00019998624689538077,
      "loss": 1.5742,
      "step": 472
    },
    {
      "epoch": 0.005290413502298478,
      "grad_norm": 0.05323821306228638,
      "learning_rate": 0.00019998618855912375,
      "loss": 1.5965,
      "step": 473
    },
    {
      "epoch": 0.005301598308857248,
      "grad_norm": 0.07279177010059357,
      "learning_rate": 0.0001999861300994151,
      "loss": 1.5954,
      "step": 474
    },
    {
      "epoch": 0.0053127831154160185,
      "grad_norm": 0.06261339038610458,
      "learning_rate": 0.00019998607151625497,
      "loss": 1.5771,
      "step": 475
    },
    {
      "epoch": 0.00532396792197479,
      "grad_norm": 0.0605684369802475,
      "learning_rate": 0.00019998601280964335,
      "loss": 1.5674,
      "step": 476
    },
    {
      "epoch": 0.00533515272853356,
      "grad_norm": 0.05855708196759224,
      "learning_rate": 0.0001999859539795804,
      "loss": 1.5641,
      "step": 477
    },
    {
      "epoch": 0.00534633753509233,
      "grad_norm": 0.04459947720170021,
      "learning_rate": 0.0001999858950260661,
      "loss": 1.5706,
      "step": 478
    },
    {
      "epoch": 0.0053575223416511015,
      "grad_norm": 0.05174221470952034,
      "learning_rate": 0.00019998583594910057,
      "loss": 1.5525,
      "step": 479
    },
    {
      "epoch": 0.005368707148209872,
      "grad_norm": 0.047726552933454514,
      "learning_rate": 0.0001999857767486839,
      "loss": 1.5613,
      "step": 480
    },
    {
      "epoch": 0.005379891954768642,
      "grad_norm": 0.056866295635700226,
      "learning_rate": 0.0001999857174248161,
      "loss": 1.5851,
      "step": 481
    },
    {
      "epoch": 0.005391076761327412,
      "grad_norm": 0.05624596029520035,
      "learning_rate": 0.00019998565797749732,
      "loss": 1.5862,
      "step": 482
    },
    {
      "epoch": 0.005402261567886184,
      "grad_norm": 0.057655058801174164,
      "learning_rate": 0.0001999855984067276,
      "loss": 1.5848,
      "step": 483
    },
    {
      "epoch": 0.005413446374444954,
      "grad_norm": 0.06511086970567703,
      "learning_rate": 0.000199985538712507,
      "loss": 1.6081,
      "step": 484
    },
    {
      "epoch": 0.005424631181003724,
      "grad_norm": 0.0913616269826889,
      "learning_rate": 0.0001999854788948356,
      "loss": 1.5973,
      "step": 485
    },
    {
      "epoch": 0.005435815987562495,
      "grad_norm": 0.11560064554214478,
      "learning_rate": 0.00019998541895371345,
      "loss": 1.5767,
      "step": 486
    },
    {
      "epoch": 0.005447000794121266,
      "grad_norm": 0.10205356776714325,
      "learning_rate": 0.00019998535888914073,
      "loss": 1.5641,
      "step": 487
    },
    {
      "epoch": 0.005458185600680036,
      "grad_norm": 0.06806248426437378,
      "learning_rate": 0.00019998529870111735,
      "loss": 1.5586,
      "step": 488
    },
    {
      "epoch": 0.005469370407238807,
      "grad_norm": 0.06459362804889679,
      "learning_rate": 0.00019998523838964355,
      "loss": 1.5532,
      "step": 489
    },
    {
      "epoch": 0.0054805552137975775,
      "grad_norm": 0.07120149582624435,
      "learning_rate": 0.00019998517795471928,
      "loss": 1.5515,
      "step": 490
    },
    {
      "epoch": 0.005491740020356348,
      "grad_norm": 0.059738751500844955,
      "learning_rate": 0.00019998511739634464,
      "loss": 1.5567,
      "step": 491
    },
    {
      "epoch": 0.005502924826915118,
      "grad_norm": 0.06517786532640457,
      "learning_rate": 0.00019998505671451976,
      "loss": 1.595,
      "step": 492
    },
    {
      "epoch": 0.005514109633473889,
      "grad_norm": 0.06841952353715897,
      "learning_rate": 0.0001999849959092447,
      "loss": 1.5805,
      "step": 493
    },
    {
      "epoch": 0.00552529444003266,
      "grad_norm": 0.06286690384149551,
      "learning_rate": 0.0001999849349805195,
      "loss": 1.571,
      "step": 494
    },
    {
      "epoch": 0.00553647924659143,
      "grad_norm": 0.0559217631816864,
      "learning_rate": 0.00019998487392834422,
      "loss": 1.5569,
      "step": 495
    },
    {
      "epoch": 0.005547664053150201,
      "grad_norm": 0.0687541738152504,
      "learning_rate": 0.000199984812752719,
      "loss": 1.557,
      "step": 496
    },
    {
      "epoch": 0.0055588488597089715,
      "grad_norm": 0.06690400838851929,
      "learning_rate": 0.00019998475145364383,
      "loss": 1.567,
      "step": 497
    },
    {
      "epoch": 0.005570033666267742,
      "grad_norm": 0.06469254940748215,
      "learning_rate": 0.00019998469003111892,
      "loss": 1.5783,
      "step": 498
    },
    {
      "epoch": 0.005581218472826512,
      "grad_norm": 0.06279771029949188,
      "learning_rate": 0.0001999846284851442,
      "loss": 1.588,
      "step": 499
    },
    {
      "epoch": 0.005592403279385283,
      "grad_norm": 0.05500609427690506,
      "learning_rate": 0.00019998456681571982,
      "loss": 1.5775,
      "step": 500
    },
    {
      "epoch": 0.005603588085944054,
      "grad_norm": 0.05660603567957878,
      "learning_rate": 0.00019998450502284584,
      "loss": 1.5847,
      "step": 501
    },
    {
      "epoch": 0.005614772892502824,
      "grad_norm": 0.06350179761648178,
      "learning_rate": 0.00019998444310652237,
      "loss": 1.5859,
      "step": 502
    },
    {
      "epoch": 0.005625957699061595,
      "grad_norm": 0.06947285681962967,
      "learning_rate": 0.00019998438106674945,
      "loss": 1.6142,
      "step": 503
    },
    {
      "epoch": 0.005637142505620365,
      "grad_norm": 0.08556090295314789,
      "learning_rate": 0.00019998431890352712,
      "loss": 1.5949,
      "step": 504
    },
    {
      "epoch": 0.005648327312179136,
      "grad_norm": 0.10163229703903198,
      "learning_rate": 0.00019998425661685553,
      "loss": 1.5819,
      "step": 505
    },
    {
      "epoch": 0.005659512118737906,
      "grad_norm": 0.11865612119436264,
      "learning_rate": 0.00019998419420673476,
      "loss": 1.5799,
      "step": 506
    },
    {
      "epoch": 0.005670696925296677,
      "grad_norm": 0.12710048258304596,
      "learning_rate": 0.0001999841316731648,
      "loss": 1.5837,
      "step": 507
    },
    {
      "epoch": 0.0056818817318554475,
      "grad_norm": 0.10893180966377258,
      "learning_rate": 0.00019998406901614583,
      "loss": 1.5547,
      "step": 508
    },
    {
      "epoch": 0.005693066538414218,
      "grad_norm": 0.06662983447313309,
      "learning_rate": 0.00019998400623567788,
      "loss": 1.5412,
      "step": 509
    },
    {
      "epoch": 0.005704251344972989,
      "grad_norm": 0.06717602163553238,
      "learning_rate": 0.000199983943331761,
      "loss": 1.5471,
      "step": 510
    },
    {
      "epoch": 0.005715436151531759,
      "grad_norm": 0.08228597790002823,
      "learning_rate": 0.0001999838803043953,
      "loss": 1.5446,
      "step": 511
    },
    {
      "epoch": 0.00572662095809053,
      "grad_norm": 0.07614196836948395,
      "learning_rate": 0.00019998381715358084,
      "loss": 1.5574,
      "step": 512
    },
    {
      "epoch": 0.005737805764649301,
      "grad_norm": 0.06075645610690117,
      "learning_rate": 0.00019998375387931774,
      "loss": 1.5597,
