{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.015341202464006533,
  "eval_steps": 500,
  "global_step": 1597,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.606263283660946e-06,
      "grad_norm": 0.14495976269245148,
      "learning_rate": 2e-05,
      "loss": 1.6676,
      "step": 1
    },
    {
      "epoch": 1.9212526567321893e-05,
      "grad_norm": 0.21062932908535004,
      "learning_rate": 4e-05,
      "loss": 2.1992,
      "step": 2
    },
    {
      "epoch": 2.881878985098284e-05,
      "grad_norm": 0.25762754678726196,
      "learning_rate": 6e-05,
      "loss": 2.391,
      "step": 3
    },
    {
      "epoch": 3.8425053134643785e-05,
      "grad_norm": 0.8984953761100769,
      "learning_rate": 8e-05,
      "loss": 1.6662,
      "step": 4
    },
    {
      "epoch": 4.803131641830473e-05,
      "grad_norm": 1.4548120498657227,
      "learning_rate": 0.0001,
      "loss": 1.8096,
      "step": 5
    },
    {
      "epoch": 5.763757970196568e-05,
      "grad_norm": 0.6851717829704285,
      "learning_rate": 9.999990264607036e-05,
      "loss": 1.8183,
      "step": 6
    },
    {
      "epoch": 6.724384298562662e-05,
      "grad_norm": 0.8171159625053406,
      "learning_rate": 9.999961058466053e-05,
      "loss": 1.9373,
      "step": 7
    },
    {
      "epoch": 7.685010626928757e-05,
      "grad_norm": 0.5197387337684631,
      "learning_rate": 9.999912381690782e-05,
      "loss": 1.5923,
      "step": 8
    },
    {
      "epoch": 8.645636955294852e-05,
      "grad_norm": 0.7395459413528442,
      "learning_rate": 9.999844234470782e-05,
      "loss": 2.0587,
      "step": 9
    },
    {
      "epoch": 9.606263283660947e-05,
      "grad_norm": 1.6935160160064697,
      "learning_rate": 9.999756617071427e-05,
      "loss": 1.6117,
      "step": 10
    },
    {
      "epoch": 0.00010566889612027041,
      "grad_norm": 2.026272773742676,
      "learning_rate": 9.999649529833915e-05,
      "loss": 1.4417,
      "step": 11
    },
    {
      "epoch": 0.00011527515940393136,
      "grad_norm": 1.0077913999557495,
      "learning_rate": 9.999522973175258e-05,
      "loss": 1.9104,
      "step": 12
    },
    {
      "epoch": 0.0001248814226875923,
      "grad_norm": 1.5325918197631836,
      "learning_rate": 9.999376947588288e-05,
      "loss": 1.2988,
      "step": 13
    },
    {
      "epoch": 0.00013448768597125324,
      "grad_norm": 1.6350547075271606,
      "learning_rate": 9.99921145364165e-05,
      "loss": 1.4681,
      "step": 14
    },
    {
      "epoch": 0.0001440939492549142,
      "grad_norm": 1.4793027639389038,
      "learning_rate": 9.999026491979808e-05,
      "loss": 1.4799,
      "step": 15
    },
    {
      "epoch": 0.00015370021253857514,
      "grad_norm": 1.3111652135849,
      "learning_rate": 9.998822063323029e-05,
      "loss": 1.1776,
      "step": 16
    },
    {
      "epoch": 0.0001633064758222361,
      "grad_norm": 1.0607692003250122,
      "learning_rate": 9.99859816846739e-05,
      "loss": 1.2731,
      "step": 17
    },
    {
      "epoch": 0.00017291273910589704,
      "grad_norm": 1.9974901676177979,
      "learning_rate": 9.998354808284774e-05,
      "loss": 1.1644,
      "step": 18
    },
    {
      "epoch": 0.000182519002389558,
      "grad_norm": 1.0126925706863403,
      "learning_rate": 9.998091983722863e-05,
      "loss": 1.2306,
      "step": 19
    },
    {
      "epoch": 0.00019212526567321893,
      "grad_norm": 1.3808321952819824,
      "learning_rate": 9.997809695805135e-05,
      "loss": 1.5281,
      "step": 20
    },
    {
      "epoch": 0.0002017315289568799,
      "grad_norm": 1.4283260107040405,
      "learning_rate": 9.99750794563087e-05,
      "loss": 1.6118,
      "step": 21
    },
    {
      "epoch": 0.00021133779224054083,
      "grad_norm": 1.3764599561691284,
      "learning_rate": 9.997186734375123e-05,
      "loss": 1.4118,
      "step": 22
    },
    {
      "epoch": 0.0002209440555242018,
      "grad_norm": 1.6656081676483154,
      "learning_rate": 9.996846063288747e-05,
      "loss": 1.2301,
      "step": 23
    },
    {
      "epoch": 0.00023055031880786273,
      "grad_norm": 1.6236650943756104,
      "learning_rate": 9.996485933698365e-05,
      "loss": 1.377,
      "step": 24
    },
    {
      "epoch": 0.0002401565820915237,
      "grad_norm": 1.5138514041900635,
      "learning_rate": 9.996106347006379e-05,
      "loss": 1.2694,
      "step": 25
    },
    {
      "epoch": 0.0002497628453751846,
      "grad_norm": 1.3256175518035889,
      "learning_rate": 9.99570730469096e-05,
      "loss": 1.1344,
      "step": 26
    },
    {
      "epoch": 0.00025936910865884556,
      "grad_norm": 1.8038383722305298,
      "learning_rate": 9.99528880830604e-05,
      "loss": 0.8756,
      "step": 27
    },
    {
      "epoch": 0.0002689753719425065,
      "grad_norm": 1.2924219369888306,
      "learning_rate": 9.994850859481313e-05,
      "loss": 1.3219,
      "step": 28
    },
    {
      "epoch": 0.0002785816352261675,
      "grad_norm": 1.938310980796814,
      "learning_rate": 9.994393459922218e-05,
      "loss": 0.826,
      "step": 29
    },
    {
      "epoch": 0.0002881878985098284,
      "grad_norm": 1.5296651124954224,
      "learning_rate": 9.993916611409941e-05,
      "loss": 1.4519,
      "step": 30
    },
    {
      "epoch": 0.00029779416179348935,
      "grad_norm": 1.5700263977050781,
      "learning_rate": 9.993420315801406e-05,
      "loss": 1.7252,
      "step": 31
    },
    {
      "epoch": 0.0003074004250771503,
      "grad_norm": 2.414926290512085,
      "learning_rate": 9.992904575029264e-05,
      "loss": 1.4879,
      "step": 32
    },
    {
      "epoch": 0.00031700668836081127,
      "grad_norm": 1.8164900541305542,
      "learning_rate": 9.992369391101895e-05,
      "loss": 0.9039,
      "step": 33
    },
    {
      "epoch": 0.0003266129516444722,
      "grad_norm": 3.1730737686157227,
      "learning_rate": 9.991814766103385e-05,
      "loss": 0.4925,
      "step": 34
    },
    {
      "epoch": 0.00033621921492813314,
