{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 1004,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00796812749003984,
      "grad_norm": 0.3647017180919647,
      "learning_rate": 1.9607843137254904e-07,
      "loss": 2.1478986740112305,
      "step": 2
    },
    {
      "epoch": 0.01593625498007968,
      "grad_norm": 0.24763226509094238,
      "learning_rate": 5.882352941176471e-07,
      "loss": 2.0969438552856445,
      "step": 4
    },
    {
      "epoch": 0.02390438247011952,
      "grad_norm": 0.9846194982528687,
      "learning_rate": 9.80392156862745e-07,
      "loss": 2.282338857650757,
      "step": 6
    },
    {
      "epoch": 0.03187250996015936,
      "grad_norm": 0.2201109677553177,
      "learning_rate": 1.3725490196078434e-06,
      "loss": 1.9020755290985107,
      "step": 8
    },
    {
      "epoch": 0.0398406374501992,
      "grad_norm": 0.5704253911972046,
      "learning_rate": 1.7647058823529414e-06,
      "loss": 1.4784005880355835,
      "step": 10
    },
    {
      "epoch": 0.04780876494023904,
      "grad_norm": 0.2062346190214157,
      "learning_rate": 2.1568627450980393e-06,
      "loss": 1.731598138809204,
      "step": 12
    },
    {
      "epoch": 0.055776892430278883,
      "grad_norm": 2.076277256011963,
      "learning_rate": 2.549019607843137e-06,
      "loss": 2.348186492919922,
      "step": 14
    },
    {
      "epoch": 0.06374501992031872,
      "grad_norm": 0.8406722545623779,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 2.586444854736328,
      "step": 16
    },
    {
      "epoch": 0.07171314741035857,
      "grad_norm": 0.3690650761127472,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.6810544729232788,
      "step": 18
    },
    {
      "epoch": 0.0796812749003984,
      "grad_norm": 0.18414445221424103,
      "learning_rate": 3.7254901960784316e-06,
      "loss": 1.6172391176223755,
      "step": 20
    },
    {
      "epoch": 0.08764940239043825,
      "grad_norm": 0.17469090223312378,
      "learning_rate": 4.11764705882353e-06,
      "loss": 1.25531005859375,
      "step": 22
    },
    {
      "epoch": 0.09561752988047809,
      "grad_norm": 1.353502631187439,
      "learning_rate": 4.509803921568628e-06,
      "loss": 1.4002299308776855,
      "step": 24
    },
    {
      "epoch": 0.10358565737051793,
      "grad_norm": 0.12341447174549103,
      "learning_rate": 4.901960784313726e-06,
      "loss": 1.5042188167572021,
      "step": 26
    },
    {
      "epoch": 0.11155378486055777,
      "grad_norm": 0.3027441203594208,
      "learning_rate": 5.294117647058824e-06,
      "loss": 1.6550602912902832,
      "step": 28
    },
    {
      "epoch": 0.11952191235059761,
      "grad_norm": 0.4921821057796478,
      "learning_rate": 5.686274509803922e-06,
      "loss": 1.3609421253204346,
      "step": 30
    },
    {
      "epoch": 0.12749003984063745,
      "grad_norm": 0.32808157801628113,
      "learning_rate": 6.07843137254902e-06,
      "loss": 1.1800657510757446,
      "step": 32
    },
    {
      "epoch": 0.13545816733067728,
      "grad_norm": 0.4257211983203888,
      "learning_rate": 6.470588235294119e-06,
      "loss": 1.305320143699646,
      "step": 34
    },
    {
      "epoch": 0.14342629482071714,
      "grad_norm": 0.132957324385643,
      "learning_rate": 6.862745098039216e-06,
      "loss": 1.4380031824111938,
      "step": 36
    },
    {
      "epoch": 0.15139442231075698,
      "grad_norm": 0.2522510588169098,
      "learning_rate": 7.2549019607843145e-06,
      "loss": 1.2642239332199097,
      "step": 38
    },
    {
      "epoch": 0.1593625498007968,
      "grad_norm": 0.2946537733078003,
      "learning_rate": 7.647058823529411e-06,
      "loss": 1.3502380847930908,
      "step": 40
    },
    {
      "epoch": 0.16733067729083664,
      "grad_norm": 0.2393134981393814,
      "learning_rate": 8.03921568627451e-06,
      "loss": 0.7276158332824707,
      "step": 42
    },
    {
      "epoch": 0.1752988047808765,
      "grad_norm": 0.16951222717761993,
      "learning_rate": 8.43137254901961e-06,
      "loss": 0.9695879817008972,
      "step": 44
    },
    {
      "epoch": 0.18326693227091634,
      "grad_norm": 0.20765990018844604,
      "learning_rate": 8.823529411764707e-06,
      "loss": 0.9420666098594666,
      "step": 46
    },
    {
      "epoch": 0.19123505976095617,
      "grad_norm": 0.5464557409286499,
      "learning_rate": 9.215686274509804e-06,
      "loss": 1.1855758428573608,
      "step": 48
    },
    {
      "epoch": 0.199203187250996,
      "grad_norm": 0.15976053476333618,
      "learning_rate": 9.607843137254903e-06,
      "loss": 1.3898124694824219,
      "step": 50
    },
    {
      "epoch": 0.20717131474103587,
      "grad_norm": 0.14374567568302155,
      "learning_rate": 1e-05,
      "loss": 1.3371469974517822,
      "step": 52
    },
    {
      "epoch": 0.2151394422310757,
      "grad_norm": 0.3317624032497406,
      "learning_rate": 9.999902196391087e-06,
      "loss": 0.9539063572883606,
      "step": 54
    },
    {
      "epoch": 0.22310756972111553,
      "grad_norm": 0.193589985370636,
      "learning_rate": 9.999608789815699e-06,
      "loss": 1.5122933387756348,
      "step": 56
    },
    {
      "epoch": 0.23107569721115537,
      "grad_norm": 0.20332682132720947,
      "learning_rate": 9.999119793027713e-06,
      "loss": 1.3573031425476074,
      "step": 58
    },
    {
      "epoch": 0.23904382470119523,
      "grad_norm": 0.11816702783107758,
      "learning_rate": 9.998435227282975e-06,
      "loss": 0.7735430598258972,
      "step": 60
    },
    {
      "epoch": 0.24701195219123506,
      "grad_norm": 0.12483549118041992,
      "learning_rate": 9.997555122338372e-06,
      "loss": 1.3591876029968262,
      "step": 62
    },
    {
      "epoch": 0.2549800796812749,
      "grad_norm": 0.4447817802429199,
      "learning_rate": 9.996479516450544e-06,
      "loss": 1.0984009504318237,
      "step": 64
    },
    {
      "epoch": 0.26294820717131473,
      "grad_norm": 0.09938079863786697,
      "learning_rate": 9.99520845637422e-06,
      "loss": 1.285765528678894,
      "step": 66
    },
    {
      "epoch": 0.27091633466135456,
      "grad_norm": 0.12960276007652283,
      "learning_rate": 9.993741997360185e-06,
      "loss": 0.7459665536880493,
      "step": 68
    },
    {
      "epoch": 0.2788844621513944,
      "grad_norm": 0.4153902530670166,
      "learning_rate": 9.992080203152872e-06,
      "loss": 1.610697627067566,
      "step": 70
    },
    {
      "epoch": 0.2868525896414343,
      "grad_norm": 0.19509944319725037,
      "learning_rate": 9.990223145987608e-06,
      "loss": 0.7440732717514038,
      "step": 72
    },
    {
      "epoch": 0.2948207171314741,
      "grad_norm": 0.2681001126766205,
      "learning_rate": 9.98817090658745e-06,
      "loss": 1.1801700592041016,
      "step": 74
    },
    {
      "epoch": 0.30278884462151395,
      "grad_norm": 0.1282671093940735,
      "learning_rate": 9.985923574159696e-06,
      "loss": 1.4409031867980957,
      "step": 76
    },
    {
      "epoch": 0.3107569721115538,
      "grad_norm": 0.4123835563659668,
      "learning_rate": 9.983481246392003e-06,
      "loss": 0.933936357498169,
      "step": 78
    },
    {
      "epoch": 0.3187250996015936,
      "grad_norm": 0.11349471658468246,
      "learning_rate": 9.980844029448135e-06,
      "loss": 0.809579074382782,
      "step": 80
    },
    {
      "epoch": 0.32669322709163345,
      "grad_norm": 0.2041967213153839,
      "learning_rate": 9.97801203796335e-06,
      "loss": 1.3015254735946655,
      "step": 82
    },
    {
      "epoch": 0.3346613545816733,
      "grad_norm": 0.315410852432251,
      "learning_rate": 9.974985395039423e-06,
      "loss": 0.8379510641098022,
      "step": 84
    },
    {
      "epoch": 0.3426294820717131,
      "grad_norm": 0.29619190096855164,
      "learning_rate": 9.971764232239287e-06,
      "loss": 1.319396734237671,
      "step": 86
    },
    {
      "epoch": 0.350597609561753,
      "grad_norm": 0.5492198467254639,
      "learning_rate": 9.968348689581315e-06,
      "loss": 0.8692243695259094,
      "step": 88
    },
    {
      "epoch": 0.35856573705179284,
      "grad_norm": 0.14868339896202087,
      "learning_rate": 9.964738915533242e-06,
      "loss": 1.327016830444336,
      "step": 90
    },
    {
      "epoch": 0.3665338645418327,
      "grad_norm": 0.24090534448623657,
      "learning_rate": 9.960935067005706e-06,
      "loss": 1.1243009567260742,
      "step": 92
    },
    {
      "epoch": 0.3745019920318725,
      "grad_norm": 0.12350978702306747,
      "learning_rate": 9.956937309345418e-06,
      "loss": 1.2415589094161987,
      "step": 94
    },
    {
      "epoch": 0.38247011952191234,
      "grad_norm": 0.26156526803970337,
      "learning_rate": 9.952745816327998e-06,
      "loss": 1.1342397928237915,
      "step": 96
    },
    {
      "epoch": 0.3904382470119522,
      "grad_norm": 0.12600702047348022,
      "learning_rate": 9.94836077015039e-06,
      "loss": 1.2638773918151855,
      "step": 98
    },
    {
      "epoch": 0.398406374501992,
      "grad_norm": 0.17442956566810608,
      "learning_rate": 9.943782361422975e-06,
      "loss": 0.8380853533744812,
      "step": 100
    },
    {
      "epoch": 0.4063745019920319,
      "grad_norm": 0.19549265503883362,
      "learning_rate": 9.939010789161259e-06,
      "loss": 1.274878978729248,
      "step": 102
    },
    {
      "epoch": 0.41434262948207173,
      "grad_norm": 0.14597226679325104,
      "learning_rate": 9.934046260777239e-06,
      "loss": 0.7483026385307312,
      "step": 104
    },
    {
      "epoch": 0.42231075697211157,
      "grad_norm": 0.24794237315654755,
      "learning_rate": 9.928888992070373e-06,
      "loss": 0.7552729845046997,
      "step": 106
    },
    {
      "epoch": 0.4302788844621514,
      "grad_norm": 0.1744687557220459,
      "learning_rate": 9.923539207218219e-06,
      "loss": 1.018141746520996,
      "step": 108
    },
    {
      "epoch": 0.43824701195219123,
      "grad_norm": 0.11073736101388931,
      "learning_rate": 9.91799713876667e-06,
      "loss": 1.285387635231018,
      "step": 110
    },
    {
      "epoch": 0.44621513944223107,
      "grad_norm": 0.35274383425712585,
      "learning_rate": 9.912263027619856e-06,
      "loss": 0.8079631328582764,
      "step": 112
    },
    {
      "epoch": 0.4541832669322709,
      "grad_norm": 0.22966524958610535,
      "learning_rate": 9.906337123029675e-06,
      "loss": 1.1002734899520874,
      "step": 114
    },
    {
      "epoch": 0.46215139442231074,
      "grad_norm": 0.1442008912563324,
      "learning_rate": 9.90021968258495e-06,
      "loss": 1.2883250713348389,
      "step": 116
    },
    {
      "epoch": 0.4701195219123506,
      "grad_norm": 0.11145839095115662,
      "learning_rate": 9.893910972200238e-06,
      "loss": 1.35982084274292,
      "step": 118
    },
    {
      "epoch": 0.47808764940239046,
      "grad_norm": 0.2911342680454254,
      "learning_rate": 9.887411266104268e-06,
      "loss": 1.2249188423156738,
      "step": 120
    },
    {
      "epoch": 0.4860557768924303,
      "grad_norm": 0.13076253235340118,
      "learning_rate": 9.880720846828026e-06,
      "loss": 0.9469150304794312,
      "step": 122
    },
    {
      "epoch": 0.4940239043824701,
      "grad_norm": 0.3520956039428711,
      "learning_rate": 9.873840005192466e-06,
      "loss": 1.0702083110809326,
      "step": 124
    },
    {
      "epoch": 0.50199203187251,
      "grad_norm": 0.8981730937957764,
      "learning_rate": 9.866769040295875e-06,
      "loss": 0.9181742668151855,
      "step": 126
    },
    {