      "step": 513
    },
    {
      "epoch": 0.005748990571208071,
      "grad_norm": 0.05882800742983818,
      "learning_rate": 0.00019998369048160604,
      "loss": 1.5559,
      "step": 514
    },
    {
      "epoch": 0.0057601753777668414,
      "grad_norm": 0.07097506523132324,
      "learning_rate": 0.0001999836269604458,
      "loss": 1.5648,
      "step": 515
    },
    {
      "epoch": 0.005771360184325612,
      "grad_norm": 0.06486310064792633,
      "learning_rate": 0.00019998356331583716,
      "loss": 1.5735,
      "step": 516
    },
    {
      "epoch": 0.005782544990884383,
      "grad_norm": 0.05333361402153969,
      "learning_rate": 0.00019998349954778016,
      "loss": 1.567,
      "step": 517
    },
    {
      "epoch": 0.005793729797443153,
      "grad_norm": 0.07817003130912781,
      "learning_rate": 0.00019998343565627488,
      "loss": 1.5902,
      "step": 518
    },
    {
      "epoch": 0.0058049146040019236,
      "grad_norm": 0.07619974762201309,
      "learning_rate": 0.00019998337164132138,
      "loss": 1.5819,
      "step": 519
    },
    {
      "epoch": 0.005816099410560695,
      "grad_norm": 0.06044092774391174,
      "learning_rate": 0.0001999833075029198,
      "loss": 1.6163,
      "step": 520
    },
    {
      "epoch": 0.005827284217119465,
      "grad_norm": 0.06666608154773712,
      "learning_rate": 0.00019998324324107015,
      "loss": 1.5977,
      "step": 521
    },
    {
      "epoch": 0.005838469023678235,
      "grad_norm": 0.06902644038200378,
      "learning_rate": 0.00019998317885577254,
      "loss": 1.5818,
      "step": 522
    },
    {
      "epoch": 0.005849653830237006,
      "grad_norm": 0.05606195330619812,
      "learning_rate": 0.00019998311434702703,
      "loss": 1.5784,
      "step": 523
    },
    {
      "epoch": 0.005860838636795777,
      "grad_norm": 0.08465290814638138,
      "learning_rate": 0.00019998304971483374,
      "loss": 1.5653,
      "step": 524
    },
    {
      "epoch": 0.005872023443354547,
      "grad_norm": 0.08544803410768509,
      "learning_rate": 0.00019998298495919274,
      "loss": 1.5616,
      "step": 525
    },
    {
      "epoch": 0.0058832082499133175,
      "grad_norm": 0.06527858972549438,
      "learning_rate": 0.0001999829200801041,
      "loss": 1.5879,
      "step": 526
    },
    {
      "epoch": 0.005894393056472089,
      "grad_norm": 0.07150562107563019,
      "learning_rate": 0.00019998285507756789,
      "loss": 1.5642,
      "step": 527
    },
    {
      "epoch": 0.005905577863030859,
      "grad_norm": 0.08152669668197632,
      "learning_rate": 0.00019998278995158418,
      "loss": 1.5744,
      "step": 528
    },
    {
      "epoch": 0.005916762669589629,
      "grad_norm": 0.09149385243654251,
      "learning_rate": 0.0001999827247021531,
      "loss": 1.5715,
      "step": 529
    },
    {
      "epoch": 0.0059279474761484,
      "grad_norm": 0.08523254096508026,
      "learning_rate": 0.00019998265932927466,
      "loss": 1.5822,
      "step": 530
    },
    {
      "epoch": 0.005939132282707171,
      "grad_norm": 0.062498655170202255,
      "learning_rate": 0.000199982593832949,
      "loss": 1.5862,
      "step": 531
    },
    {
      "epoch": 0.005950317089265941,
      "grad_norm": 0.07431355118751526,
      "learning_rate": 0.0001999825282131762,
      "loss": 1.574,
      "step": 532
    },
    {
      "epoch": 0.005961501895824711,
      "grad_norm": 0.0720391795039177,
      "learning_rate": 0.00019998246246995632,
      "loss": 1.57,
      "step": 533
    },
    {
      "epoch": 0.005972686702383483,
      "grad_norm": 0.06915175914764404,
      "learning_rate": 0.00019998239660328943,
      "loss": 1.5721,
      "step": 534
    },
    {
      "epoch": 0.005983871508942253,
      "grad_norm": 0.058932509273290634,
      "learning_rate": 0.00019998233061317561,
      "loss": 1.583,
      "step": 535
    },
    {
      "epoch": 0.005995056315501023,
      "grad_norm": 0.05271697789430618,
      "learning_rate": 0.000199982264499615,
      "loss": 1.5762,
      "step": 536
    },
    {
      "epoch": 0.006006241122059794,
      "grad_norm": 0.05659927427768707,
      "learning_rate": 0.0001999821982626076,
      "loss": 1.5718,
      "step": 537
    },
    {
      "epoch": 0.006017425928618565,
      "grad_norm": 0.05749264359474182,
      "learning_rate": 0.00019998213190215353,
      "loss": 1.5897,
      "step": 538
    },
    {
      "epoch": 0.006028610735177335,
      "grad_norm": 0.06748930364847183,
      "learning_rate": 0.0001999820654182529,
      "loss": 1.5844,
      "step": 539
    },
    {
      "epoch": 0.006039795541736105,
      "grad_norm": 0.06751269847154617,
      "learning_rate": 0.0001999819988109057,
      "loss": 1.5926,
      "step": 540
    },
    {
      "epoch": 0.0060509803482948765,
      "grad_norm": 0.06434184312820435,
      "learning_rate": 0.00019998193208011213,
      "loss": 1.5742,
      "step": 541
    },
    {
      "epoch": 0.006062165154853647,
      "grad_norm": 0.06833555549383163,
      "learning_rate": 0.0001999818652258722,
      "loss": 1.5818,
      "step": 542
    },
    {
      "epoch": 0.006073349961412417,
      "grad_norm": 0.06202944368124008,
      "learning_rate": 0.000199981798248186,
      "loss": 1.5806,
      "step": 543
    },
    {
      "epoch": 0.006084534767971188,
      "grad_norm": 0.06603850424289703,
      "learning_rate": 0.00019998173114705366,
      "loss": 1.5892,
      "step": 544
    },
    {
      "epoch": 0.006095719574529959,
      "grad_norm": 0.04424119368195534,
      "learning_rate": 0.00019998166392247522,
      "loss": 1.6017,
      "step": 545
    },
    {
      "epoch": 0.006106904381088729,
      "grad_norm": 0.04886182025074959,
      "learning_rate": 0.00019998159657445074,
      "loss": 1.5971,
      "step": 546
    },
    {
      "epoch": 0.006118089187647499,
      "grad_norm": 0.055009353905916214,
      "learning_rate": 0.00019998152910298035,
      "loss": 1.5942,
      "step": 547
    },
    {
      "epoch": 0.0061292739942062704,
      "grad_norm": 0.04949553683400154,
      "learning_rate": 0.00019998146150806411,
      "loss": 1.5851,
      "step": 548
    },
    {
      "epoch": 0.006140458800765041,
      "grad_norm": 0.05669408291578293,
      "learning_rate": 0.0001999813937897021,
      "loss": 1.5565,
      "step": 549
    },
    {
      "epoch": 0.006151643607323811,
      "grad_norm": 0.052044518291950226,
      "learning_rate": 0.00019998132594789444,
      "loss": 1.5652,
      "step": 550
    },
    {
      "epoch": 0.006162828413882582,
      "grad_norm": 0.055031049996614456,
      "learning_rate": 0.00019998125798264117,
      "loss": 1.5509,
      "step": 551
    },
    {
      "epoch": 0.006174013220441353,
      "grad_norm": 0.0653780847787857,
      "learning_rate": 0.0001999811898939424,
      "loss": 1.5672,
      "step": 552
    },
    {
      "epoch": 0.006185198027000123,
      "grad_norm": 0.06647983938455582,
      "learning_rate": 0.00019998112168179822,
      "loss": 1.5744,
      "step": 553
    },
    {
      "epoch": 0.006196382833558893,
      "grad_norm": 0.062013398855924606,
      "learning_rate": 0.00019998105334620867,
      "loss": 1.5545,
      "step": 554