      "grad_norm": 2.13222336769104,
      "learning_rate": 9.991240702193532e-05,
      "loss": 0.8911,
      "step": 35
    },
    {
      "epoch": 0.0003458254782117941,
      "grad_norm": 1.5261226892471313,
      "learning_rate": 9.99064720160783e-05,
      "loss": 0.1545,
      "step": 36
    },
    {
      "epoch": 0.00035543174149545506,
      "grad_norm": 1.5573840141296387,
      "learning_rate": 9.990034266657467e-05,
      "loss": 1.7874,
      "step": 37
    },
    {
      "epoch": 0.000365038004779116,
      "grad_norm": 1.5914417505264282,
      "learning_rate": 9.989401899729305e-05,
      "loss": 1.2142,
      "step": 38
    },
    {
      "epoch": 0.00037464426806277693,
      "grad_norm": 2.0287771224975586,
      "learning_rate": 9.988750103285883e-05,
      "loss": 1.4565,
      "step": 39
    },
    {
      "epoch": 0.00038425053134643787,
      "grad_norm": 1.8799737691879272,
      "learning_rate": 9.988078879865396e-05,
      "loss": 0.7811,
      "step": 40
    },
    {
      "epoch": 0.0003938567946300988,
      "grad_norm": 3.282158613204956,
      "learning_rate": 9.987388232081694e-05,
      "loss": 2.0671,
      "step": 41
    },
    {
      "epoch": 0.0004034630579137598,
      "grad_norm": 2.947150945663452,
      "learning_rate": 9.98667816262427e-05,
      "loss": 1.7885,
      "step": 42
    },
    {
      "epoch": 0.0004130693211974207,
      "grad_norm": 2.634065628051758,
      "learning_rate": 9.985948674258243e-05,
      "loss": 1.2769,
      "step": 43
    },
    {
      "epoch": 0.00042267558448108166,
      "grad_norm": 2.4246838092803955,
      "learning_rate": 9.985199769824358e-05,
      "loss": 1.9088,
      "step": 44
    },
    {
      "epoch": 0.0004322818477647426,
      "grad_norm": 2.908745765686035,
      "learning_rate": 9.984431452238967e-05,
      "loss": 1.8612,
      "step": 45
    },
    {
      "epoch": 0.0004418881110484036,
      "grad_norm": 5.218689918518066,
      "learning_rate": 9.983643724494016e-05,
      "loss": 2.2283,
      "step": 46
    },
    {
      "epoch": 0.0004514943743320645,
      "grad_norm": 2.420764446258545,
      "learning_rate": 9.982836589657043e-05,
      "loss": 2.9241,
      "step": 47
    },
    {
      "epoch": 0.00046110063761572545,
      "grad_norm": 4.500425338745117,
      "learning_rate": 9.982010050871159e-05,
      "loss": 1.436,
      "step": 48
    },
    {
      "epoch": 0.0004707069008993864,
      "grad_norm": 4.564813137054443,
      "learning_rate": 9.981164111355035e-05,
      "loss": 2.0793,
      "step": 49
    },
    {
      "epoch": 0.0004803131641830474,
      "grad_norm": 6.348535537719727,
      "learning_rate": 9.98029877440289e-05,
      "loss": 3.823,
      "step": 50
    },
    {
      "epoch": 0.0004899194274667083,
      "grad_norm": 0.5213695168495178,
      "learning_rate": 9.979414043384485e-05,
      "loss": 1.7958,
      "step": 51
    },
    {
      "epoch": 0.0004995256907503692,
      "grad_norm": 0.6419565081596375,
      "learning_rate": 9.978509921745101e-05,
      "loss": 1.7161,
      "step": 52
    },
    {
      "epoch": 0.0005091319540340302,
      "grad_norm": 1.0938045978546143,
      "learning_rate": 9.977586413005531e-05,
      "loss": 2.0512,
      "step": 53
    },
    {
      "epoch": 0.0005187382173176911,
      "grad_norm": 0.5223602056503296,
      "learning_rate": 9.976643520762059e-05,
      "loss": 2.1225,
      "step": 54
    },
    {
      "epoch": 0.000528344480601352,
      "grad_norm": 0.7355101108551025,
      "learning_rate": 9.975681248686461e-05,
      "loss": 1.249,
      "step": 55
    },
    {
      "epoch": 0.000537950743885013,
      "grad_norm": 0.688000500202179,
      "learning_rate": 9.974699600525972e-05,
      "loss": 2.2844,
      "step": 56
    },
    {
      "epoch": 0.000547557007168674,
      "grad_norm": 0.7172989845275879,
      "learning_rate": 9.973698580103285e-05,
      "loss": 2.6309,
      "step": 57
    },
    {
      "epoch": 0.000557163270452335,
      "grad_norm": 0.6729030013084412,
      "learning_rate": 9.972678191316532e-05,
      "loss": 1.7566,
      "step": 58
    },
    {
      "epoch": 0.0005667695337359959,
      "grad_norm": 1.0324585437774658,
      "learning_rate": 9.971638438139266e-05,
      "loss": 1.4651,
      "step": 59
    },
    {
      "epoch": 0.0005763757970196568,
      "grad_norm": 1.7930604219436646,
      "learning_rate": 9.97057932462045e-05,
      "loss": 1.812,
      "step": 60
    },
    {
      "epoch": 0.0005859820603033178,
      "grad_norm": 0.6198264360427856,
      "learning_rate": 9.96950085488444e-05,
      "loss": 1.3351,
      "step": 61
    },
    {
      "epoch": 0.0005955883235869787,
      "grad_norm": 0.6650513410568237,
      "learning_rate": 9.968403033130962e-05,
      "loss": 2.12,
      "step": 62
    },
    {
      "epoch": 0.0006051945868706396,
      "grad_norm": 0.9265512228012085,
      "learning_rate": 9.967285863635112e-05,
      "loss": 0.7946,
      "step": 63
    },
    {
      "epoch": 0.0006148008501543006,
      "grad_norm": 0.751823365688324,
      "learning_rate": 9.966149350747322e-05,
      "loss": 1.2828,
      "step": 64
    },
    {
      "epoch": 0.0006244071134379615,
      "grad_norm": 1.0016423463821411,
      "learning_rate": 9.964993498893349e-05,
      "loss": 1.4028,
      "step": 65
    },
    {
      "epoch": 0.0006340133767216225,
      "grad_norm": 0.7488006949424744,
      "learning_rate": 9.963818312574265e-05,
      "loss": 0.8453,
      "step": 66
    },
    {
      "epoch": 0.0006436196400052835,
      "grad_norm": 0.8498468399047852,
      "learning_rate": 9.962623796366429e-05,
      "loss": 1.0812,
      "step": 67
    },
    {
      "epoch": 0.0006532259032889444,
      "grad_norm": 0.8496734499931335,
      "learning_rate": 9.961409954921472e-05,
      "loss": 2.0307,
      "step": 68
    },
    {
      "epoch": 0.0006628321665726053,
      "grad_norm": 0.8753388524055481,
      "learning_rate": 9.960176792966289e-05,
      "loss": 1.1429,
      "step": 69
    },
    {