      "epoch": 0.5099601593625498,
      "grad_norm": 0.15165890753269196,
      "learning_rate": 9.859508259500869e-06,
      "loss": 1.3017489910125732,
      "step": 128
    },
    {
      "epoch": 0.5179282868525896,
      "grad_norm": 0.14621900022029877,
      "learning_rate": 9.852057978421033e-06,
      "loss": 0.9123371243476868,
      "step": 130
    },
    {
      "epoch": 0.5258964143426295,
      "grad_norm": 0.2252267450094223,
      "learning_rate": 9.844418520907198e-06,
      "loss": 0.8035184741020203,
      "step": 132
    },
    {
      "epoch": 0.5338645418326693,
      "grad_norm": 0.14569298923015594,
      "learning_rate": 9.836590219033375e-06,
      "loss": 0.868083119392395,
      "step": 134
    },
    {
      "epoch": 0.5418326693227091,
      "grad_norm": 0.15642385184764862,
      "learning_rate": 9.828573413082306e-06,
      "loss": 1.2870720624923706,
      "step": 136
    },
    {
      "epoch": 0.549800796812749,
      "grad_norm": 0.1265447586774826,
      "learning_rate": 9.820368451530681e-06,
      "loss": 0.9235859513282776,
      "step": 138
    },
    {
      "epoch": 0.5577689243027888,
      "grad_norm": 0.27114516496658325,
      "learning_rate": 9.811975691033992e-06,
      "loss": 1.0814189910888672,
      "step": 140
    },
    {
      "epoch": 0.5657370517928287,
      "grad_norm": 0.20777811110019684,
      "learning_rate": 9.80339549641102e-06,
      "loss": 1.0009130239486694,
      "step": 142
    },
    {
      "epoch": 0.5737051792828686,
      "grad_norm": 0.08995456993579865,
      "learning_rate": 9.794628240627992e-06,
      "loss": 1.2293169498443604,
      "step": 144
    },
    {
      "epoch": 0.5816733067729084,
      "grad_norm": 0.2387453019618988,
      "learning_rate": 9.785674304782349e-06,
      "loss": 1.2766615152359009,
      "step": 146
    },
    {
      "epoch": 0.5896414342629482,
      "grad_norm": 0.18870815634727478,
      "learning_rate": 9.776534078086202e-06,
      "loss": 0.9375801086425781,
      "step": 148
    },
    {
      "epoch": 0.5976095617529881,
      "grad_norm": 0.32719048857688904,
      "learning_rate": 9.767207957849396e-06,
      "loss": 0.9444013833999634,
      "step": 150
    },
    {
      "epoch": 0.6055776892430279,
      "grad_norm": 0.4171900451183319,
      "learning_rate": 9.757696349462252e-06,
      "loss": 0.9264601469039917,
      "step": 152
    },
    {
      "epoch": 0.6135458167330677,
      "grad_norm": 0.32865679264068604,
      "learning_rate": 9.747999666377934e-06,
      "loss": 1.121659755706787,
      "step": 154
    },
    {
      "epoch": 0.6215139442231076,
      "grad_norm": 0.7469711899757385,
      "learning_rate": 9.738118330094491e-06,
      "loss": 0.9346869587898254,
      "step": 156
    },
    {
      "epoch": 0.6294820717131474,
      "grad_norm": 0.1958705335855484,
      "learning_rate": 9.728052770136518e-06,
      "loss": 1.1452645063400269,
      "step": 158
    },
    {
      "epoch": 0.6374501992031872,
      "grad_norm": 0.10420547425746918,
      "learning_rate": 9.717803424036501e-06,
      "loss": 0.9993031024932861,
      "step": 160
    },
    {
      "epoch": 0.6454183266932271,
      "grad_norm": 0.07185376435518265,
      "learning_rate": 9.707370737315793e-06,
      "loss": 0.7395276427268982,
      "step": 162
    },
    {
      "epoch": 0.6533864541832669,
      "grad_norm": 0.1622873842716217,
      "learning_rate": 9.69675516346524e-06,
      "loss": 1.1817888021469116,
      "step": 164
    },
    {
      "epoch": 0.6613545816733067,
      "grad_norm": 0.22871525585651398,
      "learning_rate": 9.685957163925479e-06,
      "loss": 0.8039273619651794,
      "step": 166
    },
    {
      "epoch": 0.6693227091633466,
      "grad_norm": 0.11571378260850906,
      "learning_rate": 9.67497720806688e-06,
      "loss": 1.2390202283859253,
      "step": 168
    },
    {
      "epoch": 0.6772908366533864,
      "grad_norm": 1.3822053670883179,
      "learning_rate": 9.663815773169131e-06,
      "loss": 0.9708461761474609,
      "step": 170
    },
    {
      "epoch": 0.6852589641434262,
      "grad_norm": 0.23121927678585052,
      "learning_rate": 9.652473344400507e-06,
      "loss": 0.9113867878913879,
      "step": 172
    },
    {
      "epoch": 0.6932270916334662,
      "grad_norm": 0.11104010790586472,
      "learning_rate": 9.640950414796772e-06,
      "loss": 1.2310372591018677,
      "step": 174
    },
    {
      "epoch": 0.701195219123506,
      "grad_norm": 1.821785807609558,
      "learning_rate": 9.629247485239747e-06,
      "loss": 0.7442381381988525,
      "step": 176
    },
    {
      "epoch": 0.7091633466135459,
      "grad_norm": 0.36091160774230957,
      "learning_rate": 9.617365064435544e-06,
      "loss": 0.9631052017211914,
      "step": 178
    },
    {
      "epoch": 0.7171314741035857,
      "grad_norm": 0.29177799820899963,
      "learning_rate": 9.605303668892443e-06,
      "loss": 1.0962485074996948,
      "step": 180
    },
    {
      "epoch": 0.7250996015936255,
      "grad_norm": 0.08216815441846848,
      "learning_rate": 9.593063822898451e-06,
      "loss": 0.4498317837715149,
      "step": 182
    },
    {
      "epoch": 0.7330677290836654,
      "grad_norm": 0.09393391013145447,
      "learning_rate": 9.580646058498505e-06,
      "loss": 1.3047553300857544,
      "step": 184
    },
    {
      "epoch": 0.7410358565737052,
      "grad_norm": 0.10853663086891174,
      "learning_rate": 9.56805091547135e-06,
      "loss": 1.2840948104858398,
      "step": 186
    },
    {
      "epoch": 0.749003984063745,
      "grad_norm": 0.09809323400259018,
      "learning_rate": 9.55527894130607e-06,
      "loss": 1.0851843357086182,
      "step": 188
    },
    {
      "epoch": 0.7569721115537849,
      "grad_norm": 0.17349843680858612,
      "learning_rate": 9.542330691178296e-06,
      "loss": 0.8616277575492859,
      "step": 190
    },
    {
      "epoch": 0.7649402390438247,
      "grad_norm": 0.19055721163749695,
      "learning_rate": 9.529206727926071e-06,
      "loss": 1.2394646406173706,
      "step": 192
    },
    {
      "epoch": 0.7729083665338645,
      "grad_norm": 0.16671667993068695,
      "learning_rate": 9.515907622025376e-06,
      "loss": 0.8858673572540283,
      "step": 194
    },
    {
      "epoch": 0.7808764940239044,
      "grad_norm": 0.3057395815849304,
      "learning_rate": 9.502433951565347e-06,
      "loss": 0.5642332434654236,
      "step": 196
    },
    {
      "epoch": 0.7888446215139442,
      "grad_norm": 0.49418577551841736,
      "learning_rate": 9.48878630222314e-06,
      "loss": 0.7914112210273743,
      "step": 198
    },
    {
      "epoch": 0.796812749003984,
      "grad_norm": 0.16775543987751007,
      "learning_rate": 9.474965267238469e-06,
      "loss": 1.0121368169784546,
      "step": 200
    },
    {
      "epoch": 0.8047808764940239,
      "grad_norm": 0.10699599981307983,
      "learning_rate": 9.460971447387824e-06,
      "loss": 0.9411302804946899,
      "step": 202
    },
    {
      "epoch": 0.8127490039840638,
      "grad_norm": 0.19434906542301178,
      "learning_rate": 9.446805450958353e-06,
      "loss": 1.5670157670974731,
      "step": 204
    },
    {
      "epoch": 0.8207171314741036,
      "grad_norm": 0.3301284909248352,
      "learning_rate": 9.43246789372142e-06,
      "loss": 1.2477513551712036,
      "step": 206
    },
    {
      "epoch": 0.8286852589641435,
      "grad_norm": 0.32433968782424927,
      "learning_rate": 9.417959398905846e-06,
      "loss": 0.6038308143615723,
      "step": 208
    },
    {
      "epoch": 0.8366533864541833,
      "grad_norm": 0.13645341992378235,
      "learning_rate": 9.40328059717081e-06,
      "loss": 1.2748695611953735,
      "step": 210
    },
    {
      "epoch": 0.8446215139442231,
      "grad_norm": 0.1219300627708435,
      "learning_rate": 9.388432126578438e-06,
      "loss": 1.22676682472229,
      "step": 212
    },
    {
      "epoch": 0.852589641434263,
      "grad_norm": 0.09014702588319778,
      "learning_rate": 9.373414632566066e-06,
      "loss": 0.9495640397071838,
      "step": 214
    },
    {
      "epoch": 0.8605577689243028,
      "grad_norm": 0.2150721698999405,
      "learning_rate": 9.35822876791819e-06,
      "loss": 1.4344321489334106,
      "step": 216
    },
    {
      "epoch": 0.8685258964143426,
      "grad_norm": 0.4374437928199768,
      "learning_rate": 9.342875192738083e-06,
      "loss": 0.8705391883850098,
      "step": 218
    },
    {
      "epoch": 0.8764940239043825,
      "grad_norm": 0.10829262435436249,
      "learning_rate": 9.327354574419109e-06,
      "loss": 1.2422717809677124,
      "step": 220
    },
    {
      "epoch": 0.8844621513944223,
      "grad_norm": 0.19894227385520935,
      "learning_rate": 9.311667587615701e-06,
      "loss": 1.4882192611694336,
      "step": 222
    },
    {
      "epoch": 0.8924302788844621,
      "grad_norm": 0.42940643429756165,
      "learning_rate": 9.29581491421405e-06,
      "loss": 0.716874897480011,
      "step": 224
    },
    {
      "epoch": 0.900398406374502,
      "grad_norm": 0.12247120589017868,
      "learning_rate": 9.279797243302453e-06,
      "loss": 0.9885092377662659,
      "step": 226
    },
    {
      "epoch": 0.9083665338645418,
      "grad_norm": 0.4069404900074005,
      "learning_rate": 9.263615271141362e-06,
      "loss": 1.0850290060043335,
      "step": 228
    },
    {
      "epoch": 0.9163346613545816,
      "grad_norm": 0.20781095325946808,
      "learning_rate": 9.247269701133125e-06,
      "loss": 1.1709660291671753,
      "step": 230
    },
    {
      "epoch": 0.9243027888446215,
      "grad_norm": 0.12009521573781967,
      "learning_rate": 9.230761243791401e-06,
      "loss": 1.3472001552581787,
      "step": 232
    },
    {
      "epoch": 0.9322709163346613,
      "grad_norm": 0.11444836854934692,
      "learning_rate": 9.214090616710282e-06,
      "loss": 0.7829840779304504,
      "step": 234
    },
    {
      "epoch": 0.9402390438247012,
      "grad_norm": 0.11701781302690506,
      "learning_rate": 9.197258544533098e-06,
      "loss": 0.7907887101173401,
      "step": 236
    },
    {
      "epoch": 0.9482071713147411,
      "grad_norm": 0.21362987160682678,
      "learning_rate": 9.180265758920916e-06,
      "loss": 0.44280657172203064,
      "step": 238
    },
    {
      "epoch": 0.9561752988047809,
      "grad_norm": 0.20342251658439636,
      "learning_rate": 9.163112998520741e-06,
      "loss": 0.8508027195930481,
      "step": 240
    },
    {
      "epoch": 0.9641434262948207,
      "grad_norm": 0.20921140909194946,
      "learning_rate": 9.145801008933403e-06,
      "loss": 1.1112576723098755,
      "step": 242
    },
    {
      "epoch": 0.9721115537848606,
      "grad_norm": 0.13130858540534973,
      "learning_rate": 9.128330542681154e-06,
      "loss": 1.215651273727417,
      "step": 244
    },
    {
      "epoch": 0.9800796812749004,
      "grad_norm": 0.2843606770038605,
      "learning_rate": 9.110702359174944e-06,
      "loss": 0.9460815787315369,
      "step": 246
    },
    {
      "epoch": 0.9880478087649402,
      "grad_norm": 0.12077733874320984,
      "learning_rate": 9.092917224681427e-06,
      "loss": 0.9816232919692993,
      "step": 248
    },
    {
      "epoch": 0.9960159362549801,
      "grad_norm": 0.1792304813861847,
      "learning_rate": 9.074975912289641e-06,
      "loss": 0.9759309887886047,
      "step": 250
    },
    {
      "epoch": 1.00398406374502,
      "grad_norm": 0.13639993965625763,
      "learning_rate": 9.056879201877412e-06,
      "loss": 0.8936959505081177,
      "step": 252
    },
    {
      "epoch": 1.0119521912350598,