    },
    {
      "epoch": 0.006207567640117664,
      "grad_norm": 0.04633625969290733,
      "learning_rate": 0.00019998098488717384,
      "loss": 1.5643,
      "step": 555
    },
    {
      "epoch": 0.006218752446676435,
      "grad_norm": 0.04178265854716301,
      "learning_rate": 0.00019998091630469387,
      "loss": 1.5484,
      "step": 556
    },
    {
      "epoch": 0.006229937253235205,
      "grad_norm": 0.04647913947701454,
      "learning_rate": 0.00019998084759876883,
      "loss": 1.554,
      "step": 557
    },
    {
      "epoch": 0.006241122059793976,
      "grad_norm": 0.05254526063799858,
      "learning_rate": 0.00019998077876939876,
      "loss": 1.5671,
      "step": 558
    },
    {
      "epoch": 0.0062523068663527465,
      "grad_norm": 0.054666776210069656,
      "learning_rate": 0.00019998070981658376,
      "loss": 1.5709,
      "step": 559
    },
    {
      "epoch": 0.006263491672911517,
      "grad_norm": 0.04982587322592735,
      "learning_rate": 0.00019998064074032396,
      "loss": 1.5645,
      "step": 560
    },
    {
      "epoch": 0.006274676479470288,
      "grad_norm": 0.05644576624035835,
      "learning_rate": 0.00019998057154061938,
      "loss": 1.5512,
      "step": 561
    },
    {
      "epoch": 0.006285861286029058,
      "grad_norm": 0.05311651527881622,
      "learning_rate": 0.00019998050221747016,
      "loss": 1.5399,
      "step": 562
    },
    {
      "epoch": 0.006297046092587829,
      "grad_norm": 0.05626964941620827,
      "learning_rate": 0.00019998043277087634,
      "loss": 1.5389,
      "step": 563
    },
    {
      "epoch": 0.006308230899146599,
      "grad_norm": 0.06463497877120972,
      "learning_rate": 0.00019998036320083808,
      "loss": 1.549,
      "step": 564
    },
    {
      "epoch": 0.00631941570570537,
      "grad_norm": 0.07779069244861603,
      "learning_rate": 0.00019998029350735538,
      "loss": 1.5591,
      "step": 565
    },
    {
      "epoch": 0.00633060051226414,
      "grad_norm": 0.10419348627328873,
      "learning_rate": 0.00019998022369042837,
      "loss": 1.5669,
      "step": 566
    },
    {
      "epoch": 0.006341785318822911,
      "grad_norm": 0.11543929576873779,
      "learning_rate": 0.00019998015375005709,
      "loss": 1.5669,
      "step": 567
    },
    {
      "epoch": 0.006352970125381682,
      "grad_norm": 0.08400971442461014,
      "learning_rate": 0.0001999800836862417,
      "loss": 1.589,
      "step": 568
    },
    {
      "epoch": 0.006364154931940452,
      "grad_norm": 0.056216295808553696,
      "learning_rate": 0.00019998001349898225,
      "loss": 1.5781,
      "step": 569
    },
    {
      "epoch": 0.0063753397384992225,
      "grad_norm": 0.07171747833490372,
      "learning_rate": 0.0001999799431882788,
      "loss": 1.5722,
      "step": 570
    },
    {
      "epoch": 0.006386524545057993,
      "grad_norm": 0.07911943644285202,
      "learning_rate": 0.0001999798727541315,
      "loss": 1.5617,
      "step": 571
    },
    {
      "epoch": 0.006397709351616764,
      "grad_norm": 0.07471580803394318,
      "learning_rate": 0.0001999798021965404,
      "loss": 1.5674,
      "step": 572
    },
    {
      "epoch": 0.006408894158175534,
      "grad_norm": 0.06016454100608826,
      "learning_rate": 0.00019997973151550556,
      "loss": 1.589,
      "step": 573
    },
    {
      "epoch": 0.006420078964734305,
      "grad_norm": 0.06692295521497726,
      "learning_rate": 0.00019997966071102713,
      "loss": 1.5721,
      "step": 574
    },
    {
      "epoch": 0.006431263771293076,
      "grad_norm": 0.06640581041574478,
      "learning_rate": 0.00019997958978310514,
      "loss": 1.5781,
      "step": 575
    },
    {
      "epoch": 0.006442448577851846,
      "grad_norm": 0.058826372027397156,
      "learning_rate": 0.0001999795187317397,
      "loss": 1.5666,
      "step": 576
    },
    {
      "epoch": 0.0064536333844106165,
      "grad_norm": 0.055648185312747955,
      "learning_rate": 0.0001999794475569309,
      "loss": 1.5707,
      "step": 577
    },
    {
      "epoch": 0.006464818190969387,
      "grad_norm": 0.058248959481716156,
      "learning_rate": 0.00019997937625867884,
      "loss": 1.57,
      "step": 578
    },
    {
      "epoch": 0.006476002997528158,
      "grad_norm": 0.05667665973305702,
      "learning_rate": 0.00019997930483698357,
      "loss": 1.5715,
      "step": 579
    },
    {
      "epoch": 0.006487187804086928,
      "grad_norm": 0.051860544830560684,
      "learning_rate": 0.00019997923329184524,
      "loss": 1.5875,
      "step": 580
    },
    {
      "epoch": 0.006498372610645699,
      "grad_norm": 0.05429021269083023,
      "learning_rate": 0.00019997916162326385,
      "loss": 1.606,
      "step": 581
    },
    {
      "epoch": 0.00650955741720447,
      "grad_norm": 0.055650923401117325,
      "learning_rate": 0.00019997908983123956,
      "loss": 1.6024,
      "step": 582
    },
    {
      "epoch": 0.00652074222376324,
      "grad_norm": 0.061447255313396454,
      "learning_rate": 0.00019997901791577244,
      "loss": 1.5888,
      "step": 583
    },
    {
      "epoch": 0.00653192703032201,
      "grad_norm": 0.06065785884857178,
      "learning_rate": 0.00019997894587686255,
      "loss": 1.5739,
      "step": 584
    },
    {
      "epoch": 0.006543111836880782,
      "grad_norm": 0.07358521968126297,
      "learning_rate": 0.00019997887371451002,
      "loss": 1.5682,
      "step": 585
    },
    {
      "epoch": 0.006554296643439552,
      "grad_norm": 0.08286885917186737,
      "learning_rate": 0.00019997880142871494,
      "loss": 1.5702,
      "step": 586
    },
    {
      "epoch": 0.006565481449998322,
      "grad_norm": 0.09056065231561661,
      "learning_rate": 0.0001999787290194774,
      "loss": 1.5822,
      "step": 587
    },
    {
      "epoch": 0.0065766662565570925,
      "grad_norm": 0.08298853039741516,
      "learning_rate": 0.00019997865648679745,
      "loss": 1.5818,
      "step": 588
    },
    {
      "epoch": 0.006587851063115864,
      "grad_norm": 0.08499585837125778,
      "learning_rate": 0.00019997858383067517,
      "loss": 1.5775,
      "step": 589
    },
    {
      "epoch": 0.006599035869674634,
      "grad_norm": 0.08271525800228119,
      "learning_rate": 0.00019997851105111073,
      "loss": 1.5756,
      "step": 590
    },
    {
      "epoch": 0.006610220676233404,
      "grad_norm": 0.07318850606679916,
      "learning_rate": 0.00019997843814810416,
      "loss": 1.5674,
      "step": 591
    },
    {
      "epoch": 0.0066214054827921755,
      "grad_norm": 0.07372857630252838,
      "learning_rate": 0.00019997836512165558,
      "loss": 1.5589,
      "step": 592
    },
    {
      "epoch": 0.006632590289350946,
      "grad_norm": 0.09608045220375061,
      "learning_rate": 0.00019997829197176503,
      "loss": 1.5483,
      "step": 593
    },
    {
      "epoch": 0.006643775095909716,
      "grad_norm": 0.13775509595870972,
      "learning_rate": 0.00019997821869843264,
      "loss": 1.5534,
      "step": 594
    },
    {
      "epoch": 0.0066549599024684864,
      "grad_norm": 0.1282949000597,
      "learning_rate": 0.00019997814530165847,
      "loss": 1.5707,
      "step": 595
    },
    {