      "epoch": 0.0006724384298562663,
      "grad_norm": 0.8592175245285034,
      "learning_rate": 9.958924315303004e-05,
      "loss": 1.3216,
      "step": 70
    },
    {
      "epoch": 0.0006820446931399272,
      "grad_norm": 1.4940111637115479,
      "learning_rate": 9.95765252680896e-05,
      "loss": 0.82,
      "step": 71
    },
    {
      "epoch": 0.0006916509564235881,
      "grad_norm": 1.0287102460861206,
      "learning_rate": 9.956361432436705e-05,
      "loss": 1.072,
      "step": 72
    },
    {
      "epoch": 0.0007012572197072491,
      "grad_norm": 1.0697017908096313,
      "learning_rate": 9.95505103721396e-05,
      "loss": 0.9232,
      "step": 73
    },
    {
      "epoch": 0.0007108634829909101,
      "grad_norm": 0.9282472729682922,
      "learning_rate": 9.953721346243613e-05,
      "loss": 1.3353,
      "step": 74
    },
    {
      "epoch": 0.0007204697462745711,
      "grad_norm": 0.9869083166122437,
      "learning_rate": 9.952372364703687e-05,
      "loss": 1.6519,
      "step": 75
    },
    {
      "epoch": 0.000730076009558232,
      "grad_norm": 1.2323437929153442,
      "learning_rate": 9.95100409784733e-05,
      "loss": 0.9975,
      "step": 76
    },
    {
      "epoch": 0.0007396822728418929,
      "grad_norm": 1.076454997062683,
      "learning_rate": 9.949616551002787e-05,
      "loss": 0.8598,
      "step": 77
    },
    {
      "epoch": 0.0007492885361255539,
      "grad_norm": 1.2261958122253418,
      "learning_rate": 9.948209729573383e-05,
      "loss": 1.1187,
      "step": 78
    },
    {
      "epoch": 0.0007588947994092148,
      "grad_norm": 1.4195151329040527,
      "learning_rate": 9.946783639037504e-05,
      "loss": 1.5959,
      "step": 79
    },
    {
      "epoch": 0.0007685010626928757,
      "grad_norm": 1.1138527393341064,
      "learning_rate": 9.945338284948569e-05,
      "loss": 0.6499,
      "step": 80
    },
    {
      "epoch": 0.0007781073259765367,
      "grad_norm": 1.8946585655212402,
      "learning_rate": 9.943873672935014e-05,
      "loss": 1.0448,
      "step": 81
    },
    {
      "epoch": 0.0007877135892601976,
      "grad_norm": 1.3500112295150757,
      "learning_rate": 9.942389808700269e-05,
      "loss": 1.1712,
      "step": 82
    },
    {
      "epoch": 0.0007973198525438586,
      "grad_norm": 1.1796232461929321,
      "learning_rate": 9.940886698022734e-05,
      "loss": 0.5903,
      "step": 83
    },
    {
      "epoch": 0.0008069261158275196,
      "grad_norm": 1.4405488967895508,
      "learning_rate": 9.93936434675576e-05,
      "loss": 1.0555,
      "step": 84
    },
    {
      "epoch": 0.0008165323791111805,
      "grad_norm": 1.7087458372116089,
      "learning_rate": 9.93782276082762e-05,
      "loss": 1.0939,
      "step": 85
    },
    {
      "epoch": 0.0008261386423948414,
      "grad_norm": 2.6879634857177734,
      "learning_rate": 9.936261946241492e-05,
      "loss": 1.2703,
      "step": 86
    },
    {
      "epoch": 0.0008357449056785024,
      "grad_norm": 6.579562664031982,
      "learning_rate": 9.934681909075434e-05,
      "loss": 1.4112,
      "step": 87
    },
    {
      "epoch": 0.0008453511689621633,
      "grad_norm": 1.440399169921875,
      "learning_rate": 9.93308265548236e-05,
      "loss": 1.2241,
      "step": 88
    },
    {
      "epoch": 0.0008549574322458243,
      "grad_norm": 2.0533266067504883,
      "learning_rate": 9.931464191690015e-05,
      "loss": 2.5073,
      "step": 89
    },
    {
      "epoch": 0.0008645636955294852,
      "grad_norm": 1.9462393522262573,
      "learning_rate": 9.929826524000948e-05,
      "loss": 2.1409,
      "step": 90
    },
    {
      "epoch": 0.0008741699588131461,
      "grad_norm": 2.327108383178711,
      "learning_rate": 9.928169658792498e-05,
      "loss": 2.3141,
      "step": 91
    },
    {
      "epoch": 0.0008837762220968072,
      "grad_norm": 2.5525617599487305,
      "learning_rate": 9.926493602516758e-05,
      "loss": 0.3761,
      "step": 92
    },
    {
      "epoch": 0.0008933824853804681,
      "grad_norm": 2.11832857131958,
      "learning_rate": 9.924798361700553e-05,
      "loss": 0.6106,
      "step": 93
    },
    {
      "epoch": 0.000902988748664129,
      "grad_norm": 2.1728515625,
      "learning_rate": 9.923083942945418e-05,
      "loss": 2.361,
      "step": 94
    },
    {
      "epoch": 0.00091259501194779,
      "grad_norm": 3.2426114082336426,
      "learning_rate": 9.92135035292757e-05,
      "loss": 0.9839,
      "step": 95
    },
    {
      "epoch": 0.0009222012752314509,
      "grad_norm": 4.549430847167969,
      "learning_rate": 9.91959759839788e-05,
      "loss": 1.316,
      "step": 96
    },
    {
      "epoch": 0.0009318075385151118,
      "grad_norm": 3.251258134841919,
      "learning_rate": 9.91782568618185e-05,
      "loss": 2.9985,
      "step": 97
    },
    {
      "epoch": 0.0009414138017987728,
      "grad_norm": 3.002316951751709,
      "learning_rate": 9.916034623179584e-05,
      "loss": 2.1304,
      "step": 98
    },
    {
      "epoch": 0.0009510200650824337,
      "grad_norm": 3.6586525440216064,
      "learning_rate": 9.914224416365764e-05,
      "loss": 1.9576,
      "step": 99
    },
    {
      "epoch": 0.0009606263283660947,
      "grad_norm": 4.185449123382568,
      "learning_rate": 9.91239507278962e-05,
      "loss": 3.8204,
      "step": 100
    },
    {
      "epoch": 0.0009702325916497557,
      "grad_norm": 0.23552259802818298,
      "learning_rate": 9.910546599574902e-05,
      "loss": 1.8227,
      "step": 101
    },
    {
      "epoch": 0.0009798388549334166,
      "grad_norm": 0.27870145440101624,
      "learning_rate": 9.908679003919856e-05,
      "loss": 2.0462,
      "step": 102
    },
    {
      "epoch": 0.0009894451182170774,
      "grad_norm": 0.5098013281822205,
      "learning_rate": 9.906792293097194e-05,
      "loss": 1.9969,
      "step": 103
    },
    {
      "epoch": 0.0009990513815007385,
      "grad_norm": 0.49776753783226013,
      "learning_rate": 9.904886474454063e-05,
      "loss": 1.6834,