      "grad_norm": 0.1269155740737915,
      "learning_rate": 9.038627880077441e-06,
      "loss": 0.6980884671211243,
      "step": 254
    },
    {
      "epoch": 1.0199203187250996,
      "grad_norm": 0.09299927204847336,
      "learning_rate": 9.020222740243127e-06,
      "loss": 1.15306556224823,
      "step": 256
    },
    {
      "epoch": 1.0278884462151394,
      "grad_norm": 0.14957579970359802,
      "learning_rate": 9.001664582414066e-06,
      "loss": 1.1301281452178955,
      "step": 258
    },
    {
      "epoch": 1.0358565737051793,
      "grad_norm": 0.27969855070114136,
      "learning_rate": 8.982954213281291e-06,
      "loss": 1.0512151718139648,
      "step": 260
    },
    {
      "epoch": 1.043824701195219,
      "grad_norm": 0.15900252759456635,
      "learning_rate": 8.964092446152183e-06,
      "loss": 0.9678531885147095,
      "step": 262
    },
    {
      "epoch": 1.051792828685259,
      "grad_norm": 0.27932503819465637,
      "learning_rate": 8.945080100915145e-06,
      "loss": 1.046820044517517,
      "step": 264
    },
    {
      "epoch": 1.0597609561752988,
      "grad_norm": 0.23315030336380005,
      "learning_rate": 8.925918004003941e-06,
      "loss": 0.6050488352775574,
      "step": 266
    },
    {
      "epoch": 1.0677290836653386,
      "grad_norm": 0.09495838731527328,
      "learning_rate": 8.90660698836179e-06,
      "loss": 0.9008283019065857,
      "step": 268
    },
    {
      "epoch": 1.0756972111553784,
      "grad_norm": 0.35869407653808594,
      "learning_rate": 8.887147893405145e-06,
      "loss": 0.5234286785125732,
      "step": 270
    },
    {
      "epoch": 1.0836653386454183,
      "grad_norm": 0.15390193462371826,
      "learning_rate": 8.867541564987212e-06,
      "loss": 1.230911135673523,
      "step": 272
    },
    {
      "epoch": 1.091633466135458,
      "grad_norm": 0.25453439354896545,
      "learning_rate": 8.84778885536118e-06,
      "loss": 0.5974389910697937,
      "step": 274
    },
    {
      "epoch": 1.099601593625498,
      "grad_norm": 0.13244271278381348,
      "learning_rate": 8.827890623143172e-06,
      "loss": 0.890313446521759,
      "step": 276
    },
    {
      "epoch": 1.1075697211155378,
      "grad_norm": 0.10025046765804291,
      "learning_rate": 8.80784773327494e-06,
      "loss": 0.7988033890724182,
      "step": 278
    },
    {
      "epoch": 1.1155378486055776,
      "grad_norm": 0.1597757637500763,
      "learning_rate": 8.787661056986236e-06,
      "loss": 0.9830846786499023,
      "step": 280
    },
    {
      "epoch": 1.1235059760956174,
      "grad_norm": 0.1170232743024826,
      "learning_rate": 8.767331471756975e-06,
      "loss": 1.138309359550476,
      "step": 282
    },
    {
      "epoch": 1.1314741035856573,
      "grad_norm": 0.22332733869552612,
      "learning_rate": 8.746859861279068e-06,
      "loss": 1.1853747367858887,
      "step": 284
    },
    {
      "epoch": 1.139442231075697,
      "grad_norm": 0.06254544109106064,
      "learning_rate": 8.726247115418017e-06,
      "loss": 0.2661963999271393,
      "step": 286
    },
    {
      "epoch": 1.1474103585657371,
      "grad_norm": 0.09349869191646576,
      "learning_rate": 8.70549413017424e-06,
      "loss": 0.6934034824371338,
      "step": 288
    },
    {
      "epoch": 1.155378486055777,
      "grad_norm": 0.6584972143173218,
      "learning_rate": 8.684601807644114e-06,
      "loss": 0.9061666131019592,
      "step": 290
    },
    {
      "epoch": 1.1633466135458168,
      "grad_norm": 0.2169235199689865,
      "learning_rate": 8.663571055980769e-06,
      "loss": 0.497935026884079,
      "step": 292
    },
    {
      "epoch": 1.1713147410358566,
      "grad_norm": 0.08594366908073425,
      "learning_rate": 8.642402789354609e-06,
      "loss": 1.1939326524734497,
      "step": 294
    },
    {
      "epoch": 1.1792828685258965,
      "grad_norm": 0.08441396057605743,
      "learning_rate": 8.62109792791358e-06,
      "loss": 1.254359245300293,
      "step": 296
    },
    {
      "epoch": 1.1872509960159363,
      "grad_norm": 0.08592527359724045,
      "learning_rate": 8.599657397743163e-06,
      "loss": 1.3354953527450562,
      "step": 298
    },
    {
      "epoch": 1.1952191235059761,
      "grad_norm": 0.9391301274299622,
      "learning_rate": 8.578082130826123e-06,
      "loss": 0.5832847952842712,
      "step": 300
    },
    {
      "epoch": 1.203187250996016,
      "grad_norm": 0.14405034482479095,
      "learning_rate": 8.556373065002006e-06,
      "loss": 0.8736499547958374,
      "step": 302
    },
    {
      "epoch": 1.2111553784860558,
      "grad_norm": 0.12587150931358337,
      "learning_rate": 8.534531143926359e-06,
      "loss": 0.4843589961528778,
      "step": 304
    },
    {
      "epoch": 1.2191235059760956,
      "grad_norm": 0.09853191673755646,
      "learning_rate": 8.512557317029715e-06,
      "loss": 1.1760183572769165,
      "step": 306
    },
    {
      "epoch": 1.2270916334661355,
      "grad_norm": 0.3153192102909088,
      "learning_rate": 8.49045253947633e-06,
      "loss": 0.984307050704956,
      "step": 308
    },
    {
      "epoch": 1.2350597609561753,
      "grad_norm": 0.1911332905292511,
      "learning_rate": 8.46821777212266e-06,
      "loss": 1.1619549989700317,
      "step": 310
    },
    {
      "epoch": 1.2430278884462151,
      "grad_norm": 0.4250432252883911,
      "learning_rate": 8.445853981475587e-06,
      "loss": 1.0535634756088257,
      "step": 312
    },
    {
      "epoch": 1.250996015936255,
      "grad_norm": 0.2402748465538025,
      "learning_rate": 8.423362139650414e-06,
      "loss": 0.8381243348121643,
      "step": 314
    },
    {
      "epoch": 1.2589641434262948,
      "grad_norm": 0.573125422000885,
      "learning_rate": 8.40074322432861e-06,
      "loss": 0.719307541847229,
      "step": 316
    },
    {
      "epoch": 1.2669322709163346,
      "grad_norm": 0.19297780096530914,
      "learning_rate": 8.377998218715311e-06,
      "loss": 0.5993711948394775,
      "step": 318
    },
    {
      "epoch": 1.2749003984063745,
      "grad_norm": 0.1421806365251541,
      "learning_rate": 8.355128111496569e-06,
      "loss": 0.46684369444847107,
      "step": 320
    },
    {
      "epoch": 1.2828685258964143,
      "grad_norm": 0.4090231955051422,
      "learning_rate": 8.332133896796399e-06,
      "loss": 0.9673095345497131,
      "step": 322
    },
    {
      "epoch": 1.2908366533864541,
      "grad_norm": 0.12756745517253876,
      "learning_rate": 8.309016574133545e-06,
      "loss": 0.9027948975563049,
      "step": 324
    },
    {
      "epoch": 1.298804780876494,
      "grad_norm": 0.20138461887836456,
      "learning_rate": 8.28577714837805e-06,
      "loss": 1.2605825662612915,
      "step": 326
    },
    {
      "epoch": 1.3067729083665338,
      "grad_norm": 0.09830658882856369,
      "learning_rate": 8.262416629707555e-06,
      "loss": 0.6810956001281738,
      "step": 328
    },
    {
      "epoch": 1.3147410358565736,
      "grad_norm": 0.12664003670215607,
      "learning_rate": 8.238936033563415e-06,
      "loss": 0.8375826478004456,
      "step": 330
    },
    {
      "epoch": 1.3227091633466135,
      "grad_norm": 0.5658034682273865,
      "learning_rate": 8.215336380606533e-06,
      "loss": 0.9011093974113464,
      "step": 332
    },
    {
      "epoch": 1.3306772908366533,
      "grad_norm": 0.1833939105272293,
      "learning_rate": 8.191618696673012e-06,
      "loss": 0.6151177883148193,
      "step": 334
    },
    {
      "epoch": 1.3386454183266931,
      "grad_norm": 3.7321767807006836,
      "learning_rate": 8.167784012729555e-06,
      "loss": 0.8623945713043213,
      "step": 336
    },
    {
      "epoch": 1.3466135458167332,
      "grad_norm": 0.10215169191360474,
      "learning_rate": 8.143833364828654e-06,
      "loss": 1.1351113319396973,
      "step": 338
    },
    {
      "epoch": 1.354581673306773,
      "grad_norm": 0.06890127807855606,
      "learning_rate": 8.119767794063553e-06,
      "loss": 1.1333948373794556,
      "step": 340
    },
    {
      "epoch": 1.3625498007968129,
      "grad_norm": 0.09315377473831177,
      "learning_rate": 8.095588346523e-06,
      "loss": 1.1782711744308472,
      "step": 342
    },
    {
      "epoch": 1.3705179282868527,
      "grad_norm": 0.11636896431446075,
      "learning_rate": 8.071296073245759e-06,
      "loss": 1.1379573345184326,
      "step": 344
    },
    {
      "epoch": 1.3784860557768925,
      "grad_norm": 0.12263190001249313,
      "learning_rate": 8.046892030174938e-06,
      "loss": 0.6890803575515747,
      "step": 346
    },
    {
      "epoch": 1.3864541832669324,
      "grad_norm": 0.09175064414739609,
      "learning_rate": 8.022377278112089e-06,
      "loss": 1.159577488899231,
      "step": 348
    },
    {
      "epoch": 1.3944223107569722,
      "grad_norm": 0.14179781079292297,
      "learning_rate": 7.997752882671088e-06,
      "loss": 0.6742680668830872,
      "step": 350
    },
    {
      "epoch": 1.402390438247012,
      "grad_norm": 0.09603075683116913,
      "learning_rate": 7.97301991423182e-06,
      "loss": 1.183809757232666,
      "step": 352
    },
    {
      "epoch": 1.4103585657370519,
      "grad_norm": 0.23190699517726898,
      "learning_rate": 7.948179447893652e-06,
      "loss": 1.1228028535842896,
      "step": 354
    },
    {
      "epoch": 1.4183266932270917,
      "grad_norm": 0.41616132855415344,
      "learning_rate": 7.923232563428693e-06,
      "loss": 0.8794443011283875,
      "step": 356
    },
    {
      "epoch": 1.4262948207171315,
      "grad_norm": 0.27986449003219604,
      "learning_rate": 7.898180345234875e-06,
      "loss": 1.1482864618301392,
      "step": 358
    },
    {
      "epoch": 1.4342629482071714,
      "grad_norm": 0.12306718528270721,
      "learning_rate": 7.873023882288793e-06,
      "loss": 0.5888344049453735,
      "step": 360
    },
    {
      "epoch": 1.4422310756972112,
      "grad_norm": 0.10971551388502121,
      "learning_rate": 7.847764268098387e-06,
      "loss": 0.6951230764389038,
      "step": 362
    },
    {
      "epoch": 1.450199203187251,
      "grad_norm": 0.15229885280132294,
      "learning_rate": 7.822402600655405e-06,
      "loss": 1.0400999784469604,
      "step": 364
    },
    {
      "epoch": 1.4581673306772909,
      "grad_norm": 0.0946863442659378,
      "learning_rate": 7.79693998238767e-06,
      "loss": 1.1122182607650757,
      "step": 366
    },
    {
      "epoch": 1.4661354581673307,
      "grad_norm": 0.08339293301105499,
      "learning_rate": 7.771377520111163e-06,
      "loss": 0.604324460029602,
      "step": 368
    },
    {
      "epoch": 1.4741035856573705,
      "grad_norm": 0.16374002397060394,
      "learning_rate": 7.745716324981912e-06,
      "loss": 0.684520423412323,
      "step": 370
    },
    {
      "epoch": 1.4820717131474104,
      "grad_norm": 0.22385168075561523,
      "learning_rate": 7.71995751244769e-06,
      "loss": 1.2398829460144043,
      "step": 372
    },
    {
      "epoch": 1.4900398406374502,
      "grad_norm": 0.09120070189237595,
      "learning_rate": 7.694102202199536e-06,
      "loss": 1.1202857494354248,
      "step": 374
    },
    {
      "epoch": 1.49800796812749,
      "grad_norm": 0.15905345976352692,
      "learning_rate": 7.66815151812307e-06,
      "loss": 1.1247127056121826,
      "step": 376
    },
    {
      "epoch": 1.5059760956175299,
      "grad_norm": 0.24509046971797943,
      "learning_rate": 7.642106588249651e-06,
      "loss": 0.8915665745735168,
      "step": 378
    },
    {
      "epoch": 1.5139442231075697,
      "grad_norm": 0.11813104152679443,