      "epoch": 0.006666144709027258,
      "grad_norm": 0.09030576795339584,
      "learning_rate": 0.00019997807178144268,
      "loss": 1.5759,
      "step": 596
    },
    {
      "epoch": 0.006677329515586028,
      "grad_norm": 0.08960919827222824,
      "learning_rate": 0.00019997799813778531,
      "loss": 1.5747,
      "step": 597
    },
    {
      "epoch": 0.006688514322144798,
      "grad_norm": 0.08592968434095383,
      "learning_rate": 0.00019997792437068644,
      "loss": 1.5837,
      "step": 598
    },
    {
      "epoch": 0.0066996991287035694,
      "grad_norm": 0.07291566580533981,
      "learning_rate": 0.00019997785048014616,
      "loss": 1.5797,
      "step": 599
    },
    {
      "epoch": 0.00671088393526234,
      "grad_norm": 0.07706471532583237,
      "learning_rate": 0.0001999777764661646,
      "loss": 1.5715,
      "step": 600
    },
    {
      "epoch": 0.00672206874182111,
      "grad_norm": 0.06954386830329895,
      "learning_rate": 0.00019997770232874182,
      "loss": 1.563,
      "step": 601
    },
    {
      "epoch": 0.00673325354837988,
      "grad_norm": 0.06999648362398148,
      "learning_rate": 0.00019997762806787792,
      "loss": 1.5717,
      "step": 602
    },
    {
      "epoch": 0.0067444383549386516,
      "grad_norm": 0.05400196090340614,
      "learning_rate": 0.00019997755368357298,
      "loss": 1.5862,
      "step": 603
    },
    {
      "epoch": 0.006755623161497422,
      "grad_norm": 0.06418072432279587,
      "learning_rate": 0.00019997747917582714,
      "loss": 1.5908,
      "step": 604
    },
    {
      "epoch": 0.006766807968056192,
      "grad_norm": 0.05838518589735031,
      "learning_rate": 0.00019997740454464044,
      "loss": 1.5718,
      "step": 605
    },
    {
      "epoch": 0.006777992774614963,
      "grad_norm": 0.05882187560200691,
      "learning_rate": 0.00019997732979001298,
      "loss": 1.5769,
      "step": 606
    },
    {
      "epoch": 0.006789177581173734,
      "grad_norm": 0.05842543765902519,
      "learning_rate": 0.00019997725491194482,
      "loss": 1.5746,
      "step": 607
    },
    {
      "epoch": 0.006800362387732504,
      "grad_norm": 0.0527958944439888,
      "learning_rate": 0.00019997717991043616,
      "loss": 1.5682,
      "step": 608
    },
    {
      "epoch": 0.006811547194291274,
      "grad_norm": 0.06266690045595169,
      "learning_rate": 0.00019997710478548698,
      "loss": 1.572,
      "step": 609
    },
    {
      "epoch": 0.0068227320008500455,
      "grad_norm": 0.05554317682981491,
      "learning_rate": 0.00019997702953709746,
      "loss": 1.583,
      "step": 610
    },
    {
      "epoch": 0.006833916807408816,
      "grad_norm": 0.04651861637830734,
      "learning_rate": 0.00019997695416526761,
      "loss": 1.5738,
      "step": 611
    },
    {
      "epoch": 0.006845101613967586,
      "grad_norm": 0.053717561066150665,
      "learning_rate": 0.0001999768786699976,
      "loss": 1.5766,
      "step": 612
    },
    {
      "epoch": 0.006856286420526357,
      "grad_norm": 0.050605516880750656,
      "learning_rate": 0.0001999768030512875,
      "loss": 1.606,
      "step": 613
    },
    {
      "epoch": 0.006867471227085128,
      "grad_norm": 0.054307371377944946,
      "learning_rate": 0.00019997672730913735,
      "loss": 1.6187,
      "step": 614
    },
    {
      "epoch": 0.006878656033643898,
      "grad_norm": 0.06506580859422684,
      "learning_rate": 0.00019997665144354728,
      "loss": 1.6113,
      "step": 615
    },
    {
      "epoch": 0.006889840840202669,
      "grad_norm": 0.06480210274457932,
      "learning_rate": 0.00019997657545451744,
      "loss": 1.594,
      "step": 616
    },
    {
      "epoch": 0.006901025646761439,
      "grad_norm": 0.04906410723924637,
      "learning_rate": 0.00019997649934204784,
      "loss": 1.5809,
      "step": 617
    },
    {
      "epoch": 0.00691221045332021,
      "grad_norm": 0.05194586515426636,
      "learning_rate": 0.00019997642310613857,
      "loss": 1.5913,
      "step": 618
    },
    {
      "epoch": 0.00692339525987898,
      "grad_norm": 0.05839546024799347,
      "learning_rate": 0.0001999763467467898,
      "loss": 1.5746,
      "step": 619
    },
    {
      "epoch": 0.006934580066437751,
      "grad_norm": 0.06750357896089554,
      "learning_rate": 0.0001999762702640016,
      "loss": 1.575,
      "step": 620
    },
    {
      "epoch": 0.0069457648729965215,
      "grad_norm": 0.07982991635799408,
      "learning_rate": 0.00019997619365777402,
      "loss": 1.576,
      "step": 621
    },
    {
      "epoch": 0.006956949679555292,
      "grad_norm": 0.08816216886043549,
      "learning_rate": 0.00019997611692810718,
      "loss": 1.5744,
      "step": 622
    },
    {
      "epoch": 0.006968134486114063,
      "grad_norm": 0.09619053453207016,
      "learning_rate": 0.0001999760400750012,
      "loss": 1.5791,
      "step": 623
    },
    {
      "epoch": 0.006979319292672833,
      "grad_norm": 0.09412987530231476,
      "learning_rate": 0.00019997596309845612,
      "loss": 1.5799,
      "step": 624
    },
    {
      "epoch": 0.006990504099231604,
      "grad_norm": 0.09227743744850159,
      "learning_rate": 0.0001999758859984721,
      "loss": 1.574,
      "step": 625
    },
    {
      "epoch": 0.007001688905790374,
      "grad_norm": 0.07984034717082977,
      "learning_rate": 0.00019997580877504918,
      "loss": 1.5612,
      "step": 626
    },
    {
      "epoch": 0.007012873712349145,
      "grad_norm": 0.05941009521484375,
      "learning_rate": 0.00019997573142818752,
      "loss": 1.5655,
      "step": 627
    },
    {
      "epoch": 0.0070240585189079155,
      "grad_norm": 0.06787893921136856,
      "learning_rate": 0.0001999756539578871,
      "loss": 1.5662,
      "step": 628
    },
    {
      "epoch": 0.007035243325466686,
      "grad_norm": 0.07556013017892838,
      "learning_rate": 0.00019997557636414816,
      "loss": 1.5877,
      "step": 629
    },
    {
      "epoch": 0.007046428132025457,
      "grad_norm": 0.06565730273723602,
      "learning_rate": 0.0001999754986469707,
      "loss": 1.5701,
      "step": 630
    },
    {
      "epoch": 0.007057612938584227,
      "grad_norm": 0.05801456421613693,
      "learning_rate": 0.00019997542080635482,
      "loss": 1.5604,
      "step": 631
    },
    {
      "epoch": 0.007068797745142998,
      "grad_norm": 0.058777451515197754,
      "learning_rate": 0.00019997534284230066,
      "loss": 1.5695,
      "step": 632
    },
    {
      "epoch": 0.007079982551701768,
      "grad_norm": 0.0650622621178627,
      "learning_rate": 0.0001999752647548083,
      "loss": 1.5612,
      "step": 633
    },
    {
      "epoch": 0.007091167358260539,
      "grad_norm": 0.05834876000881195,
      "learning_rate": 0.00019997518654387783,
      "loss": 1.5617,
      "step": 634
    },
    {
      "epoch": 0.007102352164819309,
      "grad_norm": 0.06384813785552979,
      "learning_rate": 0.00019997510820950933,
      "loss": 1.5567,
      "step": 635
    },
    {
      "epoch": 0.00711353697137808,
      "grad_norm": 0.05401776731014252,
      "learning_rate": 0.00019997502975170291,
      "loss": 1.5558,
      "step": 636
    },
    {
      "epoch": 0.007124721777936851,