      "step": 104
    },
    {
      "epoch": 0.0010086576447843995,
      "grad_norm": 0.4735666811466217,
      "learning_rate": 9.90296155541202e-05,
      "loss": 1.9661,
      "step": 105
    },
    {
      "epoch": 0.0010182639080680604,
      "grad_norm": 0.5975028276443481,
      "learning_rate": 9.901017543467005e-05,
      "loss": 2.1454,
      "step": 106
    },
    {
      "epoch": 0.0010278701713517214,
      "grad_norm": 0.7630074620246887,
      "learning_rate": 9.899054446189304e-05,
      "loss": 1.8832,
      "step": 107
    },
    {
      "epoch": 0.0010374764346353822,
      "grad_norm": 1.503172755241394,
      "learning_rate": 9.897072271223527e-05,
      "loss": 0.4378,
      "step": 108
    },
    {
      "epoch": 0.0010470826979190433,
      "grad_norm": 0.5636879801750183,
      "learning_rate": 9.895071026288574e-05,
      "loss": 0.1684,
      "step": 109
    },
    {
      "epoch": 0.001056688961202704,
      "grad_norm": 1.4080603122711182,
      "learning_rate": 9.893050719177608e-05,
      "loss": 0.1514,
      "step": 110
    },
    {
      "epoch": 0.0010662952244863651,
      "grad_norm": 0.9469501376152039,
      "learning_rate": 9.891011357758022e-05,
      "loss": 2.6235,
      "step": 111
    },
    {
      "epoch": 0.001075901487770026,
      "grad_norm": 1.0777138471603394,
      "learning_rate": 9.88895294997141e-05,
      "loss": 1.7301,
      "step": 112
    },
    {
      "epoch": 0.001085507751053687,
      "grad_norm": 0.896313488483429,
      "learning_rate": 9.886875503833536e-05,
      "loss": 1.0684,
      "step": 113
    },
    {
      "epoch": 0.001095114014337348,
      "grad_norm": 1.190860629081726,
      "learning_rate": 9.884779027434303e-05,
      "loss": 1.537,
      "step": 114
    },
    {
      "epoch": 0.0011047202776210089,
      "grad_norm": 1.179147481918335,
      "learning_rate": 9.882663528937717e-05,
      "loss": 0.8047,
      "step": 115
    },
    {
      "epoch": 0.00111432654090467,
      "grad_norm": 1.5248749256134033,
      "learning_rate": 9.880529016581863e-05,
      "loss": 1.1823,
      "step": 116
    },
    {
      "epoch": 0.0011239328041883307,
      "grad_norm": 1.5252033472061157,
      "learning_rate": 9.87837549867887e-05,
      "loss": 0.457,
      "step": 117
    },
    {
      "epoch": 0.0011335390674719918,
      "grad_norm": 0.7578396201133728,
      "learning_rate": 9.876202983614869e-05,
      "loss": 1.1555,
      "step": 118
    },
    {
      "epoch": 0.0011431453307556526,
      "grad_norm": 0.7523555159568787,
      "learning_rate": 9.87401147984998e-05,
      "loss": 0.6587,
      "step": 119
    },
    {
      "epoch": 0.0011527515940393137,
      "grad_norm": 0.8681097030639648,
      "learning_rate": 9.871800995918263e-05,
      "loss": 1.3397,
      "step": 120
    },
    {
      "epoch": 0.0011623578573229745,
      "grad_norm": 0.8374543786048889,
      "learning_rate": 9.869571540427689e-05,
      "loss": 1.6629,
      "step": 121
    },
    {
      "epoch": 0.0011719641206066355,
      "grad_norm": 0.9632848501205444,
      "learning_rate": 9.867323122060108e-05,
      "loss": 1.3911,
      "step": 122
    },
    {
      "epoch": 0.0011815703838902966,
      "grad_norm": 0.8265724182128906,
      "learning_rate": 9.865055749571215e-05,
      "loss": 1.8432,
      "step": 123
    },
    {
      "epoch": 0.0011911766471739574,
      "grad_norm": 1.1174460649490356,
      "learning_rate": 9.862769431790513e-05,
      "loss": 1.5187,
      "step": 124
    },
    {
      "epoch": 0.0012007829104576184,
      "grad_norm": 0.9457104206085205,
      "learning_rate": 9.860464177621284e-05,
      "loss": 1.6047,
      "step": 125
    },
    {
      "epoch": 0.0012103891737412793,
      "grad_norm": 0.9461504220962524,
      "learning_rate": 9.858139996040553e-05,
      "loss": 1.0638,
      "step": 126
    },
    {
      "epoch": 0.0012199954370249403,
      "grad_norm": 1.2864220142364502,
      "learning_rate": 9.855796896099045e-05,
      "loss": 1.1593,
      "step": 127
    },
    {
      "epoch": 0.0012296017003086011,
      "grad_norm": 0.9703513979911804,
      "learning_rate": 9.853434886921158e-05,
      "loss": 0.6943,
      "step": 128
    },
    {
      "epoch": 0.0012392079635922622,
      "grad_norm": 1.7747596502304077,
      "learning_rate": 9.851053977704931e-05,
      "loss": 1.4437,
      "step": 129
    },
    {
      "epoch": 0.001248814226875923,
      "grad_norm": 1.3455934524536133,
      "learning_rate": 9.848654177721998e-05,
      "loss": 0.7812,
      "step": 130
    },
    {
      "epoch": 0.001258420490159584,
      "grad_norm": 1.3344652652740479,
      "learning_rate": 9.846235496317555e-05,
      "loss": 1.1207,
      "step": 131
    },
    {
      "epoch": 0.001268026753443245,
      "grad_norm": 1.470723032951355,
      "learning_rate": 9.843797942910328e-05,
      "loss": 1.7458,
      "step": 132
    },
    {
      "epoch": 0.001277633016726906,
      "grad_norm": 1.446967363357544,
      "learning_rate": 9.841341526992536e-05,
      "loss": 1.3464,
      "step": 133
    },
    {
      "epoch": 0.001287239280010567,
      "grad_norm": 1.2281447649002075,
      "learning_rate": 9.838866258129847e-05,
      "loss": 1.2668,
      "step": 134
    },
    {
      "epoch": 0.0012968455432942278,
      "grad_norm": 1.6069281101226807,
      "learning_rate": 9.836372145961345e-05,
      "loss": 0.848,
      "step": 135
    },
    {
      "epoch": 0.0013064518065778888,
      "grad_norm": 1.3499243259429932,
      "learning_rate": 9.833859200199497e-05,
      "loss": 0.7165,
      "step": 136
    },
    {
      "epoch": 0.0013160580698615496,
      "grad_norm": 1.5303913354873657,
      "learning_rate": 9.83132743063011e-05,
      "loss": 0.3277,
      "step": 137
    },
    {
      "epoch": 0.0013256643331452107,
      "grad_norm": 1.4204318523406982,
      "learning_rate": 9.82877684711229e-05,
      "loss": 2.6048,
      "step": 138
    },
    {
      "epoch": 0.0013352705964288717,
      "grad_norm": 1.4058451652526855,