      "learning_rate": 7.6159685447073376e-06,
      "loss": 1.1169109344482422,
      "step": 380
    },
    {
      "epoch": 1.5219123505976095,
      "grad_norm": 0.10362762212753296,
      "learning_rate": 7.589738523671685e-06,
      "loss": 0.7195178866386414,
      "step": 382
    },
    {
      "epoch": 1.5298804780876494,
      "grad_norm": 0.13230100274085999,
      "learning_rate": 7.563417665316338e-06,
      "loss": 0.7218353152275085,
      "step": 384
    },
    {
      "epoch": 1.5378486055776892,
      "grad_norm": 0.4582996964454651,
      "learning_rate": 7.5370071137634984e-06,
      "loss": 1.0153872966766357,
      "step": 386
    },
    {
      "epoch": 1.545816733067729,
      "grad_norm": 0.17963430285453796,
      "learning_rate": 7.510508017034165e-06,
      "loss": 1.1490799188613892,
      "step": 388
    },
    {
      "epoch": 1.5537848605577689,
      "grad_norm": 0.17351095378398895,
      "learning_rate": 7.483921526998247e-06,
      "loss": 0.5256304740905762,
      "step": 390
    },
    {
      "epoch": 1.5617529880478087,
      "grad_norm": 0.15164697170257568,
      "learning_rate": 7.4572487993244865e-06,
      "loss": 0.5711215138435364,
      "step": 392
    },
    {
      "epoch": 1.5697211155378485,
      "grad_norm": 0.9964547753334045,
      "learning_rate": 7.430490993430232e-06,
      "loss": 1.0898221731185913,
      "step": 394
    },
    {
      "epoch": 1.5776892430278884,
      "grad_norm": 0.19943900406360626,
      "learning_rate": 7.403649272431027e-06,
      "loss": 1.0307166576385498,
      "step": 396
    },
    {
      "epoch": 1.5856573705179282,
      "grad_norm": 0.0867285281419754,
      "learning_rate": 7.376724803090069e-06,
      "loss": 1.1423487663269043,
      "step": 398
    },
    {
      "epoch": 1.593625498007968,
      "grad_norm": 1.1143836975097656,
      "learning_rate": 7.349718755767473e-06,
      "loss": 0.5850595235824585,
      "step": 400
    },
    {
      "epoch": 1.6015936254980079,
      "grad_norm": 0.3400084972381592,
      "learning_rate": 7.322632304369416e-06,
      "loss": 1.1260920763015747,
      "step": 402
    },
    {
      "epoch": 1.6095617529880477,
      "grad_norm": 0.11252368986606598,
      "learning_rate": 7.2954666262970965e-06,
      "loss": 1.2705897092819214,
      "step": 404
    },
    {
      "epoch": 1.6175298804780875,
      "grad_norm": 0.1879090517759323,
      "learning_rate": 7.268222902395562e-06,
      "loss": 0.7150183320045471,
      "step": 406
    },
    {
      "epoch": 1.6254980079681274,
      "grad_norm": 0.07393921166658401,
      "learning_rate": 7.240902316902373e-06,
      "loss": 0.5556809306144714,
      "step": 408
    },
    {
      "epoch": 1.6334661354581672,
      "grad_norm": 0.2739734649658203,
      "learning_rate": 7.213506057396134e-06,
      "loss": 0.767079770565033,
      "step": 410
    },
    {
      "epoch": 1.641434262948207,
      "grad_norm": 0.12828262150287628,
      "learning_rate": 7.186035314744872e-06,
      "loss": 0.4897518754005432,
      "step": 412
    },
    {
      "epoch": 1.6494023904382469,
      "grad_norm": 0.0791148766875267,
      "learning_rate": 7.1584912830542575e-06,
      "loss": 0.8231608271598816,
      "step": 414
    },
    {
      "epoch": 1.6573705179282867,
      "grad_norm": 0.19900332391262054,
      "learning_rate": 7.130875159615716e-06,
      "loss": 0.7192663550376892,
      "step": 416
    },
    {
      "epoch": 1.6653386454183265,
      "grad_norm": 0.21691080927848816,
      "learning_rate": 7.103188144854376e-06,
      "loss": 0.9949262738227844,
      "step": 418
    },
    {
      "epoch": 1.6733067729083664,
      "grad_norm": 0.08056575804948807,
      "learning_rate": 7.075431442276888e-06,
      "loss": 1.1483328342437744,
      "step": 420
    },
    {
      "epoch": 1.6812749003984062,
      "grad_norm": 0.11413298547267914,
      "learning_rate": 7.047606258419108e-06,
      "loss": 1.019271731376648,
      "step": 422
    },
    {
      "epoch": 1.6892430278884463,
      "grad_norm": 0.13839687407016754,
      "learning_rate": 7.019713802793663e-06,
      "loss": 0.5812303423881531,
      "step": 424
    },
    {
      "epoch": 1.697211155378486,
      "grad_norm": 0.0764610543847084,
      "learning_rate": 6.99175528783736e-06,
      "loss": 1.199522614479065,
      "step": 426
    },
    {
      "epoch": 1.705179282868526,
      "grad_norm": 0.07787472009658813,
      "learning_rate": 6.963731928858495e-06,
      "loss": 1.1365389823913574,
      "step": 428
    },
    {
      "epoch": 1.7131474103585658,
      "grad_norm": 0.17501701414585114,
      "learning_rate": 6.935644943984017e-06,
      "loss": 0.7606980204582214,
      "step": 430
    },
    {
      "epoch": 1.7211155378486056,
      "grad_norm": 0.5705646276473999,
      "learning_rate": 6.9074955541065865e-06,
      "loss": 0.902596116065979,
      "step": 432
    },
    {
      "epoch": 1.7290836653386454,
      "grad_norm": 1.084190011024475,
      "learning_rate": 6.8792849828315054e-06,
      "loss": 0.6085677742958069,
      "step": 434
    },
    {
      "epoch": 1.7370517928286853,
      "grad_norm": 0.434283047914505,
      "learning_rate": 6.851014456423515e-06,
      "loss": 0.3681115508079529,
      "step": 436
    },
    {
      "epoch": 1.745019920318725,
      "grad_norm": 0.1507345736026764,
      "learning_rate": 6.822685203753508e-06,
      "loss": 0.7172828316688538,
      "step": 438
    },
    {
      "epoch": 1.752988047808765,
      "grad_norm": 0.33439791202545166,
      "learning_rate": 6.794298456245111e-06,
      "loss": 0.6734229326248169,
      "step": 440
    },
    {
      "epoch": 1.7609561752988048,
      "grad_norm": 0.2586667835712433,
      "learning_rate": 6.765855447821144e-06,
      "loss": 1.4859187602996826,
      "step": 442
    },
    {
      "epoch": 1.7689243027888446,
      "grad_norm": 0.16227927803993225,
      "learning_rate": 6.737357414849993e-06,
      "loss": 0.7400438785552979,
      "step": 444
    },
    {
      "epoch": 1.7768924302788844,
      "grad_norm": 0.11800998449325562,
      "learning_rate": 6.70880559609187e-06,
      "loss": 1.030636191368103,
      "step": 446
    },
    {
      "epoch": 1.7848605577689243,
      "grad_norm": 0.6456466317176819,
      "learning_rate": 6.680201232644958e-06,
      "loss": 1.1084392070770264,
      "step": 448
    },
    {
      "epoch": 1.792828685258964,
      "grad_norm": 0.14494922757148743,
      "learning_rate": 6.651545567891471e-06,
      "loss": 0.8555322885513306,
      "step": 450
    },
    {
      "epoch": 1.800796812749004,
      "grad_norm": 0.05881403386592865,
      "learning_rate": 6.622839847443596e-06,
      "loss": 0.493237167596817,
      "step": 452
    },
    {
      "epoch": 1.8087649402390438,
      "grad_norm": 0.2267322540283203,
      "learning_rate": 6.59408531908936e-06,
      "loss": 0.982709527015686,
      "step": 454
    },
    {
      "epoch": 1.8167330677290838,
      "grad_norm": 3.470564126968384,
      "learning_rate": 6.565283232738386e-06,
      "loss": 0.6639694571495056,
      "step": 456
    },
    {
      "epoch": 1.8247011952191237,
      "grad_norm": 0.12834540009498596,
      "learning_rate": 6.536434840367555e-06,
      "loss": 1.0367791652679443,
      "step": 458
    },
    {
      "epoch": 1.8326693227091635,
      "grad_norm": 0.18368105590343475,
      "learning_rate": 6.507541395966591e-06,
      "loss": 0.7434819340705872,
      "step": 460
    },
    {
      "epoch": 1.8406374501992033,
      "grad_norm": 0.08568334579467773,
      "learning_rate": 6.47860415548356e-06,
      "loss": 0.8052833676338196,
      "step": 462
    },
    {
      "epoch": 1.8486055776892432,
      "grad_norm": 0.11108598113059998,
      "learning_rate": 6.449624376770259e-06,
      "loss": 0.12546391785144806,
      "step": 464
    },
    {
      "epoch": 1.856573705179283,
      "grad_norm": 0.20673945546150208,
      "learning_rate": 6.420603319527554e-06,
      "loss": 1.1594171524047852,
      "step": 466
    },
    {
      "epoch": 1.8645418326693228,
      "grad_norm": 0.1927093267440796,
      "learning_rate": 6.3915422452506146e-06,
      "loss": 0.5444157719612122,
      "step": 468
    },
    {
      "epoch": 1.8725099601593627,
      "grad_norm": 0.21195177733898163,
      "learning_rate": 6.362442417174082e-06,
      "loss": 1.1595547199249268,
      "step": 470
    },
    {
      "epoch": 1.8804780876494025,
      "grad_norm": 0.34100231528282166,
      "learning_rate": 6.33330510021716e-06,
      "loss": 1.0272108316421509,
      "step": 472
    },
    {
      "epoch": 1.8884462151394423,
      "grad_norm": 0.3166549503803253,
      "learning_rate": 6.304131560928622e-06,
      "loss": 1.1524168252944946,
      "step": 474
    },
    {
      "epoch": 1.8964143426294822,
      "grad_norm": 0.07201140373945236,
      "learning_rate": 6.274923067431775e-06,
      "loss": 0.949006974697113,
      "step": 476
    },
    {
      "epoch": 1.904382470119522,
      "grad_norm": 0.07126747071743011,
      "learning_rate": 6.2456808893693155e-06,
      "loss": 0.8311147093772888,
      "step": 478
    },
    {
      "epoch": 1.9123505976095618,
      "grad_norm": 0.9960137605667114,
      "learning_rate": 6.216406297848155e-06,
      "loss": 0.5735684633255005,
      "step": 480
    },
    {
      "epoch": 1.9203187250996017,
      "grad_norm": 0.3506461977958679,
      "learning_rate": 6.187100565384162e-06,
      "loss": 0.7413168549537659,
      "step": 482
    },
    {
      "epoch": 1.9282868525896415,
      "grad_norm": 0.07256942987442017,
      "learning_rate": 6.157764965846841e-06,
      "loss": 1.1099603176116943,
      "step": 484
    },
    {
      "epoch": 1.9362549800796813,
      "grad_norm": 0.13000288605690002,
      "learning_rate": 6.128400774403974e-06,
      "loss": 1.0959395170211792,
      "step": 486
    },
    {
      "epoch": 1.9442231075697212,
      "grad_norm": 0.06632355600595474,
      "learning_rate": 6.0990092674661834e-06,
      "loss": 1.102731704711914,
      "step": 488
    },
    {
      "epoch": 1.952191235059761,
      "grad_norm": 0.0742376446723938,
      "learning_rate": 6.069591722631446e-06,
      "loss": 1.1146429777145386,
      "step": 490
    },
    {
      "epoch": 1.9601593625498008,
      "grad_norm": 0.09307283163070679,
      "learning_rate": 6.040149418629562e-06,
      "loss": 0.7695032358169556,
      "step": 492
    },
    {
      "epoch": 1.9681274900398407,
      "grad_norm": 1.1231991052627563,
      "learning_rate": 6.010683635266569e-06,
      "loss": 0.705299973487854,
      "step": 494
    },
    {
      "epoch": 1.9760956175298805,
      "grad_norm": 0.11147436499595642,
      "learning_rate": 5.981195653369113e-06,
      "loss": 1.0949732065200806,
      "step": 496
    },
    {
      "epoch": 1.9840637450199203,
      "grad_norm": 0.5126568078994751,
      "learning_rate": 5.95168675472877e-06,
      "loss": 0.8323699235916138,
      "step": 498
    },
    {
      "epoch": 1.9920318725099602,
      "grad_norm": 0.12495867908000946,
      "learning_rate": 5.922158222046334e-06,
      "loss": 1.1727375984191895,
      "step": 500
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.17220130562782288,
      "learning_rate": 5.892611338876053e-06,
      "loss": 0.6304442286491394,
      "step": 502
    },
    {
      "epoch": 2.00796812749004,
      "grad_norm": 0.10112476348876953,
      "learning_rate": 5.8630473895698426e-06,
      "loss": 0.6072199940681458,
      "step": 504
    },
    {
      "epoch": 2.0159362549800797,
      "grad_norm": 0.23855921626091003,
      "learning_rate": 5.833467659221452e-06,