      "grad_norm": 0.06590646505355835,
      "learning_rate": 0.00019997495117045867,
      "loss": 1.5474,
      "step": 637
    },
    {
      "epoch": 0.007135906584495621,
      "grad_norm": 0.0560823492705822,
      "learning_rate": 0.00019997487246577674,
      "loss": 1.5404,
      "step": 638
    },
    {
      "epoch": 0.0071470913910543915,
      "grad_norm": 0.0544624887406826,
      "learning_rate": 0.00019997479363765717,
      "loss": 1.5321,
      "step": 639
    },
    {
      "epoch": 0.007158276197613163,
      "grad_norm": 0.04914103075861931,
      "learning_rate": 0.00019997471468610005,
      "loss": 1.5395,
      "step": 640
    },
    {
      "epoch": 0.007169461004171933,
      "grad_norm": 0.0481346994638443,
      "learning_rate": 0.00019997463561110553,
      "loss": 1.5513,
      "step": 641
    },
    {
      "epoch": 0.007180645810730703,
      "grad_norm": 0.04910167306661606,
      "learning_rate": 0.00019997455641267367,
      "loss": 1.5436,
      "step": 642
    },
    {
      "epoch": 0.007191830617289474,
      "grad_norm": 0.05214869976043701,
      "learning_rate": 0.00019997447709080456,
      "loss": 1.5589,
      "step": 643
    },
    {
      "epoch": 0.007203015423848245,
      "grad_norm": 0.06242618337273598,
      "learning_rate": 0.00019997439764549832,
      "loss": 1.5395,
      "step": 644
    },
    {
      "epoch": 0.007214200230407015,
      "grad_norm": 0.07024102658033371,
      "learning_rate": 0.00019997431807675505,
      "loss": 1.5624,
      "step": 645
    },
    {
      "epoch": 0.0072253850369657854,
      "grad_norm": 0.07082174718379974,
      "learning_rate": 0.0001999742383845748,
      "loss": 1.5505,
      "step": 646
    },
    {
      "epoch": 0.007236569843524557,
      "grad_norm": 0.06821414083242416,
      "learning_rate": 0.00019997415856895775,
      "loss": 1.5489,
      "step": 647
    },
    {
      "epoch": 0.007247754650083327,
      "grad_norm": 0.062424443662166595,
      "learning_rate": 0.00019997407862990395,
      "loss": 1.5341,
      "step": 648
    },
    {
      "epoch": 0.007258939456642097,
      "grad_norm": 0.05251247063279152,
      "learning_rate": 0.00019997399856741348,
      "loss": 1.5257,
      "step": 649
    },
    {
      "epoch": 0.0072701242632008676,
      "grad_norm": 0.04626723378896713,
      "learning_rate": 0.0001999739183814865,
      "loss": 1.5387,
      "step": 650
    },
    {
      "epoch": 0.007281309069759639,
      "grad_norm": 0.05231785774230957,
      "learning_rate": 0.00019997383807212306,
      "loss": 1.5378,
      "step": 651
    },
    {
      "epoch": 0.007292493876318409,
      "grad_norm": 0.06309188902378082,
      "learning_rate": 0.00019997375763932323,
      "loss": 1.5531,
      "step": 652
    },
    {
      "epoch": 0.007303678682877179,
      "grad_norm": 0.05309786647558212,
      "learning_rate": 0.0001999736770830872,
      "loss": 1.5694,
      "step": 653
    },
    {
      "epoch": 0.0073148634894359505,
      "grad_norm": 0.04923882707953453,
      "learning_rate": 0.000199973596403415,
      "loss": 1.5837,
      "step": 654
    },
    {
      "epoch": 0.007326048295994721,
      "grad_norm": 0.05534524843096733,
      "learning_rate": 0.00019997351560030677,
      "loss": 1.5918,
      "step": 655
    },
    {
      "epoch": 0.007337233102553491,
      "grad_norm": 0.1240246444940567,
      "learning_rate": 0.00019997343467376258,
      "loss": 1.5881,
      "step": 656
    },
    {
      "epoch": 0.0073484179091122615,
      "grad_norm": 0.06112068518996239,
      "learning_rate": 0.00019997335362378254,
      "loss": 1.5761,
      "step": 657
    },
    {
      "epoch": 0.007359602715671033,
      "grad_norm": 0.06589160114526749,
      "learning_rate": 0.00019997327245036673,
      "loss": 1.5641,
      "step": 658
    },
    {
      "epoch": 0.007370787522229803,
      "grad_norm": 0.060181207954883575,
      "learning_rate": 0.0001999731911535153,
      "loss": 1.5573,
      "step": 659
    },
    {
      "epoch": 0.007381972328788573,
      "grad_norm": 0.06380990892648697,
      "learning_rate": 0.0001999731097332283,
      "loss": 1.5624,
      "step": 660
    },
    {
      "epoch": 0.0073931571353473445,
      "grad_norm": 0.06176357343792915,
      "learning_rate": 0.00019997302818950584,
      "loss": 1.5499,
      "step": 661
    },
    {
      "epoch": 0.007404341941906115,
      "grad_norm": 0.055721499025821686,
      "learning_rate": 0.00019997294652234805,
      "loss": 1.557,
      "step": 662
    },
    {
      "epoch": 0.007415526748464885,
      "grad_norm": 0.051978956907987595,
      "learning_rate": 0.000199972864731755,
      "loss": 1.5623,
      "step": 663
    },
    {
      "epoch": 0.007426711555023656,
      "grad_norm": 0.04754827544093132,
      "learning_rate": 0.00019997278281772682,
      "loss": 1.5465,
      "step": 664
    },
    {
      "epoch": 0.007437896361582427,
      "grad_norm": 0.0538279265165329,
      "learning_rate": 0.0001999727007802636,
      "loss": 1.5326,
      "step": 665
    },
    {
      "epoch": 0.007449081168141197,
      "grad_norm": 0.0629352405667305,
      "learning_rate": 0.00019997261861936543,
      "loss": 1.5365,
      "step": 666
    },
    {
      "epoch": 0.007460265974699967,
      "grad_norm": 0.06892745941877365,
      "learning_rate": 0.00019997253633503238,
      "loss": 1.5607,
      "step": 667
    },
    {
      "epoch": 0.007471450781258738,
      "grad_norm": 0.07525767385959625,
      "learning_rate": 0.00019997245392726465,
      "loss": 1.5728,
      "step": 668
    },
    {
      "epoch": 0.007482635587817509,
      "grad_norm": 0.10010071098804474,
      "learning_rate": 0.00019997237139606224,
      "loss": 1.559,
      "step": 669
    },
    {
      "epoch": 0.007493820394376279,
      "grad_norm": 0.12011202424764633,
      "learning_rate": 0.0001999722887414253,
      "loss": 1.5694,
      "step": 670
    },
    {
      "epoch": 0.00750500520093505,
      "grad_norm": 0.12278566509485245,
      "learning_rate": 0.00019997220596335393,
      "loss": 1.5939,
      "step": 671
    },
    {
      "epoch": 0.0075161900074938205,
      "grad_norm": 0.10163605213165283,
      "learning_rate": 0.00019997212306184823,
      "loss": 1.5722,
      "step": 672
    },
    {
      "epoch": 0.007527374814052591,
      "grad_norm": 0.09386469423770905,
      "learning_rate": 0.00019997204003690828,
      "loss": 1.5748,
      "step": 673
    },
    {
      "epoch": 0.007538559620611361,
      "grad_norm": 0.1031983494758606,
      "learning_rate": 0.00019997195688853422,
      "loss": 1.5778,
      "step": 674
    },
    {
      "epoch": 0.007549744427170132,
      "grad_norm": 0.09082422405481339,
      "learning_rate": 0.00019997187361672615,
      "loss": 1.5904,
      "step": 675
    },
    {
      "epoch": 0.007560929233728903,
      "grad_norm": 0.05761239677667618,
      "learning_rate": 0.0001999717902214841,
      "loss": 1.5753,
      "step": 676
    },
    {
      "epoch": 0.007572114040287673,
      "grad_norm": 0.0772882029414177,
      "learning_rate": 0.0001999717067028083,
      "loss": 1.5631,
      "step": 677
    },
    {
      "epoch": 0.007583298846846444,
      "grad_norm": 0.09266892075538635,