      "learning_rate": 9.826207459578411e-05,
      "loss": 0.6269,
      "step": 139
    },
    {
      "epoch": 0.0013448768597125326,
      "grad_norm": 2.362095355987549,
      "learning_rate": 9.823619278034072e-05,
      "loss": 1.9204,
      "step": 140
    },
    {
      "epoch": 0.0013544831229961936,
      "grad_norm": 2.1079070568084717,
      "learning_rate": 9.821012312558058e-05,
      "loss": 1.2442,
      "step": 141
    },
    {
      "epoch": 0.0013640893862798544,
      "grad_norm": 1.9698988199234009,
      "learning_rate": 9.818386573302305e-05,
      "loss": 1.8193,
      "step": 142
    },
    {
      "epoch": 0.0013736956495635155,
      "grad_norm": 2.0987484455108643,
      "learning_rate": 9.815742070491852e-05,
      "loss": 1.5768,
      "step": 143
    },
    {
      "epoch": 0.0013833019128471763,
      "grad_norm": 2.337153673171997,
      "learning_rate": 9.81307881442481e-05,
      "loss": 1.6459,
      "step": 144
    },
    {
      "epoch": 0.0013929081761308373,
      "grad_norm": 3.7250924110412598,
      "learning_rate": 9.810396815472314e-05,
      "loss": 1.1638,
      "step": 145
    },
    {
      "epoch": 0.0014025144394144982,
      "grad_norm": 5.047917366027832,
      "learning_rate": 9.807696084078493e-05,
      "loss": 2.0148,
      "step": 146
    },
    {
      "epoch": 0.0014121207026981592,
      "grad_norm": 3.966524124145508,
      "learning_rate": 9.804976630760419e-05,
      "loss": 1.9081,
      "step": 147
    },
    {
      "epoch": 0.0014217269659818203,
      "grad_norm": 2.8529458045959473,
      "learning_rate": 9.802238466108068e-05,
      "loss": 2.399,
      "step": 148
    },
    {
      "epoch": 0.001431333229265481,
      "grad_norm": 2.899770975112915,
      "learning_rate": 9.799481600784286e-05,
      "loss": 2.42,
      "step": 149
    },
    {
      "epoch": 0.0014409394925491421,
      "grad_norm": 5.4804463386535645,
      "learning_rate": 9.796706045524738e-05,
      "loss": 2.6402,
      "step": 150
    },
    {
      "epoch": 0.001450545755832803,
      "grad_norm": 0.24980565905570984,
      "learning_rate": 9.793911811137875e-05,
      "loss": 1.963,
      "step": 151
    },
    {
      "epoch": 0.001460152019116464,
      "grad_norm": 0.3728490173816681,
      "learning_rate": 9.791098908504884e-05,
      "loss": 1.8289,
      "step": 152
    },
    {
      "epoch": 0.0014697582824001248,
      "grad_norm": 0.4433547556400299,
      "learning_rate": 9.788267348579648e-05,
      "loss": 1.1267,
      "step": 153
    },
    {
      "epoch": 0.0014793645456837859,
      "grad_norm": 0.8421958684921265,
      "learning_rate": 9.785417142388709e-05,
      "loss": 1.1472,
      "step": 154
    },
    {
      "epoch": 0.0014889708089674467,
      "grad_norm": 0.6756650805473328,
      "learning_rate": 9.782548301031217e-05,
      "loss": 1.2683,
      "step": 155
    },
    {
      "epoch": 0.0014985770722511077,
      "grad_norm": 0.7339028120040894,
      "learning_rate": 9.77966083567889e-05,
      "loss": 1.6852,
      "step": 156
    },
    {
      "epoch": 0.0015081833355347688,
      "grad_norm": 0.4794408679008484,
      "learning_rate": 9.776754757575975e-05,
      "loss": 2.2113,
      "step": 157
    },
    {
      "epoch": 0.0015177895988184296,
      "grad_norm": 1.1240463256835938,
      "learning_rate": 9.773830078039193e-05,
      "loss": 1.2598,
      "step": 158
    },
    {
      "epoch": 0.0015273958621020906,
      "grad_norm": 0.7506759762763977,
      "learning_rate": 9.770886808457709e-05,
      "loss": 1.3998,
      "step": 159
    },
    {
      "epoch": 0.0015370021253857515,
      "grad_norm": 1.0879539251327515,
      "learning_rate": 9.767924960293076e-05,
      "loss": 0.6049,
      "step": 160
    },
    {
      "epoch": 0.0015466083886694125,
      "grad_norm": 0.59916090965271,
      "learning_rate": 9.764944545079196e-05,
      "loss": 1.3074,
      "step": 161
    },
    {
      "epoch": 0.0015562146519530733,
      "grad_norm": 0.9513159990310669,
      "learning_rate": 9.761945574422276e-05,
      "loss": 1.9838,
      "step": 162
    },
    {
      "epoch": 0.0015658209152367344,
      "grad_norm": 0.9434720277786255,
      "learning_rate": 9.758928060000778e-05,
      "loss": 1.7917,
      "step": 163
    },
    {
      "epoch": 0.0015754271785203952,
      "grad_norm": 0.9315809607505798,
      "learning_rate": 9.755892013565376e-05,
      "loss": 0.5713,
      "step": 164
    },
    {
      "epoch": 0.0015850334418040562,
      "grad_norm": 1.1091091632843018,
      "learning_rate": 9.752837446938915e-05,
      "loss": 0.5943,
      "step": 165
    },
    {
      "epoch": 0.0015946397050877173,
      "grad_norm": 0.6209042072296143,
      "learning_rate": 9.749764372016355e-05,
      "loss": 1.1236,
      "step": 166
    },
    {
      "epoch": 0.0016042459683713781,
      "grad_norm": 0.8369060754776001,
      "learning_rate": 9.746672800764735e-05,
      "loss": 0.9256,
      "step": 167
    },
    {
      "epoch": 0.0016138522316550392,
      "grad_norm": 1.1757186651229858,
      "learning_rate": 9.743562745223117e-05,
      "loss": 0.9426,
      "step": 168
    },
    {
      "epoch": 0.0016234584949387,
      "grad_norm": 0.9146075248718262,
      "learning_rate": 9.740434217502547e-05,
      "loss": 1.1608,
      "step": 169
    },
    {
      "epoch": 0.001633064758222361,
      "grad_norm": 0.856109082698822,
      "learning_rate": 9.737287229786006e-05,
      "loss": 1.0219,
      "step": 170
    },
    {
      "epoch": 0.0016426710215060219,
      "grad_norm": 0.8393319845199585,
      "learning_rate": 9.734121794328357e-05,
      "loss": 1.0764,
      "step": 171
    },
    {
      "epoch": 0.001652277284789683,
      "grad_norm": 0.7857136726379395,
      "learning_rate": 9.730937923456302e-05,
      "loss": 0.7437,
      "step": 172
    },
    {
      "epoch": 0.0016618835480733437,
      "grad_norm": 1.101627230644226,
      "learning_rate": 9.727735629568336e-05,
      "loss": 1.4182,
      "step": 173