      "loss": 0.5671996474266052,
      "step": 506
    },
    {
      "epoch": 2.0239043824701195,
      "grad_norm": 0.028424056246876717,
      "learning_rate": 5.803873433610604e-06,
      "loss": 0.5381027460098267,
      "step": 508
    },
    {
      "epoch": 2.0318725099601593,
      "grad_norm": 0.19135037064552307,
      "learning_rate": 5.774265999147108e-06,
      "loss": 0.8364197611808777,
      "step": 510
    },
    {
      "epoch": 2.039840637450199,
      "grad_norm": 0.10862982273101807,
      "learning_rate": 5.744646642814935e-06,
      "loss": 0.7114787101745605,
      "step": 512
    },
    {
      "epoch": 2.047808764940239,
      "grad_norm": 0.23314134776592255,
      "learning_rate": 5.715016652116284e-06,
      "loss": 0.31099826097488403,
      "step": 514
    },
    {
      "epoch": 2.055776892430279,
      "grad_norm": 0.08411658555269241,
      "learning_rate": 5.68537731501561e-06,
      "loss": 0.8799335956573486,
      "step": 516
    },
    {
      "epoch": 2.0637450199203187,
      "grad_norm": 0.5435796976089478,
      "learning_rate": 5.655729919883639e-06,
      "loss": 0.5037041902542114,
      "step": 518
    },
    {
      "epoch": 2.0717131474103585,
      "grad_norm": 0.33835330605506897,
      "learning_rate": 5.626075755441366e-06,
      "loss": 0.7988424897193909,
      "step": 520
    },
    {
      "epoch": 2.0796812749003983,
      "grad_norm": 0.1254897266626358,
      "learning_rate": 5.596416110704034e-06,
      "loss": 0.9907560348510742,
      "step": 522
    },
    {
      "epoch": 2.087649402390438,
      "grad_norm": 0.2384515255689621,
      "learning_rate": 5.566752274925109e-06,
      "loss": 0.8815075755119324,
      "step": 524
    },
    {
      "epoch": 2.095617529880478,
      "grad_norm": 0.26632511615753174,
      "learning_rate": 5.5370855375402345e-06,
      "loss": 0.4096245765686035,
      "step": 526
    },
    {
      "epoch": 2.103585657370518,
      "grad_norm": 0.14012040197849274,
      "learning_rate": 5.507417188111176e-06,
      "loss": 0.7404940724372864,
      "step": 528
    },
    {
      "epoch": 2.1115537848605577,
      "grad_norm": 0.08850028365850449,
      "learning_rate": 5.477748516269779e-06,
      "loss": 0.6362339854240417,
      "step": 530
    },
    {
      "epoch": 2.1195219123505975,
      "grad_norm": 0.14766618609428406,
      "learning_rate": 5.448080811661897e-06,
      "loss": 0.20267254114151,
      "step": 532
    },
    {
      "epoch": 2.1274900398406373,
      "grad_norm": 0.17179229855537415,
      "learning_rate": 5.418415363891346e-06,
      "loss": 0.7317008376121521,
      "step": 534
    },
    {
      "epoch": 2.135458167330677,
      "grad_norm": 0.3195139765739441,
      "learning_rate": 5.3887534624638335e-06,
      "loss": 0.640765368938446,
      "step": 536
    },
    {
      "epoch": 2.143426294820717,
      "grad_norm": 0.3261059820652008,
      "learning_rate": 5.359096396730924e-06,
      "loss": 0.684579074382782,
      "step": 538
    },
    {
      "epoch": 2.151394422310757,
      "grad_norm": 0.115351602435112,
      "learning_rate": 5.329445455833975e-06,
      "loss": 0.9158648252487183,
      "step": 540
    },
    {
      "epoch": 2.1593625498007967,
      "grad_norm": 0.34780338406562805,
      "learning_rate": 5.299801928648105e-06,
      "loss": 0.8859466910362244,
      "step": 542
    },
    {
      "epoch": 2.1673306772908365,
      "grad_norm": 0.08513224869966507,
      "learning_rate": 5.270167103726181e-06,
      "loss": 0.7413093447685242,
      "step": 544
    },
    {
      "epoch": 2.1752988047808763,
      "grad_norm": 0.12682124972343445,
      "learning_rate": 5.240542269242794e-06,
      "loss": 0.6223682165145874,
      "step": 546
    },
    {
      "epoch": 2.183266932270916,
      "grad_norm": 0.12175296247005463,
      "learning_rate": 5.210928712938262e-06,
      "loss": 0.9578088521957397,
      "step": 548
    },
    {
      "epoch": 2.191235059760956,
      "grad_norm": 0.1025998443365097,
      "learning_rate": 5.181327722062667e-06,
      "loss": 1.019105076789856,
      "step": 550
    },
    {
      "epoch": 2.199203187250996,
      "grad_norm": 1.0863326787948608,
      "learning_rate": 5.151740583319892e-06,
      "loss": 0.6743606925010681,
      "step": 552
    },
    {
      "epoch": 2.2071713147410357,
      "grad_norm": 0.5417909026145935,
      "learning_rate": 5.122168582811689e-06,
      "loss": 0.42565545439720154,
      "step": 554
    },
    {
      "epoch": 2.2151394422310755,
      "grad_norm": 0.266326904296875,
      "learning_rate": 5.092613005981781e-06,
      "loss": 0.6403976082801819,
      "step": 556
    },
    {
      "epoch": 2.2231075697211153,
      "grad_norm": 0.16834671795368195,
      "learning_rate": 5.063075137559979e-06,
      "loss": 0.8421195149421692,
      "step": 558
    },
    {
      "epoch": 2.231075697211155,
      "grad_norm": 0.13817018270492554,
      "learning_rate": 5.033556261506343e-06,
      "loss": 0.8897129893302917,
      "step": 560
    },
    {
      "epoch": 2.239043824701195,
      "grad_norm": 0.14422644674777985,
      "learning_rate": 5.0040576609553645e-06,
      "loss": 1.0258100032806396,
      "step": 562
    },
    {
      "epoch": 2.247011952191235,
      "grad_norm": 0.24181491136550903,
      "learning_rate": 4.974580618160194e-06,
      "loss": 0.7488424181938171,
      "step": 564
    },
    {
      "epoch": 2.2549800796812747,
      "grad_norm": 0.1938558667898178,
      "learning_rate": 4.945126414436908e-06,
      "loss": 0.6393429040908813,
      "step": 566
    },
    {
      "epoch": 2.2629482071713145,
      "grad_norm": 0.08594364672899246,
      "learning_rate": 4.915696330108804e-06,
      "loss": 0.7282727360725403,
      "step": 568
    },
    {
      "epoch": 2.2709163346613543,
      "grad_norm": 0.1412476748228073,
      "learning_rate": 4.886291644450751e-06,
      "loss": 0.996859610080719,
      "step": 570
    },
    {
      "epoch": 2.278884462151394,
      "grad_norm": 0.35477742552757263,
      "learning_rate": 4.8569136356335845e-06,
      "loss": 0.5815088152885437,
      "step": 572
    },
    {
      "epoch": 2.2868525896414345,
      "grad_norm": 0.14603076875209808,
      "learning_rate": 4.827563580668544e-06,
      "loss": 0.6880822777748108,
      "step": 574
    },
    {
      "epoch": 2.2948207171314743,
      "grad_norm": 0.10120602697134018,
      "learning_rate": 4.798242755351755e-06,
      "loss": 0.2764871418476105,
      "step": 576
    },
    {
      "epoch": 2.302788844621514,
      "grad_norm": 0.15307286381721497,
      "learning_rate": 4.768952434208795e-06,
      "loss": 0.6583220362663269,
      "step": 578
    },
    {
      "epoch": 2.310756972111554,
      "grad_norm": 0.14034956693649292,
      "learning_rate": 4.7396938904392656e-06,
      "loss": 0.6165253520011902,
      "step": 580
    },
    {
      "epoch": 2.318725099601594,
      "grad_norm": 0.25004416704177856,
      "learning_rate": 4.71046839586147e-06,
      "loss": 0.5720459818840027,
      "step": 582
    },
    {
      "epoch": 2.3266932270916336,
      "grad_norm": 0.09810896217823029,
      "learning_rate": 4.681277220857108e-06,
      "loss": 0.7465008497238159,
      "step": 584
    },
    {
      "epoch": 2.3346613545816735,
      "grad_norm": 0.08897019922733307,
      "learning_rate": 4.6521216343160824e-06,
      "loss": 0.6230015158653259,
      "step": 586
    },
    {
      "epoch": 2.3426294820717133,
      "grad_norm": 0.38632476329803467,
      "learning_rate": 4.623002903581316e-06,
      "loss": 0.8418217897415161,
      "step": 588
    },
    {
      "epoch": 2.350597609561753,
      "grad_norm": 0.13882328569889069,
      "learning_rate": 4.593922294393677e-06,
      "loss": 0.9831987619400024,
      "step": 590
    },
    {
      "epoch": 2.358565737051793,
      "grad_norm": 0.14402514696121216,
      "learning_rate": 4.564881070836955e-06,
      "loss": 1.0264031887054443,
      "step": 592
    },
    {
      "epoch": 2.366533864541833,
      "grad_norm": 0.12101496756076813,
      "learning_rate": 4.535880495282916e-06,
      "loss": 0.6720498204231262,
      "step": 594
    },
    {
      "epoch": 2.3745019920318726,
      "grad_norm": 0.0773487314581871,
      "learning_rate": 4.506921828336427e-06,
      "loss": 0.9208852052688599,
      "step": 596
    },
    {
      "epoch": 2.3824701195219125,
      "grad_norm": 0.07198522239923477,
      "learning_rate": 4.478006328780657e-06,
      "loss": 0.6842321753501892,
      "step": 598
    },
    {
      "epoch": 2.3904382470119523,
      "grad_norm": 0.804380476474762,
      "learning_rate": 4.44913525352237e-06,
      "loss": 0.8522815704345703,
      "step": 600
    },
    {
      "epoch": 2.398406374501992,
      "grad_norm": 0.08560741692781448,
      "learning_rate": 4.420309857537277e-06,
      "loss": 1.0049006938934326,
      "step": 602
    },
    {
      "epoch": 2.406374501992032,
      "grad_norm": 0.14616362750530243,
      "learning_rate": 4.39153139381549e-06,
      "loss": 0.5706896781921387,
      "step": 604
    },
    {
      "epoch": 2.414342629482072,
      "grad_norm": 0.1779346466064453,
      "learning_rate": 4.3628011133070605e-06,
      "loss": 0.2019987553358078,
      "step": 606
    },
    {
      "epoch": 2.4223107569721116,
      "grad_norm": 0.049101900309324265,
      "learning_rate": 4.334120264867596e-06,
      "loss": 0.4708358347415924,
      "step": 608
    },
    {
      "epoch": 2.4302788844621515,
      "grad_norm": 0.07417161762714386,
      "learning_rate": 4.305490095203979e-06,
      "loss": 0.8848264217376709,
      "step": 610
    },
    {
      "epoch": 2.4382470119521913,
      "grad_norm": 0.08338167518377304,
      "learning_rate": 4.27691184882017e-06,
      "loss": 0.6636566519737244,
      "step": 612
    },
    {
      "epoch": 2.446215139442231,
      "grad_norm": 0.2790423631668091,
      "learning_rate": 4.248386767963121e-06,
      "loss": 0.5570278167724609,
      "step": 614
    },
    {
      "epoch": 2.454183266932271,
      "grad_norm": 0.11870516836643219,
      "learning_rate": 4.219916092568764e-06,
      "loss": 0.8892803192138672,
      "step": 616
    },
    {
      "epoch": 2.462151394422311,
      "grad_norm": 0.20905977487564087,
      "learning_rate": 4.191501060208121e-06,
      "loss": 0.679407000541687,
      "step": 618
    },
    {
      "epoch": 2.4701195219123506,
      "grad_norm": 0.6341357827186584,
      "learning_rate": 4.163142906033512e-06,
      "loss": 0.7536314129829407,
      "step": 620
    },
    {
      "epoch": 2.4780876494023905,
      "grad_norm": 2.335040807723999,
      "learning_rate": 4.134842862724854e-06,
      "loss": 0.8042733669281006,
      "step": 622
    },
    {
      "epoch": 2.4860557768924303,
      "grad_norm": 0.4020189642906189,
      "learning_rate": 4.106602160436088e-06,
      "loss": 0.7613189220428467,
      "step": 624
    },
    {
      "epoch": 2.49402390438247,
      "grad_norm": 0.24192647635936737,
      "learning_rate": 4.078422026741706e-06,
      "loss": 0.5949269533157349,
      "step": 626
    },
    {
      "epoch": 2.50199203187251,
      "grad_norm": 0.26743894815444946,
      "learning_rate": 4.050303686583386e-06,
      "loss": 0.9827489852905273,
      "step": 628
    },
    {
      "epoch": 2.50996015936255,
      "grad_norm": 0.11117087304592133,
      "learning_rate": 4.0222483622167465e-06,
      "loss": 0.440542995929718,
      "step": 630
    },
    {
      "epoch": 2.5179282868525896,
      "grad_norm": 0.29913899302482605,
      "learning_rate": 3.9942572731582185e-06,
      "loss": 0.4106953740119934,