      "learning_rate": 0.00019997162306069875,
      "loss": 1.5798,
      "step": 678
    },
    {
      "epoch": 0.0075944836534052144,
      "grad_norm": 0.07755053043365479,
      "learning_rate": 0.00019997153929515558,
      "loss": 1.5969,
      "step": 679
    },
    {
      "epoch": 0.007605668459963985,
      "grad_norm": 0.061833951622247696,
      "learning_rate": 0.0001999714554061789,
      "loss": 1.5781,
      "step": 680
    },
    {
      "epoch": 0.007616853266522755,
      "grad_norm": 0.07911964505910873,
      "learning_rate": 0.00019997137139376883,
      "loss": 1.611,
      "step": 681
    },
    {
      "epoch": 0.007628038073081526,
      "grad_norm": 0.07502644509077072,
      "learning_rate": 0.00019997128725792544,
      "loss": 1.6201,
      "step": 682
    },
    {
      "epoch": 0.007639222879640297,
      "grad_norm": 0.11084458976984024,
      "learning_rate": 0.00019997120299864886,
      "loss": 1.5882,
      "step": 683
    },
    {
      "epoch": 0.007650407686199067,
      "grad_norm": 0.1428053230047226,
      "learning_rate": 0.00019997111861593921,
      "loss": 1.5737,
      "step": 684
    },
    {
      "epoch": 0.007661592492757838,
      "grad_norm": 0.1456058770418167,
      "learning_rate": 0.00019997103410979652,
      "loss": 1.583,
      "step": 685
    },
    {
      "epoch": 0.007672777299316608,
      "grad_norm": 0.10741148889064789,
      "learning_rate": 0.00019997094948022098,
      "loss": 1.5736,
      "step": 686
    },
    {
      "epoch": 0.007683962105875379,
      "grad_norm": 0.08248301595449448,
      "learning_rate": 0.00019997086472721263,
      "loss": 1.5559,
      "step": 687
    },
    {
      "epoch": 0.00769514691243415,
      "grad_norm": 0.09595336019992828,
      "learning_rate": 0.00019997077985077163,
      "loss": 1.5513,
      "step": 688
    },
    {
      "epoch": 0.00770633171899292,
      "grad_norm": 0.06806618720293045,
      "learning_rate": 0.00019997069485089804,
      "loss": 1.5624,
      "step": 689
    },
    {
      "epoch": 0.0077175165255516905,
      "grad_norm": 0.07510481029748917,
      "learning_rate": 0.00019997060972759198,
      "loss": 1.5401,
      "step": 690
    },
    {
      "epoch": 0.007728701332110461,
      "grad_norm": 0.06557908654212952,
      "learning_rate": 0.00019997052448085358,
      "loss": 1.5507,
      "step": 691
    },
    {
      "epoch": 0.007739886138669232,
      "grad_norm": 0.07286231964826584,
      "learning_rate": 0.0001999704391106829,
      "loss": 1.5597,
      "step": 692
    },
    {
      "epoch": 0.007751070945228002,
      "grad_norm": 0.06357460469007492,
      "learning_rate": 0.0001999703536170801,
      "loss": 1.5591,
      "step": 693
    },
    {
      "epoch": 0.007762255751786773,
      "grad_norm": 0.06700731813907623,
      "learning_rate": 0.00019997026800004522,
      "loss": 1.5521,
      "step": 694
    },
    {
      "epoch": 0.007773440558345544,
      "grad_norm": 0.05840053781867027,
      "learning_rate": 0.00019997018225957839,
      "loss": 1.5688,
      "step": 695
    },
    {
      "epoch": 0.007784625364904314,
      "grad_norm": 0.06327050924301147,
      "learning_rate": 0.00019997009639567974,
      "loss": 1.5547,
      "step": 696
    },
    {
      "epoch": 0.007795810171463084,
      "grad_norm": 0.06035961955785751,
      "learning_rate": 0.00019997001040834936,
      "loss": 1.5701,
      "step": 697
    },
    {
      "epoch": 0.007806994978021855,
      "grad_norm": 0.05573936179280281,
      "learning_rate": 0.0001999699242975874,
      "loss": 1.5878,
      "step": 698
    },
    {
      "epoch": 0.007818179784580626,
      "grad_norm": 0.05611170828342438,
      "learning_rate": 0.00019996983806339387,
      "loss": 1.5899,
      "step": 699
    },
    {
      "epoch": 0.007829364591139396,
      "grad_norm": 0.05826570466160774,
      "learning_rate": 0.00019996975170576896,
      "loss": 1.5533,
      "step": 700
    },
    {
      "epoch": 0.007840549397698167,
      "grad_norm": 0.050937049090862274,
      "learning_rate": 0.00019996966522471273,
      "loss": 1.5545,
      "step": 701
    },
    {
      "epoch": 0.007851734204256937,
      "grad_norm": 0.06593479216098785,
      "learning_rate": 0.0001999695786202253,
      "loss": 1.5685,
      "step": 702
    },
    {
      "epoch": 0.007862919010815707,
      "grad_norm": 0.06405465304851532,
      "learning_rate": 0.0001999694918923068,
      "loss": 1.6121,
      "step": 703
    },
    {
      "epoch": 0.00787410381737448,
      "grad_norm": 0.052820343524217606,
      "learning_rate": 0.0001999694050409573,
      "loss": 1.6007,
      "step": 704
    },
    {
      "epoch": 0.00788528862393325,
      "grad_norm": 0.05512186512351036,
      "learning_rate": 0.00019996931806617695,
      "loss": 1.5792,
      "step": 705
    },
    {
      "epoch": 0.00789647343049202,
      "grad_norm": 0.0432184673845768,
      "learning_rate": 0.0001999692309679658,
      "loss": 1.5852,
      "step": 706
    },
    {
      "epoch": 0.00790765823705079,
      "grad_norm": 0.05248282849788666,
      "learning_rate": 0.00019996914374632402,
      "loss": 1.5917,
      "step": 707
    },
    {
      "epoch": 0.00791884304360956,
      "grad_norm": 0.04413476958870888,
      "learning_rate": 0.00019996905640125165,
      "loss": 1.5584,
      "step": 708
    },
    {
      "epoch": 0.00793002785016833,
      "grad_norm": 0.04878908023238182,
      "learning_rate": 0.00019996896893274886,
      "loss": 1.5506,
      "step": 709
    },
    {
      "epoch": 0.007941212656727101,
      "grad_norm": 0.04344234988093376,
      "learning_rate": 0.00019996888134081575,
      "loss": 1.5684,
      "step": 710
    },
    {
      "epoch": 0.007952397463285873,
      "grad_norm": 0.047158923000097275,
      "learning_rate": 0.0001999687936254524,
      "loss": 1.6133,
      "step": 711
    },
    {
      "epoch": 0.007963582269844643,
      "grad_norm": 0.050282686948776245,
      "learning_rate": 0.00019996870578665893,
      "loss": 1.6021,
      "step": 712
    },
    {
      "epoch": 0.007974767076403414,
      "grad_norm": 0.043916840106248856,
      "learning_rate": 0.0001999686178244354,
      "loss": 1.5794,
      "step": 713
    },
    {
      "epoch": 0.007985951882962184,
      "grad_norm": 0.079423688352108,
      "learning_rate": 0.00019996852973878205,
      "loss": 1.5666,
      "step": 714
    },
    {
      "epoch": 0.007997136689520954,
      "grad_norm": 0.047040194272994995,
      "learning_rate": 0.00019996844152969884,
      "loss": 1.5643,
      "step": 715
    },
    {
      "epoch": 0.008008321496079725,
      "grad_norm": 0.04954817518591881,
      "learning_rate": 0.00019996835319718596,
      "loss": 1.5756,
      "step": 716
    },
    {
      "epoch": 0.008019506302638495,
      "grad_norm": 0.0529201366007328,
      "learning_rate": 0.00019996826474124352,
      "loss": 1.5693,
      "step": 717
    },
    {
      "epoch": 0.008030691109197267,
      "grad_norm": 0.0555887334048748,
      "learning_rate": 0.00019996817616187162,
      "loss": 1.5699,
      "step": 718
    },
    {
      "epoch": 0.008041875915756037,
      "grad_norm": 0.05515376478433609,