    },
    {
      "epoch": 0.0016714898113570048,
      "grad_norm": 1.2909857034683228,
      "learning_rate": 9.724514925134694e-05,
      "loss": 1.6612,
      "step": 174
    },
    {
      "epoch": 0.0016810960746406658,
      "grad_norm": 1.5066879987716675,
      "learning_rate": 9.721275822697306e-05,
      "loss": 1.6387,
      "step": 175
    },
    {
      "epoch": 0.0016907023379243266,
      "grad_norm": 1.166881799697876,
      "learning_rate": 9.718018334869747e-05,
      "loss": 1.5044,
      "step": 176
    },
    {
      "epoch": 0.0017003086012079877,
      "grad_norm": 1.3287938833236694,
      "learning_rate": 9.714742474337186e-05,
      "loss": 1.3843,
      "step": 177
    },
    {
      "epoch": 0.0017099148644916485,
      "grad_norm": 1.5930442810058594,
      "learning_rate": 9.711448253856337e-05,
      "loss": 1.2141,
      "step": 178
    },
    {
      "epoch": 0.0017195211277753095,
      "grad_norm": 0.8705190420150757,
      "learning_rate": 9.708135686255416e-05,
      "loss": 0.6912,
      "step": 179
    },
    {
      "epoch": 0.0017291273910589704,
      "grad_norm": 1.4970552921295166,
      "learning_rate": 9.704804784434076e-05,
      "loss": 1.2565,
      "step": 180
    },
    {
      "epoch": 0.0017387336543426314,
      "grad_norm": 1.582898497581482,
      "learning_rate": 9.701455561363379e-05,
      "loss": 1.4725,
      "step": 181
    },
    {
      "epoch": 0.0017483399176262922,
      "grad_norm": 1.9624470472335815,
      "learning_rate": 9.698088030085722e-05,
      "loss": 0.7901,
      "step": 182
    },
    {
      "epoch": 0.0017579461809099533,
      "grad_norm": 0.8613381385803223,
      "learning_rate": 9.6947022037148e-05,
      "loss": 1.0455,
      "step": 183
    },
    {
      "epoch": 0.0017675524441936143,
      "grad_norm": 1.2065166234970093,
      "learning_rate": 9.69129809543556e-05,
      "loss": 0.5656,
      "step": 184
    },
    {
      "epoch": 0.0017771587074772752,
      "grad_norm": 1.3584308624267578,
      "learning_rate": 9.687875718504126e-05,
      "loss": 0.8785,
      "step": 185
    },
    {
      "epoch": 0.0017867649707609362,
      "grad_norm": 1.8783706426620483,
      "learning_rate": 9.684435086247777e-05,
      "loss": 1.0543,
      "step": 186
    },
    {
      "epoch": 0.001796371234044597,
      "grad_norm": 1.729237675666809,
      "learning_rate": 9.680976212064874e-05,
      "loss": 1.411,
      "step": 187
    },
    {
      "epoch": 0.001805977497328258,
      "grad_norm": 1.2262932062149048,
      "learning_rate": 9.677499109424818e-05,
      "loss": 1.4733,
      "step": 188
    },
    {
      "epoch": 0.0018155837606119189,
      "grad_norm": 1.9267388582229614,
      "learning_rate": 9.674003791867991e-05,
      "loss": 2.1442,
      "step": 189
    },
    {
      "epoch": 0.00182519002389558,
      "grad_norm": 1.4836162328720093,
      "learning_rate": 9.670490273005712e-05,
      "loss": 1.3679,
      "step": 190
    },
    {
      "epoch": 0.0018347962871792408,
      "grad_norm": 1.253038763999939,
      "learning_rate": 9.666958566520174e-05,
      "loss": 0.9467,
      "step": 191
    },
    {
      "epoch": 0.0018444025504629018,
      "grad_norm": 1.407757043838501,
      "learning_rate": 9.663408686164399e-05,
      "loss": 0.8622,
      "step": 192
    },
    {
      "epoch": 0.0018540088137465628,
      "grad_norm": 1.9592434167861938,
      "learning_rate": 9.659840645762175e-05,
      "loss": 1.3271,
      "step": 193
    },
    {
      "epoch": 0.0018636150770302237,
      "grad_norm": 2.5765817165374756,
      "learning_rate": 9.656254459208015e-05,
      "loss": 2.7011,
      "step": 194
    },
    {
      "epoch": 0.0018732213403138847,
      "grad_norm": 2.1232407093048096,
      "learning_rate": 9.652650140467093e-05,
      "loss": 2.0552,
      "step": 195
    },
    {
      "epoch": 0.0018828276035975455,
      "grad_norm": 2.4416000843048096,
      "learning_rate": 9.649027703575193e-05,
      "loss": 2.0104,
      "step": 196
    },
    {
      "epoch": 0.0018924338668812066,
      "grad_norm": 3.398444890975952,
      "learning_rate": 9.645387162638652e-05,
      "loss": 2.5781,
      "step": 197
    },
    {
      "epoch": 0.0019020401301648674,
      "grad_norm": 3.549269676208496,
      "learning_rate": 9.641728531834313e-05,
      "loss": 1.8268,
      "step": 198
    },
    {
      "epoch": 0.0019116463934485284,
      "grad_norm": 2.9552695751190186,
      "learning_rate": 9.638051825409453e-05,
      "loss": 2.6035,
      "step": 199
    },
    {
      "epoch": 0.0019212526567321895,
      "grad_norm": 3.7393410205841064,
      "learning_rate": 9.634357057681749e-05,
      "loss": 3.7814,
      "step": 200
    },
    {
      "epoch": 0.0019308589200158503,
      "grad_norm": 0.2191723883152008,
      "learning_rate": 9.630644243039207e-05,
      "loss": 1.8964,
      "step": 201
    },
    {
      "epoch": 0.0019404651832995114,
      "grad_norm": 0.30522456765174866,
      "learning_rate": 9.626913395940109e-05,
      "loss": 2.0684,
      "step": 202
    },
    {
      "epoch": 0.0019500714465831722,
      "grad_norm": 0.2926991879940033,
      "learning_rate": 9.623164530912963e-05,
      "loss": 2.1595,
      "step": 203
    },
    {
      "epoch": 0.0019596777098668332,
      "grad_norm": 0.8309841156005859,
      "learning_rate": 9.619397662556435e-05,
      "loss": 0.4848,
      "step": 204
    },
    {
      "epoch": 0.0019692839731504943,
      "grad_norm": 0.5335772037506104,
      "learning_rate": 9.615612805539305e-05,
      "loss": 1.4813,
      "step": 205
    },
    {
      "epoch": 0.001978890236434155,
      "grad_norm": 0.574340283870697,
      "learning_rate": 9.6118099746004e-05,
      "loss": 1.8122,
      "step": 206
    },
    {
      "epoch": 0.001988496499717816,
      "grad_norm": 0.5693013072013855,
      "learning_rate": 9.607989184548543e-05,
      "loss": 1.7344,
      "step": 207
    },
    {
      "epoch": 0.001998102763001477,
      "grad_norm": 1.1894689798355103,