      "step": 632
    },
    {
      "epoch": 2.5258964143426295,
      "grad_norm": 0.10013361275196075,
      "learning_rate": 3.966331636132039e-06,
      "loss": 1.0027564764022827,
      "step": 634
    },
    {
      "epoch": 2.5338645418326693,
      "grad_norm": 0.21742190420627594,
      "learning_rate": 3.938472665017355e-06,
      "loss": 0.9385004639625549,
      "step": 636
    },
    {
      "epoch": 2.541832669322709,
      "grad_norm": 0.25196295976638794,
      "learning_rate": 3.91068157079546e-06,
      "loss": 0.9623706936836243,
      "step": 638
    },
    {
      "epoch": 2.549800796812749,
      "grad_norm": 0.04752785339951515,
      "learning_rate": 3.882959561497159e-06,
      "loss": 0.5855326652526855,
      "step": 640
    },
    {
      "epoch": 2.557768924302789,
      "grad_norm": 0.14722587168216705,
      "learning_rate": 3.855307842150259e-06,
      "loss": 0.9499075412750244,
      "step": 642
    },
    {
      "epoch": 2.5657370517928286,
      "grad_norm": 0.09414464980363846,
      "learning_rate": 3.827727614727177e-06,
      "loss": 1.0164247751235962,
      "step": 644
    },
    {
      "epoch": 2.5737051792828685,
      "grad_norm": 0.1502121239900589,
      "learning_rate": 3.8002200780927e-06,
      "loss": 0.5638771057128906,
      "step": 646
    },
    {
      "epoch": 2.5816733067729083,
      "grad_norm": 0.15606221556663513,
      "learning_rate": 3.772786427951878e-06,
      "loss": 0.9924614429473877,
      "step": 648
    },
    {
      "epoch": 2.589641434262948,
      "grad_norm": 0.10345933586359024,
      "learning_rate": 3.7454278567980406e-06,
      "loss": 0.7498367428779602,
      "step": 650
    },
    {
      "epoch": 2.597609561752988,
      "grad_norm": 0.16408374905586243,
      "learning_rate": 3.7181455538609577e-06,
      "loss": 0.7573265433311462,
      "step": 652
    },
    {
      "epoch": 2.605577689243028,
      "grad_norm": 0.10498823970556259,
      "learning_rate": 3.690940705055163e-06,
      "loss": 0.6695098280906677,
      "step": 654
    },
    {
      "epoch": 2.6135458167330676,
      "grad_norm": 0.14632731676101685,
      "learning_rate": 3.6638144929283824e-06,
      "loss": 0.29860806465148926,
      "step": 656
    },
    {
      "epoch": 2.6215139442231075,
      "grad_norm": 0.10948789119720459,
      "learning_rate": 3.6367680966101483e-06,
      "loss": 0.5482282042503357,
      "step": 658
    },
    {
      "epoch": 2.6294820717131473,
      "grad_norm": 0.07876993715763092,
      "learning_rate": 3.6098026917605334e-06,
      "loss": 0.8400536775588989,
      "step": 660
    },
    {
      "epoch": 2.637450199203187,
      "grad_norm": 0.25182756781578064,
      "learning_rate": 3.5829194505190534e-06,
      "loss": 0.9076658487319946,
      "step": 662
    },
    {
      "epoch": 2.645418326693227,
      "grad_norm": 0.2158064991235733,
      "learning_rate": 3.5561195414537164e-06,
      "loss": 0.5271828174591064,
      "step": 664
    },
    {
      "epoch": 2.653386454183267,
      "grad_norm": 0.18953320384025574,
      "learning_rate": 3.529404129510221e-06,
      "loss": 0.9022460579872131,
      "step": 666
    },
    {
      "epoch": 2.6613545816733066,
      "grad_norm": 0.03914133086800575,
      "learning_rate": 3.502774375961321e-06,
      "loss": 0.5820519328117371,
      "step": 668
    },
    {
      "epoch": 2.6693227091633465,
      "grad_norm": 0.3090229332447052,
      "learning_rate": 3.4762314383563533e-06,
      "loss": 0.7122098207473755,
      "step": 670
    },
    {
      "epoch": 2.6772908366533863,
      "grad_norm": 0.0372186154127121,
      "learning_rate": 3.4497764704709143e-06,
      "loss": 0.32675519585609436,
      "step": 672
    },
    {
      "epoch": 2.685258964143426,
      "grad_norm": 0.07880748063325882,
      "learning_rate": 3.4234106222567055e-06,
      "loss": 0.999765932559967,
      "step": 674
    },
    {
      "epoch": 2.6932270916334664,
      "grad_norm": 0.5355376601219177,
      "learning_rate": 3.3971350397915504e-06,
      "loss": 0.6687252521514893,
      "step": 676
    },
    {
      "epoch": 2.7011952191235062,
      "grad_norm": 0.23132771253585815,
      "learning_rate": 3.370950865229583e-06,
      "loss": 0.5730953812599182,
      "step": 678
    },
    {
      "epoch": 2.709163346613546,
      "grad_norm": 0.12966091930866241,
      "learning_rate": 3.344859236751585e-06,
      "loss": 0.9984683394432068,
      "step": 680
    },
    {
      "epoch": 2.717131474103586,
      "grad_norm": 0.1126488447189331,
      "learning_rate": 3.3188612885155257e-06,
      "loss": 1.0314698219299316,
      "step": 682
    },
    {
      "epoch": 2.7250996015936257,
      "grad_norm": 0.09455672651529312,
      "learning_rate": 3.2929581506072596e-06,
      "loss": 0.6806710958480835,
      "step": 684
    },
    {
      "epoch": 2.7330677290836656,
      "grad_norm": 0.3752633333206177,
      "learning_rate": 3.267150948991387e-06,
      "loss": 0.4101208448410034,
      "step": 686
    },
    {
      "epoch": 2.7410358565737054,
      "grad_norm": 0.10289090871810913,
      "learning_rate": 3.241440805462337e-06,
      "loss": 1.0035781860351562,
      "step": 688
    },
    {
      "epoch": 2.7490039840637452,
      "grad_norm": 0.17575260996818542,
      "learning_rate": 3.215828837595585e-06,
      "loss": 0.4762260913848877,
      "step": 690
    },
    {
      "epoch": 2.756972111553785,
      "grad_norm": 0.09884456545114517,
      "learning_rate": 3.190316158699084e-06,
      "loss": 1.0761598348617554,
      "step": 692
    },
    {
      "epoch": 2.764940239043825,
      "grad_norm": 0.27310195565223694,
      "learning_rate": 3.1649038777648622e-06,
      "loss": 0.8205622434616089,
      "step": 694
    },
    {
      "epoch": 2.7729083665338647,
      "grad_norm": 0.21241426467895508,
      "learning_rate": 3.1395930994208234e-06,
      "loss": 0.6146988868713379,
      "step": 696
    },
    {
      "epoch": 2.7808764940239046,
      "grad_norm": 0.09035513550043106,
      "learning_rate": 3.114384923882734e-06,
      "loss": 0.8542025089263916,
      "step": 698
    },
    {
      "epoch": 2.7888446215139444,
      "grad_norm": 0.07631594687700272,
      "learning_rate": 3.089280446906387e-06,
      "loss": 1.030548334121704,
      "step": 700
    },
    {
      "epoch": 2.7968127490039842,
      "grad_norm": 0.5440763831138611,
      "learning_rate": 3.0642807597399847e-06,
      "loss": 0.637408435344696,
      "step": 702
    },
    {
      "epoch": 2.804780876494024,
      "grad_norm": 0.16706421971321106,
      "learning_rate": 3.0393869490766913e-06,
      "loss": 1.169363021850586,
      "step": 704
    },
    {
      "epoch": 2.812749003984064,
      "grad_norm": 0.10243828594684601,
      "learning_rate": 3.0146000970074087e-06,
      "loss": 1.1671249866485596,
      "step": 706
    },
    {
      "epoch": 2.8207171314741037,
      "grad_norm": 0.1540513038635254,
      "learning_rate": 2.9899212809737275e-06,
      "loss": 1.0633411407470703,
      "step": 708
    },
    {
      "epoch": 2.8286852589641436,
      "grad_norm": 0.7056778073310852,
      "learning_rate": 2.9653515737211037e-06,
      "loss": 0.520878255367279,
      "step": 710
    },
    {
      "epoch": 2.8366533864541834,
      "grad_norm": 0.11693540215492249,
      "learning_rate": 2.940892043252225e-06,
      "loss": 1.0443507432937622,
      "step": 712
    },
    {
      "epoch": 2.8446215139442232,
      "grad_norm": 0.12096396833658218,
      "learning_rate": 2.916543752780574e-06,
      "loss": 0.5934324860572815,
      "step": 714
    },
    {
      "epoch": 2.852589641434263,
      "grad_norm": 0.09405473619699478,
      "learning_rate": 2.892307760684234e-06,
      "loss": 1.0029265880584717,
      "step": 716
    },
    {
      "epoch": 2.860557768924303,
      "grad_norm": 0.511938214302063,
      "learning_rate": 2.8681851204598675e-06,
      "loss": 0.5955791473388672,
      "step": 718
    },
    {
      "epoch": 2.8685258964143427,
      "grad_norm": 0.10217857360839844,
      "learning_rate": 2.844176880676931e-06,
      "loss": 0.5846378207206726,
      "step": 720
    },
    {
      "epoch": 2.8764940239043826,
      "grad_norm": 0.31432026624679565,
      "learning_rate": 2.8202840849320867e-06,
      "loss": 0.6362673044204712,
      "step": 722
    },
    {
      "epoch": 2.8844621513944224,
      "grad_norm": 0.09505125880241394,
      "learning_rate": 2.7965077718038437e-06,
      "loss": 0.6044365763664246,
      "step": 724
    },
    {
      "epoch": 2.8924302788844622,
      "grad_norm": 0.09395304322242737,
      "learning_rate": 2.7728489748074193e-06,
      "loss": 0.9599644541740417,
      "step": 726
    },
    {
      "epoch": 2.900398406374502,
      "grad_norm": 0.36247771978378296,
      "learning_rate": 2.7493087223498015e-06,
      "loss": 0.5444557666778564,
      "step": 728
    },
    {
      "epoch": 2.908366533864542,
      "grad_norm": 0.5664392113685608,
      "learning_rate": 2.725888037685057e-06,
      "loss": 0.9327759742736816,
      "step": 730
    },
    {
      "epoch": 2.9163346613545817,
      "grad_norm": 0.8039626479148865,
      "learning_rate": 2.702587938869842e-06,
      "loss": 0.3262653648853302,
      "step": 732
    },
    {
      "epoch": 2.9243027888446216,
      "grad_norm": 0.08595606684684753,
      "learning_rate": 2.6794094387191617e-06,
      "loss": 1.0276615619659424,
      "step": 734
    },
    {
      "epoch": 2.9322709163346614,
      "grad_norm": 0.35183244943618774,
      "learning_rate": 2.6563535447623302e-06,
      "loss": 0.40339601039886475,
      "step": 736
    },
    {
      "epoch": 2.9402390438247012,
      "grad_norm": 0.23763059079647064,
      "learning_rate": 2.6334212591991847e-06,
      "loss": 0.7936598062515259,
      "step": 738
    },
    {
      "epoch": 2.948207171314741,
      "grad_norm": 0.08670516312122345,
      "learning_rate": 2.6106135788565248e-06,
      "loss": 1.1121079921722412,
      "step": 740
    },
    {
      "epoch": 2.956175298804781,
      "grad_norm": 0.3945688009262085,
      "learning_rate": 2.5879314951447688e-06,
      "loss": 0.3207225799560547,
      "step": 742
    },
    {
      "epoch": 2.9641434262948207,
      "grad_norm": 0.10507278144359589,
      "learning_rate": 2.5653759940148692e-06,
      "loss": 0.8529314398765564,
      "step": 744
    },
    {
      "epoch": 2.9721115537848606,
      "grad_norm": 0.18137764930725098,
      "learning_rate": 2.542948055915454e-06,
      "loss": 0.8667153716087341,
      "step": 746
    },
    {
      "epoch": 2.9800796812749004,
      "grad_norm": 0.1851310282945633,
      "learning_rate": 2.5206486557502087e-06,
      "loss": 0.5118793249130249,
      "step": 748
    },
    {
      "epoch": 2.9880478087649402,
      "grad_norm": 0.11910346150398254,
      "learning_rate": 2.498478762835492e-06,
      "loss": 0.13238655030727386,
      "step": 750
    },
    {
      "epoch": 2.99601593625498,
      "grad_norm": 0.1383173167705536,
      "learning_rate": 2.476439340858208e-06,
      "loss": 0.9977914094924927,
      "step": 752
    },
    {
      "epoch": 3.00398406374502,
      "grad_norm": 0.08733584731817245,
      "learning_rate": 2.4545313478339178e-06,
      "loss": 0.5296513438224792,
      "step": 754
    },
    {
      "epoch": 3.0119521912350598,
      "grad_norm": 0.103951096534729,
      "learning_rate": 2.4327557360651895e-06,
      "loss": 0.5676113367080688,
      "step": 756
    },
    {
      "epoch": 3.0199203187250996,
      "grad_norm": 0.07021525502204895,
      "learning_rate": 2.411113452100211e-06,
      "loss": 0.2701667249202728,