      "learning_rate": 0.00019996808745907036,
      "loss": 1.5729,
      "step": 719
    },
    {
      "epoch": 0.008053060722314808,
      "grad_norm": 0.05125884339213371,
      "learning_rate": 0.0001999679986328398,
      "loss": 1.5618,
      "step": 720
    },
    {
      "epoch": 0.008064245528873578,
      "grad_norm": 0.046284329146146774,
      "learning_rate": 0.0001999679096831802,
      "loss": 1.5723,
      "step": 721
    },
    {
      "epoch": 0.008075430335432348,
      "grad_norm": 0.07273488491773605,
      "learning_rate": 0.0001999678206100915,
      "loss": 1.5701,
      "step": 722
    },
    {
      "epoch": 0.008086615141991119,
      "grad_norm": 0.047560565173625946,
      "learning_rate": 0.0001999677314135739,
      "loss": 1.5579,
      "step": 723
    },
    {
      "epoch": 0.00809779994854989,
      "grad_norm": 0.060283761471509933,
      "learning_rate": 0.00019996764209362749,
      "loss": 1.5615,
      "step": 724
    },
    {
      "epoch": 0.008108984755108661,
      "grad_norm": 0.0602620430290699,
      "learning_rate": 0.00019996755265025236,
      "loss": 1.5602,
      "step": 725
    },
    {
      "epoch": 0.008120169561667431,
      "grad_norm": 0.05383098125457764,
      "learning_rate": 0.00019996746308344868,
      "loss": 1.5769,
      "step": 726
    },
    {
      "epoch": 0.008131354368226202,
      "grad_norm": 0.04577267915010452,
      "learning_rate": 0.0001999673733932165,
      "loss": 1.619,
      "step": 727
    },
    {
      "epoch": 0.008142539174784972,
      "grad_norm": 0.04550078883767128,
      "learning_rate": 0.00019996728357955595,
      "loss": 1.5907,
      "step": 728
    },
    {
      "epoch": 0.008153723981343742,
      "grad_norm": 0.050287820398807526,
      "learning_rate": 0.00019996719364246714,
      "loss": 1.5809,
      "step": 729
    },
    {
      "epoch": 0.008164908787902513,
      "grad_norm": 0.05841783806681633,
      "learning_rate": 0.00019996710358195018,
      "loss": 1.5521,
      "step": 730
    },
    {
      "epoch": 0.008176093594461285,
      "grad_norm": 0.07749857753515244,
      "learning_rate": 0.0001999670133980052,
      "loss": 1.5848,
      "step": 731
    },
    {
      "epoch": 0.008187278401020055,
      "grad_norm": 0.08802466839551926,
      "learning_rate": 0.00019996692309063232,
      "loss": 1.6046,
      "step": 732
    },
    {
      "epoch": 0.008198463207578825,
      "grad_norm": 0.09324830025434494,
      "learning_rate": 0.00019996683265983162,
      "loss": 1.5969,
      "step": 733
    },
    {
      "epoch": 0.008209648014137596,
      "grad_norm": 0.07845516502857208,
      "learning_rate": 0.0001999667421056032,
      "loss": 1.5831,
      "step": 734
    },
    {
      "epoch": 0.008220832820696366,
      "grad_norm": 0.06912586092948914,
      "learning_rate": 0.0001999666514279472,
      "loss": 1.5706,
      "step": 735
    },
    {
      "epoch": 0.008232017627255136,
      "grad_norm": 0.0572381317615509,
      "learning_rate": 0.00019996656062686374,
      "loss": 1.5609,
      "step": 736
    },
    {
      "epoch": 0.008243202433813906,
      "grad_norm": 0.06219245865941048,
      "learning_rate": 0.00019996646970235287,
      "loss": 1.5541,
      "step": 737
    },
    {
      "epoch": 0.008254387240372679,
      "grad_norm": 0.0628521591424942,
      "learning_rate": 0.0001999663786544148,
      "loss": 1.5556,
      "step": 738
    },
    {
      "epoch": 0.008265572046931449,
      "grad_norm": 0.06389934569597244,
      "learning_rate": 0.0001999662874830496,
      "loss": 1.5235,
      "step": 739
    },
    {
      "epoch": 0.00827675685349022,
      "grad_norm": 0.052320901304483414,
      "learning_rate": 0.00019996619618825733,
      "loss": 1.539,
      "step": 740
    },
    {
      "epoch": 0.00828794166004899,
      "grad_norm": 0.05470295995473862,
      "learning_rate": 0.00019996610477003817,
      "loss": 1.5415,
      "step": 741
    },
    {
      "epoch": 0.00829912646660776,
      "grad_norm": 0.06103771552443504,
      "learning_rate": 0.00019996601322839222,
      "loss": 1.5422,
      "step": 742
    },
    {
      "epoch": 0.00831031127316653,
      "grad_norm": 0.06434791535139084,
      "learning_rate": 0.00019996592156331958,
      "loss": 1.5527,
      "step": 743
    },
    {
      "epoch": 0.0083214960797253,
      "grad_norm": 0.06087024137377739,
      "learning_rate": 0.00019996582977482036,
      "loss": 1.562,
      "step": 744
    },
    {
      "epoch": 0.008332680886284072,
      "grad_norm": 0.060757141560316086,
      "learning_rate": 0.00019996573786289465,
      "loss": 1.5641,
      "step": 745
    },
    {
      "epoch": 0.008343865692842843,
      "grad_norm": 0.07097544521093369,
      "learning_rate": 0.00019996564582754265,
      "loss": 1.542,
      "step": 746
    },
    {
      "epoch": 0.008355050499401613,
      "grad_norm": 0.07591135054826736,
      "learning_rate": 0.00019996555366876437,
      "loss": 1.5557,
      "step": 747
    },
    {
      "epoch": 0.008366235305960383,
      "grad_norm": 0.07860101759433746,
      "learning_rate": 0.00019996546138655998,
      "loss": 1.5592,
      "step": 748
    },
    {
      "epoch": 0.008377420112519154,
      "grad_norm": 0.08454012125730515,
      "learning_rate": 0.00019996536898092958,
      "loss": 1.5428,
      "step": 749
    },
    {
      "epoch": 0.008388604919077924,
      "grad_norm": 0.08686886727809906,
      "learning_rate": 0.0001999652764518733,
      "loss": 1.5604,
      "step": 750
    },
    {
      "epoch": 0.008399789725636694,
      "grad_norm": 0.07752903550863266,
      "learning_rate": 0.00019996518379939126,
      "loss": 1.5663,
      "step": 751
    },
    {
      "epoch": 0.008410974532195466,
      "grad_norm": 0.07272690534591675,
      "learning_rate": 0.00019996509102348356,
      "loss": 1.5463,
      "step": 752
    },
    {
      "epoch": 0.008422159338754237,
      "grad_norm": 0.07069668918848038,
      "learning_rate": 0.00019996499812415026,
      "loss": 1.5403,
      "step": 753
    },
    {
      "epoch": 0.008433344145313007,
      "grad_norm": 0.06617298722267151,
      "learning_rate": 0.00019996490510139155,
      "loss": 1.5452,
      "step": 754
    },
    {
      "epoch": 0.008444528951871777,
      "grad_norm": 0.06795412302017212,
      "learning_rate": 0.00019996481195520756,
      "loss": 1.5355,
      "step": 755
    },
    {
      "epoch": 0.008455713758430548,
      "grad_norm": 0.06670048087835312,
      "learning_rate": 0.00019996471868559832,
      "loss": 1.5529,
      "step": 756
    },
    {
      "epoch": 0.008466898564989318,
      "grad_norm": 0.06750231981277466,
      "learning_rate": 0.000199964625292564,
      "loss": 1.5626,
      "step": 757
    },
    {
      "epoch": 0.008478083371548088,
      "grad_norm": 0.06446841359138489,
      "learning_rate": 0.0001999645317761047,
      "loss": 1.5599,
      "step": 758
    },
    {
      "epoch": 0.00848926817810686,
      "grad_norm": 0.0593569353222847,
      "learning_rate": 0.00019996443813622057,
      "loss": 1.5538,
      "step": 759
    },
    {
      "epoch": 0.00850045298466563,
      "grad_norm": 0.0496729277074337,
      "learning_rate": 0.00019996434437291168,