      "learning_rate": 9.604150450262488e-05,
      "loss": 0.8135,
      "step": 208
    },
    {
      "epoch": 0.002007709026285138,
      "grad_norm": 0.5404132008552551,
      "learning_rate": 9.600293786690872e-05,
      "loss": 0.6558,
      "step": 209
    },
    {
      "epoch": 0.002017315289568799,
      "grad_norm": 0.9760227799415588,
      "learning_rate": 9.596419208852151e-05,
      "loss": 0.2519,
      "step": 210
    },
    {
      "epoch": 0.0020269215528524597,
      "grad_norm": 0.8883073329925537,
      "learning_rate": 9.592526731834537e-05,
      "loss": 0.5408,
      "step": 211
    },
    {
      "epoch": 0.0020365278161361207,
      "grad_norm": 0.7274699211120605,
      "learning_rate": 9.588616370795947e-05,
      "loss": 2.3094,
      "step": 212
    },
    {
      "epoch": 0.0020461340794197817,
      "grad_norm": 1.019194483757019,
      "learning_rate": 9.584688140963944e-05,
      "loss": 0.4492,
      "step": 213
    },
    {
      "epoch": 0.002055740342703443,
      "grad_norm": 0.9480285048484802,
      "learning_rate": 9.580742057635673e-05,
      "loss": 1.6375,
      "step": 214
    },
    {
      "epoch": 0.0020653466059871034,
      "grad_norm": 0.7404997944831848,
      "learning_rate": 9.576778136177798e-05,
      "loss": 1.1661,
      "step": 215
    },
    {
      "epoch": 0.0020749528692707644,
      "grad_norm": 1.0438519716262817,
      "learning_rate": 9.572796392026456e-05,
      "loss": 1.685,
      "step": 216
    },
    {
      "epoch": 0.0020845591325544255,
      "grad_norm": 1.13591468334198,
      "learning_rate": 9.568796840687184e-05,
      "loss": 1.5669,
      "step": 217
    },
    {
      "epoch": 0.0020941653958380865,
      "grad_norm": 0.9734657406806946,
      "learning_rate": 9.564779497734864e-05,
      "loss": 1.2723,
      "step": 218
    },
    {
      "epoch": 0.0021037716591217476,
      "grad_norm": 0.9094304442405701,
      "learning_rate": 9.560744378813659e-05,
      "loss": 0.6023,
      "step": 219
    },
    {
      "epoch": 0.002113377922405408,
      "grad_norm": 0.9920042157173157,
      "learning_rate": 9.556691499636959e-05,
      "loss": 1.9717,
      "step": 220
    },
    {
      "epoch": 0.0021229841856890692,
      "grad_norm": 1.5184837579727173,
      "learning_rate": 9.552620875987311e-05,
      "loss": 0.8786,
      "step": 221
    },
    {
      "epoch": 0.0021325904489727303,
      "grad_norm": 0.8060744404792786,
      "learning_rate": 9.548532523716364e-05,
      "loss": 0.7567,
      "step": 222
    },
    {
      "epoch": 0.0021421967122563913,
      "grad_norm": 0.8059457540512085,
      "learning_rate": 9.544426458744804e-05,
      "loss": 1.0441,
      "step": 223
    },
    {
      "epoch": 0.002151802975540052,
      "grad_norm": 1.0007010698318481,
      "learning_rate": 9.540302697062293e-05,
      "loss": 1.0348,
      "step": 224
    },
    {
      "epoch": 0.002161409238823713,
      "grad_norm": 0.9019158482551575,
      "learning_rate": 9.536161254727408e-05,
      "loss": 0.6413,
      "step": 225
    },
    {
      "epoch": 0.002171015502107374,
      "grad_norm": 0.8721305727958679,
      "learning_rate": 9.532002147867575e-05,
      "loss": 2.1991,
      "step": 226
    },
    {
      "epoch": 0.002180621765391035,
      "grad_norm": 1.0612828731536865,
      "learning_rate": 9.527825392679012e-05,
      "loss": 0.7658,
      "step": 227
    },
    {
      "epoch": 0.002190228028674696,
      "grad_norm": 1.195090889930725,
      "learning_rate": 9.523631005426658e-05,
      "loss": 1.424,
      "step": 228
    },
    {
      "epoch": 0.0021998342919583567,
      "grad_norm": 1.0510435104370117,
      "learning_rate": 9.51941900244412e-05,
      "loss": 1.3153,
      "step": 229
    },
    {
      "epoch": 0.0022094405552420177,
      "grad_norm": 1.0679879188537598,
      "learning_rate": 9.515189400133593e-05,
      "loss": 1.3591,
      "step": 230
    },
    {
      "epoch": 0.002219046818525679,
      "grad_norm": 1.1337720155715942,
      "learning_rate": 9.51094221496582e-05,
      "loss": 1.2074,
      "step": 231
    },
    {
      "epoch": 0.00222865308180934,
      "grad_norm": 1.204034447669983,
      "learning_rate": 9.506677463480003e-05,
      "loss": 1.3149,
      "step": 232
    },
    {
      "epoch": 0.0022382593450930004,
      "grad_norm": 1.8811888694763184,
      "learning_rate": 9.502395162283759e-05,
      "loss": 1.5045,
      "step": 233
    },
    {
      "epoch": 0.0022478656083766615,
      "grad_norm": 1.2537976503372192,
      "learning_rate": 9.498095328053039e-05,
      "loss": 1.0997,
      "step": 234
    },
    {
      "epoch": 0.0022574718716603225,
      "grad_norm": 1.4280935525894165,
      "learning_rate": 9.493777977532072e-05,
      "loss": 1.3604,
      "step": 235
    },
    {
      "epoch": 0.0022670781349439836,
      "grad_norm": 1.7210010290145874,
      "learning_rate": 9.489443127533304e-05,
      "loss": 1.6693,
      "step": 236
    },
    {
      "epoch": 0.0022766843982276446,
      "grad_norm": 1.517876386642456,
      "learning_rate": 9.485090794937319e-05,
      "loss": 1.4608,
      "step": 237
    },
    {
      "epoch": 0.0022862906615113052,
      "grad_norm": 1.1853421926498413,
      "learning_rate": 9.480720996692784e-05,
      "loss": 0.82,
      "step": 238
    },
    {
      "epoch": 0.0022958969247949663,
      "grad_norm": 1.859249234199524,
      "learning_rate": 9.476333749816382e-05,
      "loss": 1.4565,
      "step": 239
    },
    {
      "epoch": 0.0023055031880786273,
      "grad_norm": 2.3003361225128174,
      "learning_rate": 9.47192907139274e-05,
      "loss": 1.1078,
      "step": 240
    },
    {
      "epoch": 0.0023151094513622883,
      "grad_norm": 1.7903460264205933,
      "learning_rate": 9.467506978574371e-05,
      "loss": 1.8671,
      "step": 241
    },
    {
      "epoch": 0.002324715714645949,
      "grad_norm": 1.7605727910995483,
      "learning_rate": 9.463067488581597e-05,
      "loss": 1.7443,
      "step": 242
    },
    {