      "step": 758
    },
    {
      "epoch": 3.0278884462151394,
      "grad_norm": 0.37724074721336365,
      "learning_rate": 2.3896054366916365e-06,
      "loss": 0.6014183759689331,
      "step": 760
    },
    {
      "epoch": 3.0358565737051793,
      "grad_norm": 0.2236529439687729,
      "learning_rate": 2.3682326247557e-06,
      "loss": 0.39284777641296387,
      "step": 762
    },
    {
      "epoch": 3.043824701195219,
      "grad_norm": 0.0903429388999939,
      "learning_rate": 2.346995945331576e-06,
      "loss": 0.9162108302116394,
      "step": 764
    },
    {
      "epoch": 3.051792828685259,
      "grad_norm": 0.19768160581588745,
      "learning_rate": 2.325896321540993e-06,
      "loss": 0.1558755487203598,
      "step": 766
    },
    {
      "epoch": 3.0597609561752988,
      "grad_norm": 0.10917969048023224,
      "learning_rate": 2.3049346705481086e-06,
      "loss": 0.5417059063911438,
      "step": 768
    },
    {
      "epoch": 3.0677290836653386,
      "grad_norm": 0.18912799656391144,
      "learning_rate": 2.284111903519641e-06,
      "loss": 0.7214385867118835,
      "step": 770
    },
    {
      "epoch": 3.0756972111553784,
      "grad_norm": 0.20590689778327942,
      "learning_rate": 2.26342892558526e-06,
      "loss": 0.9078834652900696,
      "step": 772
    },
    {
      "epoch": 3.0836653386454183,
      "grad_norm": 0.8302822709083557,
      "learning_rate": 2.2428866357982515e-06,
      "loss": 0.7995004057884216,
      "step": 774
    },
    {
      "epoch": 3.091633466135458,
      "grad_norm": 0.020072488114237785,
      "learning_rate": 2.222485927096424e-06,
      "loss": 0.5199475288391113,
      "step": 776
    },
    {
      "epoch": 3.099601593625498,
      "grad_norm": 0.16181273758411407,
      "learning_rate": 2.2022276862633075e-06,
      "loss": 0.5194544196128845,
      "step": 778
    },
    {
      "epoch": 3.1075697211155378,
      "grad_norm": 0.09146042913198471,
      "learning_rate": 2.1821127938895938e-06,
      "loss": 0.8923799395561218,
      "step": 780
    },
    {
      "epoch": 3.1155378486055776,
      "grad_norm": 0.05091450363397598,
      "learning_rate": 2.162142124334871e-06,
      "loss": 0.2357436567544937,
      "step": 782
    },
    {
      "epoch": 3.1235059760956174,
      "grad_norm": 0.1348317712545395,
      "learning_rate": 2.1423165456896068e-06,
      "loss": 0.9547862410545349,
      "step": 784
    },
    {
      "epoch": 3.1314741035856573,
      "grad_norm": 0.10621712356805801,
      "learning_rate": 2.1226369197374176e-06,
      "loss": 0.6909422278404236,
      "step": 786
    },
    {
      "epoch": 3.139442231075697,
      "grad_norm": 0.2938082814216614,
      "learning_rate": 2.103104101917615e-06,
      "loss": 0.5021622180938721,
      "step": 788
    },
    {
      "epoch": 3.147410358565737,
      "grad_norm": 0.07215439528226852,
      "learning_rate": 2.0837189412880044e-06,
      "loss": 0.48427554965019226,
      "step": 790
    },
    {
      "epoch": 3.1553784860557768,
      "grad_norm": 0.37235894799232483,
      "learning_rate": 2.0644822804879964e-06,
      "loss": 0.6456067562103271,
      "step": 792
    },
    {
      "epoch": 3.1633466135458166,
      "grad_norm": 0.023045551031827927,
      "learning_rate": 2.045394955701969e-06,
      "loss": 0.16015684604644775,
      "step": 794
    },
    {
      "epoch": 3.1713147410358564,
      "grad_norm": 0.291789710521698,
      "learning_rate": 2.0264577966229222e-06,
      "loss": 0.5443094372749329,
      "step": 796
    },
    {
      "epoch": 3.1792828685258963,
      "grad_norm": 0.16915495693683624,
      "learning_rate": 2.007671626416412e-06,
      "loss": 0.505410373210907,
      "step": 798
    },
    {
      "epoch": 3.187250996015936,
      "grad_norm": 0.23237206041812897,
      "learning_rate": 1.9890372616847663e-06,
      "loss": 0.7769385576248169,
      "step": 800
    },
    {
      "epoch": 3.195219123505976,
      "grad_norm": 0.09916416555643082,
      "learning_rate": 1.9705555124315965e-06,
      "loss": 0.655504584312439,
      "step": 802
    },
    {
      "epoch": 3.2031872509960158,
      "grad_norm": 0.3141724169254303,
      "learning_rate": 1.95222718202658e-06,
      "loss": 0.5474756360054016,
      "step": 804
    },
    {
      "epoch": 3.2111553784860556,
      "grad_norm": 0.2176314890384674,
      "learning_rate": 1.934053067170545e-06,
      "loss": 0.9463359713554382,
      "step": 806
    },
    {
      "epoch": 3.2191235059760954,
      "grad_norm": 0.3539562523365021,
      "learning_rate": 1.9160339578608305e-06,
      "loss": 0.2166862040758133,
      "step": 808
    },
    {
      "epoch": 3.2270916334661353,
      "grad_norm": 0.11151967942714691,
      "learning_rate": 1.8981706373569602e-06,
      "loss": 0.6182011961936951,
      "step": 810
    },
    {
      "epoch": 3.235059760956175,
      "grad_norm": 0.18874166905879974,
      "learning_rate": 1.8804638821465821e-06,
      "loss": 0.31370559334754944,
      "step": 812
    },
    {
      "epoch": 3.243027888446215,
      "grad_norm": 0.08038028329610825,
      "learning_rate": 1.8629144619117234e-06,
      "loss": 0.7675144076347351,
      "step": 814
    },
    {
      "epoch": 3.2509960159362548,
      "grad_norm": 0.3184245824813843,
      "learning_rate": 1.8455231394953348e-06,
      "loss": 0.22720375657081604,
      "step": 816
    },
    {
      "epoch": 3.2589641434262946,
      "grad_norm": 0.10131336003541946,
      "learning_rate": 1.8282906708681238e-06,
      "loss": 0.9015382528305054,
      "step": 818
    },
    {
      "epoch": 3.2669322709163344,
      "grad_norm": 0.39507946372032166,
      "learning_rate": 1.811217805095699e-06,
      "loss": 0.43418973684310913,
      "step": 820
    },
    {
      "epoch": 3.2749003984063743,
      "grad_norm": 0.29091838002204895,
      "learning_rate": 1.7943052843060133e-06,
      "loss": 0.6147860884666443,
      "step": 822
    },
    {
      "epoch": 3.2828685258964145,
      "grad_norm": 0.33153676986694336,
      "learning_rate": 1.7775538436570937e-06,
      "loss": 0.5188818573951721,
      "step": 824
    },
    {
      "epoch": 3.2908366533864544,
      "grad_norm": 0.09359156340360641,
      "learning_rate": 1.7609642113050996e-06,
      "loss": 0.8976441025733948,
      "step": 826
    },
    {
      "epoch": 3.298804780876494,
      "grad_norm": 0.16510267555713654,
      "learning_rate": 1.7445371083726567e-06,
      "loss": 0.8939739465713501,
      "step": 828
    },
    {
      "epoch": 3.306772908366534,
      "grad_norm": 0.1346827745437622,
      "learning_rate": 1.7282732489175224e-06,
      "loss": 0.8164635896682739,
      "step": 830
    },
    {
      "epoch": 3.314741035856574,
      "grad_norm": 0.2717640995979309,
      "learning_rate": 1.7121733399015406e-06,
      "loss": 0.37919336557388306,
      "step": 832
    },
    {
      "epoch": 3.3227091633466137,
      "grad_norm": 0.637654185295105,
      "learning_rate": 1.6962380811599138e-06,
      "loss": 0.24662509560585022,
      "step": 834
    },
    {
      "epoch": 3.3306772908366535,
      "grad_norm": 0.39412349462509155,
      "learning_rate": 1.6804681653707816e-06,
      "loss": 0.46995463967323303,
      "step": 836
    },
    {
      "epoch": 3.3386454183266934,
      "grad_norm": 0.1424451619386673,
      "learning_rate": 1.6648642780251096e-06,
      "loss": 0.7019530534744263,
      "step": 838
    },
    {
      "epoch": 3.346613545816733,
      "grad_norm": 0.29868650436401367,
      "learning_rate": 1.6494270973968974e-06,
      "loss": 0.1480187326669693,
      "step": 840
    },
    {
      "epoch": 3.354581673306773,
      "grad_norm": 0.30046868324279785,
      "learning_rate": 1.6341572945136902e-06,
      "loss": 0.580310583114624,
      "step": 842
    },
    {
      "epoch": 3.362549800796813,
      "grad_norm": 0.08614935725927353,
      "learning_rate": 1.6190555331274122e-06,
      "loss": 0.5572804808616638,
      "step": 844
    },
    {
      "epoch": 3.3705179282868527,
      "grad_norm": 0.13543984293937683,
      "learning_rate": 1.6041224696855145e-06,
      "loss": 0.9477728009223938,
      "step": 846
    },
    {
      "epoch": 3.3784860557768925,
      "grad_norm": 0.11601915210485458,
      "learning_rate": 1.589358753302439e-06,
      "loss": 0.8788732886314392,
      "step": 848
    },
    {
      "epoch": 3.3864541832669324,
      "grad_norm": 0.2447684109210968,
      "learning_rate": 1.5747650257314068e-06,
      "loss": 0.8367201685905457,
      "step": 850
    },
    {
      "epoch": 3.394422310756972,
      "grad_norm": 0.3203533887863159,
      "learning_rate": 1.5603419213365146e-06,
      "loss": 0.8870067596435547,
      "step": 852
    },
    {
      "epoch": 3.402390438247012,
      "grad_norm": 0.2368336319923401,
      "learning_rate": 1.5460900670651705e-06,
      "loss": 0.5136474967002869,
      "step": 854
    },
    {
      "epoch": 3.410358565737052,
      "grad_norm": 0.15598580241203308,
      "learning_rate": 1.5320100824208306e-06,
      "loss": 0.8361679315567017,
      "step": 856
    },
    {
      "epoch": 3.4183266932270917,
      "grad_norm": 0.12319839000701904,
      "learning_rate": 1.51810257943608e-06,
      "loss": 0.9815875887870789,
      "step": 858
    },
    {
      "epoch": 3.4262948207171315,
      "grad_norm": 0.16267569363117218,
      "learning_rate": 1.5043681626460196e-06,
      "loss": 0.9458333849906921,
      "step": 860
    },
    {
      "epoch": 3.4342629482071714,
      "grad_norm": 0.08212095499038696,
      "learning_rate": 1.4908074290619972e-06,
      "loss": 0.6067045331001282,
      "step": 862
    },
    {
      "epoch": 3.442231075697211,
      "grad_norm": 0.2448061853647232,
      "learning_rate": 1.477420968145652e-06,
      "loss": 0.44623082876205444,
      "step": 864
    },
    {
      "epoch": 3.450199203187251,
      "grad_norm": 0.1172880157828331,
      "learning_rate": 1.4642093617832853e-06,
      "loss": 0.7335717678070068,
      "step": 866
    },
    {
      "epoch": 3.458167330677291,
      "grad_norm": 0.11858659982681274,
      "learning_rate": 1.4511731842605805e-06,
      "loss": 0.64848393201828,
      "step": 868
    },
    {
      "epoch": 3.4661354581673307,
      "grad_norm": 0.32108765840530396,
      "learning_rate": 1.4383130022376298e-06,
      "loss": 0.48904505372047424,
      "step": 870
    },
    {
      "epoch": 3.4741035856573705,
      "grad_norm": 0.03352862969040871,
      "learning_rate": 1.4256293747243065e-06,
      "loss": 0.5316774845123291,
      "step": 872
    },
    {
      "epoch": 3.4820717131474104,
      "grad_norm": 0.20545020699501038,
      "learning_rate": 1.4131228530559637e-06,
      "loss": 0.6059108376502991,
      "step": 874
    },
    {
      "epoch": 3.49003984063745,
      "grad_norm": 0.11125019192695618,
      "learning_rate": 1.400793980869468e-06,
      "loss": 0.48970475792884827,
      "step": 876
    },
    {
      "epoch": 3.49800796812749,
      "grad_norm": 0.15669061243534088,
      "learning_rate": 1.3886432940795744e-06,
      "loss": 0.49637895822525024,
      "step": 878
    },
    {
      "epoch": 3.50597609561753,
      "grad_norm": 0.04867348074913025,
      "learning_rate": 1.3766713208556232e-06,
      "loss": 0.32917848229408264,
      "step": 880
    },
    {
      "epoch": 3.5139442231075697,
      "grad_norm": 0.19297191500663757,
      "learning_rate": 1.3648785815985876e-06,
      "loss": 0.5148631930351257,
      "step": 882
    },
    {
      "epoch": 3.5219123505976095,
      "grad_norm": 0.5676918029785156,
      "learning_rate": 1.353265588918448e-06,