      "loss": 1.5427,
      "step": 760
    },
    {
      "epoch": 0.008511637791224401,
      "grad_norm": 0.04995394125580788,
      "learning_rate": 0.00019996425048617814,
      "loss": 1.5326,
      "step": 761
    },
    {
      "epoch": 0.008522822597783171,
      "grad_norm": 0.061305031180381775,
      "learning_rate": 0.00019996415647602014,
      "loss": 1.5553,
      "step": 762
    },
    {
      "epoch": 0.008534007404341942,
      "grad_norm": 0.07046514004468918,
      "learning_rate": 0.0001999640623424377,
      "loss": 1.5305,
      "step": 763
    },
    {
      "epoch": 0.008545192210900712,
      "grad_norm": 0.0729839950799942,
      "learning_rate": 0.00019996396808543102,
      "loss": 1.5448,
      "step": 764
    },
    {
      "epoch": 0.008556377017459482,
      "grad_norm": 0.07597866654396057,
      "learning_rate": 0.00019996387370500016,
      "loss": 1.5339,
      "step": 765
    },
    {
      "epoch": 0.008567561824018254,
      "grad_norm": 0.07808911800384521,
      "learning_rate": 0.00019996377920114525,
      "loss": 1.5474,
      "step": 766
    },
    {
      "epoch": 0.008578746630577025,
      "grad_norm": 0.07077853381633759,
      "learning_rate": 0.0001999636845738664,
      "loss": 1.5456,
      "step": 767
    },
    {
      "epoch": 0.008589931437135795,
      "grad_norm": 0.05932854115962982,
      "learning_rate": 0.00019996358982316378,
      "loss": 1.5581,
      "step": 768
    },
    {
      "epoch": 0.008601116243694565,
      "grad_norm": 0.055467307567596436,
      "learning_rate": 0.00019996349494903743,
      "loss": 1.552,
      "step": 769
    },
    {
      "epoch": 0.008612301050253336,
      "grad_norm": 0.0684780701994896,
      "learning_rate": 0.0001999633999514875,
      "loss": 1.5691,
      "step": 770
    },
    {
      "epoch": 0.008623485856812106,
      "grad_norm": 0.07558051496744156,
      "learning_rate": 0.0001999633048305141,
      "loss": 1.5704,
      "step": 771
    },
    {
      "epoch": 0.008634670663370878,
      "grad_norm": 0.07451245933771133,
      "learning_rate": 0.0001999632095861174,
      "loss": 1.562,
      "step": 772
    },
    {
      "epoch": 0.008645855469929648,
      "grad_norm": 0.06852173060178757,
      "learning_rate": 0.00019996311421829744,
      "loss": 1.5582,
      "step": 773
    },
    {
      "epoch": 0.008657040276488418,
      "grad_norm": 0.05201677978038788,
      "learning_rate": 0.00019996301872705438,
      "loss": 1.5405,
      "step": 774
    },
    {
      "epoch": 0.008668225083047189,
      "grad_norm": 0.05331201106309891,
      "learning_rate": 0.00019996292311238832,
      "loss": 1.5389,
      "step": 775
    },
    {
      "epoch": 0.008679409889605959,
      "grad_norm": 0.06298915296792984,
      "learning_rate": 0.00019996282737429942,
      "loss": 1.5544,
      "step": 776
    },
    {
      "epoch": 0.00869059469616473,
      "grad_norm": 0.06354406476020813,
      "learning_rate": 0.00019996273151278774,
      "loss": 1.5654,
      "step": 777
    },
    {
      "epoch": 0.0087017795027235,
      "grad_norm": 0.0683928057551384,
      "learning_rate": 0.00019996263552785344,
      "loss": 1.5515,
      "step": 778
    },
    {
      "epoch": 0.008712964309282272,
      "grad_norm": 0.08236062526702881,
      "learning_rate": 0.0001999625394194966,
      "loss": 1.5617,
      "step": 779
    },
    {
      "epoch": 0.008724149115841042,
      "grad_norm": 0.07203904539346695,
      "learning_rate": 0.0001999624431877174,
      "loss": 1.555,
      "step": 780
    },
    {
      "epoch": 0.008735333922399812,
      "grad_norm": 0.06245394051074982,
      "learning_rate": 0.0001999623468325159,
      "loss": 1.5696,
      "step": 781
    },
    {
      "epoch": 0.008746518728958583,
      "grad_norm": 0.06451458483934402,
      "learning_rate": 0.00019996225035389222,
      "loss": 1.5695,
      "step": 782
    },
    {
      "epoch": 0.008757703535517353,
      "grad_norm": 0.06131128594279289,
      "learning_rate": 0.00019996215375184652,
      "loss": 1.5749,
      "step": 783
    },
    {
      "epoch": 0.008768888342076123,
      "grad_norm": 0.07188650965690613,
      "learning_rate": 0.00019996205702637888,
      "loss": 1.5647,
      "step": 784
    },
    {
      "epoch": 0.008780073148634894,
      "grad_norm": 0.0647532194852829,
      "learning_rate": 0.00019996196017748948,
      "loss": 1.5906,
      "step": 785
    },
    {
      "epoch": 0.008791257955193666,
      "grad_norm": 0.06955672800540924,
      "learning_rate": 0.00019996186320517836,
      "loss": 1.5923,
      "step": 786
    },
    {
      "epoch": 0.008802442761752436,
      "grad_norm": 0.08375248312950134,
      "learning_rate": 0.00019996176610944568,
      "loss": 1.5949,
      "step": 787
    },
    {
      "epoch": 0.008813627568311206,
      "grad_norm": 0.07389501482248306,
      "learning_rate": 0.00019996166889029156,
      "loss": 1.5859,
      "step": 788
    },
    {
      "epoch": 0.008824812374869977,
      "grad_norm": 0.08891887962818146,
      "learning_rate": 0.0001999615715477161,
      "loss": 1.5799,
      "step": 789
    },
    {
      "epoch": 0.008835997181428747,
      "grad_norm": 0.08630786836147308,
      "learning_rate": 0.00019996147408171948,
      "loss": 1.5648,
      "step": 790
    },
    {
      "epoch": 0.008847181987987517,
      "grad_norm": 0.09771011024713516,
      "learning_rate": 0.00019996137649230176,
      "loss": 1.5505,
      "step": 791
    },
    {
      "epoch": 0.008858366794546288,
      "grad_norm": 0.09192012995481491,
      "learning_rate": 0.00019996127877946307,
      "loss": 1.5704,
      "step": 792
    },
    {
      "epoch": 0.00886955160110506,
      "grad_norm": 0.07876724004745483,
      "learning_rate": 0.00019996118094320355,
      "loss": 1.5822,
      "step": 793
    },
    {
      "epoch": 0.00888073640766383,
      "grad_norm": 0.06203979253768921,
      "learning_rate": 0.00019996108298352328,
      "loss": 1.5599,
      "step": 794
    },
    {
      "epoch": 0.0088919212142226,
      "grad_norm": 0.06725753843784332,
      "learning_rate": 0.00019996098490042242,
      "loss": 1.562,
      "step": 795
    },
    {
      "epoch": 0.00890310602078137,
      "grad_norm": 0.07860880345106125,
      "learning_rate": 0.0001999608866939011,
      "loss": 1.5554,
      "step": 796
    },
    {
      "epoch": 0.008914290827340141,
      "grad_norm": 0.07922618836164474,
      "learning_rate": 0.0001999607883639594,
      "loss": 1.564,
      "step": 797
    },
    {
      "epoch": 0.008925475633898911,
      "grad_norm": 0.07509887218475342,
      "learning_rate": 0.0001999606899105975,
      "loss": 1.5531,
      "step": 798
    },
    {
      "epoch": 0.008936660440457682,
      "grad_norm": 0.0813961774110794,
      "learning_rate": 0.00019996059133381547,
      "loss": 1.5479,
      "step": 799
    },
    {
      "epoch": 0.008947845247016454,
      "grad_norm": 0.09687768667936325,
      "learning_rate": 0.00019996049263361343,
      "loss": 1.5527,
      "step": 800
    }
  ],
  "logging_steps": 1,
  "max_steps": 89407,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.6661988212201226e+20,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}