      "loss": 0.10626660287380219,
      "step": 884
    },
    {
      "epoch": 3.5298804780876494,
      "grad_norm": 0.1393618881702423,
      "learning_rate": 1.3418328476119136e-06,
      "loss": 0.7224240303039551,
      "step": 886
    },
    {
      "epoch": 3.537848605577689,
      "grad_norm": 0.16956079006195068,
      "learning_rate": 1.330580854640477e-06,
      "loss": 0.861225962638855,
      "step": 888
    },
    {
      "epoch": 3.545816733067729,
      "grad_norm": 0.15440139174461365,
      "learning_rate": 1.3195100991088142e-06,
      "loss": 0.47051650285720825,
      "step": 890
    },
    {
      "epoch": 3.553784860557769,
      "grad_norm": 0.3990451395511627,
      "learning_rate": 1.308621062243524e-06,
      "loss": 0.8131183385848999,
      "step": 892
    },
    {
      "epoch": 3.5617529880478087,
      "grad_norm": 0.893460214138031,
      "learning_rate": 1.2979142173722054e-06,
      "loss": 0.9933807253837585,
      "step": 894
    },
    {
      "epoch": 3.5697211155378485,
      "grad_norm": 0.0942659005522728,
      "learning_rate": 1.2873900299028896e-06,
      "loss": 0.18559087812900543,
      "step": 896
    },
    {
      "epoch": 3.5776892430278884,
      "grad_norm": 0.11877981573343277,
      "learning_rate": 1.2770489573038073e-06,
      "loss": 0.5669529438018799,
      "step": 898
    },
    {
      "epoch": 3.585657370517928,
      "grad_norm": 0.27595198154449463,
      "learning_rate": 1.2668914490834983e-06,
      "loss": 0.1947745680809021,
      "step": 900
    },
    {
      "epoch": 3.593625498007968,
      "grad_norm": 0.12320013344287872,
      "learning_rate": 1.2569179467712805e-06,
      "loss": 0.08118575066328049,
      "step": 902
    },
    {
      "epoch": 3.601593625498008,
      "grad_norm": 0.15320217609405518,
      "learning_rate": 1.2471288838980505e-06,
      "loss": 0.7362831234931946,
      "step": 904
    },
    {
      "epoch": 3.6095617529880477,
      "grad_norm": 0.15283599495887756,
      "learning_rate": 1.237524685977444e-06,
      "loss": 0.5521562695503235,
      "step": 906
    },
    {
      "epoch": 3.6175298804780875,
      "grad_norm": 0.12375420331954956,
      "learning_rate": 1.2281057704873331e-06,
      "loss": 0.8319555521011353,
      "step": 908
    },
    {
      "epoch": 3.6254980079681274,
      "grad_norm": 0.2633494734764099,
      "learning_rate": 1.2188725468516876e-06,
      "loss": 0.7656584978103638,
      "step": 910
    },
    {
      "epoch": 3.633466135458167,
      "grad_norm": 0.048641543835401535,
      "learning_rate": 1.2098254164227706e-06,
      "loss": 0.04072123020887375,
      "step": 912
    },
    {
      "epoch": 3.641434262948207,
      "grad_norm": 0.14492478966712952,
      "learning_rate": 1.200964772463695e-06,
      "loss": 0.7114333510398865,
      "step": 914
    },
    {
      "epoch": 3.649402390438247,
      "grad_norm": 0.18942925333976746,
      "learning_rate": 1.1922910001313312e-06,
      "loss": 0.7078197002410889,
      "step": 916
    },
    {
      "epoch": 3.6573705179282867,
      "grad_norm": 0.3490183651447296,
      "learning_rate": 1.1838044764595617e-06,
      "loss": 0.1272016018629074,
      "step": 918
    },
    {
      "epoch": 3.6653386454183265,
      "grad_norm": 0.13709823787212372,
      "learning_rate": 1.1755055703428958e-06,
      "loss": 0.8840264678001404,
      "step": 920
    },
    {
      "epoch": 3.6733067729083664,
      "grad_norm": 0.29470717906951904,
      "learning_rate": 1.1673946425204288e-06,
      "loss": 0.12348813563585281,
      "step": 922
    },
    {
      "epoch": 3.681274900398406,
      "grad_norm": 0.25545525550842285,
      "learning_rate": 1.1594720455601664e-06,
      "loss": 0.8577460050582886,
      "step": 924
    },
    {
      "epoch": 3.6892430278884465,
      "grad_norm": 0.10373926907777786,
      "learning_rate": 1.1517381238436986e-06,
      "loss": 0.22830277681350708,
      "step": 926
    },
    {
      "epoch": 3.6972111553784863,
      "grad_norm": 0.08529862016439438,
      "learning_rate": 1.1441932135512262e-06,
      "loss": 0.5053238868713379,
      "step": 928
    },
    {
      "epoch": 3.705179282868526,
      "grad_norm": 0.09483988583087921,
      "learning_rate": 1.1368376426469526e-06,
      "loss": 0.7587119936943054,
      "step": 930
    },
    {
      "epoch": 3.713147410358566,
      "grad_norm": 0.13993456959724426,
      "learning_rate": 1.1296717308648237e-06,
      "loss": 0.8940988779067993,
      "step": 932
    },
    {
      "epoch": 3.721115537848606,
      "grad_norm": 0.17292512953281403,
      "learning_rate": 1.1226957896946332e-06,
      "loss": 0.4452460706233978,
      "step": 934
    },
    {
      "epoch": 3.7290836653386457,
      "grad_norm": 0.20580518245697021,
      "learning_rate": 1.115910122368479e-06,
      "loss": 0.8040222525596619,
      "step": 936
    },
    {
      "epoch": 3.7370517928286855,
      "grad_norm": 0.266689270734787,
      "learning_rate": 1.1093150238475847e-06,
      "loss": 0.46449920535087585,
      "step": 938
    },
    {
      "epoch": 3.7450199203187253,
      "grad_norm": 0.11229197680950165,
      "learning_rate": 1.1029107808094789e-06,
      "loss": 0.7149961590766907,
      "step": 940
    },
    {
      "epoch": 3.752988047808765,
      "grad_norm": 0.5672265291213989,
      "learning_rate": 1.0966976716355294e-06,
      "loss": 0.5650360584259033,
      "step": 942
    },
    {
      "epoch": 3.760956175298805,
      "grad_norm": 0.3948570191860199,
      "learning_rate": 1.0906759663988475e-06,
      "loss": 0.4824882447719574,
      "step": 944
    },
    {
      "epoch": 3.768924302788845,
      "grad_norm": 0.09813003987073898,
      "learning_rate": 1.084845926852547e-06,
      "loss": 0.9347279667854309,
      "step": 946
    },
    {
      "epoch": 3.7768924302788847,
      "grad_norm": 0.1827877312898636,
      "learning_rate": 1.0792078064183652e-06,
      "loss": 0.5004470348358154,
      "step": 948
    },
    {
      "epoch": 3.7848605577689245,
      "grad_norm": 0.08153803646564484,
      "learning_rate": 1.0737618501756455e-06,
      "loss": 0.898761510848999,
      "step": 950
    },
    {
      "epoch": 3.7928286852589643,
      "grad_norm": 0.13369211554527283,
      "learning_rate": 1.0685082948506875e-06,
      "loss": 0.9007242321968079,
      "step": 952
    },
    {
      "epoch": 3.800796812749004,
      "grad_norm": 0.11834340542554855,
      "learning_rate": 1.063447368806457e-06,
      "loss": 0.5072510838508606,
      "step": 954
    },
    {
      "epoch": 3.808764940239044,
      "grad_norm": 0.6668749451637268,
      "learning_rate": 1.0585792920326565e-06,
      "loss": 0.6785795092582703,
      "step": 956
    },
    {
      "epoch": 3.816733067729084,
      "grad_norm": 1.107836127281189,
      "learning_rate": 1.0539042761361646e-06,
      "loss": 0.477740615606308,
      "step": 958
    },
    {
      "epoch": 3.8247011952191237,
      "grad_norm": 0.19571809470653534,
      "learning_rate": 1.0494225243318373e-06,
      "loss": 0.597763180732727,
      "step": 960
    },
    {
      "epoch": 3.8326693227091635,
      "grad_norm": 0.04349878430366516,
      "learning_rate": 1.0451342314336755e-06,
      "loss": 0.513166069984436,
      "step": 962
    },
    {
      "epoch": 3.8406374501992033,
      "grad_norm": 0.20118169486522675,
      "learning_rate": 1.0410395838463547e-06,
      "loss": 0.9219191074371338,
      "step": 964
    },
    {
      "epoch": 3.848605577689243,
      "grad_norm": 0.2624046206474304,
      "learning_rate": 1.0371387595571248e-06,
      "loss": 0.30333149433135986,
      "step": 966
    },
    {
      "epoch": 3.856573705179283,
      "grad_norm": 0.09783243387937546,
      "learning_rate": 1.0334319281280715e-06,
      "loss": 0.5276415348052979,
      "step": 968
    },
    {
      "epoch": 3.864541832669323,
      "grad_norm": 0.2174321860074997,
      "learning_rate": 1.0299192506887467e-06,
      "loss": 0.6463159918785095,
      "step": 970
    },
    {
      "epoch": 3.8725099601593627,
      "grad_norm": 0.02547277696430683,
      "learning_rate": 1.0266008799291635e-06,
      "loss": 0.4986192584037781,
      "step": 972
    },
    {
      "epoch": 3.8804780876494025,
      "grad_norm": 0.041136253625154495,
      "learning_rate": 1.0234769600931612e-06,
      "loss": 0.39934349060058594,
      "step": 974
    },
    {
      "epoch": 3.8884462151394423,
      "grad_norm": 0.09584596008062363,
      "learning_rate": 1.0205476269721312e-06,
      "loss": 0.9003692269325256,
      "step": 976
    },
    {
      "epoch": 3.896414342629482,
      "grad_norm": 0.1499820202589035,
      "learning_rate": 1.0178130078991196e-06,
      "loss": 0.5276603102684021,
      "step": 978
    },
    {
      "epoch": 3.904382470119522,
      "grad_norm": 0.062052126973867416,
      "learning_rate": 1.0152732217432875e-06,
      "loss": 0.35644063353538513,
      "step": 980
    },
    {
      "epoch": 3.912350597609562,
      "grad_norm": 0.21071912348270416,
      "learning_rate": 1.0129283789047473e-06,
      "loss": 0.40043482184410095,
      "step": 982
    },
    {
      "epoch": 3.9203187250996017,
      "grad_norm": 0.0894634947180748,
      "learning_rate": 1.0107785813097621e-06,
      "loss": 0.8511362671852112,
      "step": 984
    },
    {
      "epoch": 3.9282868525896415,
      "grad_norm": 0.23283158242702484,
      "learning_rate": 1.0088239224063153e-06,
      "loss": 0.1116846427321434,
      "step": 986
    },
    {
      "epoch": 3.9362549800796813,
      "grad_norm": 0.4885595738887787,
      "learning_rate": 1.0070644871600495e-06,
      "loss": 0.5695192813873291,
      "step": 988
    },
    {
      "epoch": 3.944223107569721,
      "grad_norm": 0.09019921720027924,
      "learning_rate": 1.0055003520505714e-06,
      "loss": 0.5186564326286316,
      "step": 990
    },
    {
      "epoch": 3.952191235059761,
      "grad_norm": 0.11849397420883179,
      "learning_rate": 1.00413158506813e-06,
      "loss": 0.9052659869194031,
      "step": 992
    },
    {
      "epoch": 3.960159362549801,
      "grad_norm": 0.07879645377397537,
      "learning_rate": 1.0029582457106588e-06,
      "loss": 0.20878446102142334,
      "step": 994
    },
    {
      "epoch": 3.9681274900398407,
      "grad_norm": 0.2216099500656128,
      "learning_rate": 1.00198038498119e-06,
      "loss": 0.7964245676994324,
      "step": 996
    },
    {
      "epoch": 3.9760956175298805,
      "grad_norm": 0.04027597978711128,
      "learning_rate": 1.0011980453856391e-06,
      "loss": 0.5180699229240417,
      "step": 998
    },
    {
      "epoch": 3.9840637450199203,
      "grad_norm": 0.3593519330024719,
      "learning_rate": 1.0006112609309548e-06,
      "loss": 0.37535738945007324,
      "step": 1000
    },
    {
      "epoch": 3.99203187250996,
      "grad_norm": 0.12222480773925781,
      "learning_rate": 1.0002200571236429e-06,
      "loss": 0.9074435234069824,
      "step": 1002
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.051510538905858994,
      "learning_rate": 1.0000244509686564e-06,
      "loss": 0.2501225471496582,
      "step": 1004
    },
    {
      "epoch": 4.0,
      "step": 1004,
      "total_flos": 3.954132003103703e+18,
      "train_loss": 0.8403717435273041,
      "train_runtime": 34650.8856,
      "train_samples_per_second": 1.738,
      "train_steps_per_second": 0.029
    }
  ],
  "logging_steps": 2,
  "max_steps": 1004,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 99999,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.954132003103703e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}