      "epoch": 0.00233432197792961,
      "grad_norm": 1.580639362335205,
      "learning_rate": 9.45861061870249e-05,
      "loss": 1.794,
      "step": 243
    },
    {
      "epoch": 0.002343928241213271,
      "grad_norm": 1.8136929273605347,
      "learning_rate": 9.454136386292804e-05,
      "loss": 1.4805,
      "step": 244
    },
    {
      "epoch": 0.002353534504496932,
      "grad_norm": 2.0361876487731934,
      "learning_rate": 9.449644808775902e-05,
      "loss": 1.6655,
      "step": 245
    },
    {
      "epoch": 0.002363140767780593,
      "grad_norm": 3.427790880203247,
      "learning_rate": 9.445135903642693e-05,
      "loss": 1.4443,
      "step": 246
    },
    {
      "epoch": 0.0023727470310642537,
      "grad_norm": 2.3827383518218994,
      "learning_rate": 9.44060968845156e-05,
      "loss": 1.4395,
      "step": 247
    },
    {
      "epoch": 0.0023823532943479148,
      "grad_norm": 2.609541177749634,
      "learning_rate": 9.4360661808283e-05,
      "loss": 1.6436,
      "step": 248
    },
    {
      "epoch": 0.002391959557631576,
      "grad_norm": 3.636117935180664,
      "learning_rate": 9.431505398466045e-05,
      "loss": 1.7045,
      "step": 249
    },
    {
      "epoch": 0.002401565820915237,
      "grad_norm": 3.470595359802246,
      "learning_rate": 9.426927359125195e-05,
      "loss": 2.6241,
      "step": 250
    },
    {
      "epoch": 0.0024111720841988975,
      "grad_norm": 0.2589930593967438,
      "learning_rate": 9.42233208063336e-05,
      "loss": 1.943,
      "step": 251
    },
    {
      "epoch": 0.0024207783474825585,
      "grad_norm": 0.29414838552474976,
      "learning_rate": 9.417719580885274e-05,
      "loss": 2.131,
      "step": 252
    },
    {
      "epoch": 0.0024303846107662196,
      "grad_norm": 0.5100934505462646,
      "learning_rate": 9.413089877842736e-05,
      "loss": 2.0999,
      "step": 253
    },
    {
      "epoch": 0.0024399908740498806,
      "grad_norm": 0.37613093852996826,
      "learning_rate": 9.408442989534536e-05,
      "loss": 1.84,
      "step": 254
    },
    {
      "epoch": 0.0024495971373335416,
      "grad_norm": 0.5497903227806091,
      "learning_rate": 9.403778934056391e-05,
      "loss": 2.0774,
      "step": 255
    },
    {
      "epoch": 0.0024592034006172023,
      "grad_norm": 0.4737769067287445,
      "learning_rate": 9.399097729570865e-05,
      "loss": 2.3613,
      "step": 256
    },
    {
      "epoch": 0.0024688096639008633,
      "grad_norm": 0.8594376444816589,
      "learning_rate": 9.394399394307303e-05,
      "loss": 1.9636,
      "step": 257
    },
    {
      "epoch": 0.0024784159271845243,
      "grad_norm": 0.862769603729248,
      "learning_rate": 9.389683946561761e-05,
      "loss": 1.6446,
      "step": 258
    },
    {
      "epoch": 0.0024880221904681854,
      "grad_norm": 0.6548369526863098,
      "learning_rate": 9.384951404696933e-05,
      "loss": 1.2672,
      "step": 259
    },
    {
      "epoch": 0.002497628453751846,
      "grad_norm": 0.9146730303764343,
      "learning_rate": 9.380201787142085e-05,
      "loss": 1.9785,
      "step": 260
    },
    {
      "epoch": 0.002507234717035507,
      "grad_norm": 1.1508115530014038,
      "learning_rate": 9.375435112392969e-05,
      "loss": 0.596,
      "step": 261
    },
    {
      "epoch": 0.002516840980319168,
      "grad_norm": 0.7571190595626831,
      "learning_rate": 9.37065139901177e-05,
      "loss": 2.4006,
      "step": 262
    },
    {
      "epoch": 0.002526447243602829,
      "grad_norm": 1.2918925285339355,
      "learning_rate": 9.365850665627016e-05,
      "loss": 1.6134,
      "step": 263
    },
    {
      "epoch": 0.00253605350688649,
      "grad_norm": 1.0938349962234497,
      "learning_rate": 9.361032930933519e-05,
      "loss": 1.5395,
      "step": 264
    },
    {
      "epoch": 0.0025456597701701508,
      "grad_norm": 0.8132850527763367,
      "learning_rate": 9.356198213692297e-05,
      "loss": 1.3871,
      "step": 265
    },
    {
      "epoch": 0.002555266033453812,
      "grad_norm": 0.7870836853981018,
      "learning_rate": 9.351346532730498e-05,
      "loss": 1.4398,
      "step": 266
    },
    {
      "epoch": 0.002564872296737473,
      "grad_norm": 0.8855361342430115,
      "learning_rate": 9.346477906941331e-05,
      "loss": 1.5178,
      "step": 267
    },
    {
      "epoch": 0.002574478560021134,
      "grad_norm": 1.8722375631332397,
      "learning_rate": 9.341592355283986e-05,
      "loss": 1.2086,
      "step": 268
    },
    {
      "epoch": 0.0025840848233047945,
      "grad_norm": 1.4349762201309204,
      "learning_rate": 9.336689896783573e-05,
      "loss": 0.5813,
      "step": 269
    },
    {
      "epoch": 0.0025936910865884556,
      "grad_norm": 0.8224892020225525,
      "learning_rate": 9.331770550531036e-05,
      "loss": 1.1649,
      "step": 270
    },
    {
      "epoch": 0.0026032973498721166,
      "grad_norm": 1.2980382442474365,
      "learning_rate": 9.32683433568308e-05,
      "loss": 1.4244,
      "step": 271
    },
    {
      "epoch": 0.0026129036131557776,
      "grad_norm": 1.167806625366211,
      "learning_rate": 9.321881271462104e-05,
      "loss": 0.8606,
      "step": 272
    },
    {
      "epoch": 0.0026225098764394387,
      "grad_norm": 0.6378973126411438,
      "learning_rate": 9.316911377156117e-05,
      "loss": 0.7707,
      "step": 273
    },
    {
      "epoch": 0.0026321161397230993,
      "grad_norm": 1.2472058534622192,
      "learning_rate": 9.311924672118671e-05,
      "loss": 1.3648,
      "step": 274
    },
    {
      "epoch": 0.0026417224030067603,
      "grad_norm": 0.9658389091491699,
      "learning_rate": 9.306921175768775e-05,
      "loss": 0.8266,
      "step": 275
    },
    {
      "epoch": 0.0026513286662904214,
      "grad_norm": 1.186858057975769,
      "learning_rate": 9.301900907590834e-05,
      "loss": 0.6171,
      "step": 276
    },
    {
      "epoch": 0.0026609349295740824,
      "grad_norm": 0.9427700638771057,
      "learning_rate": 9.29686388713456e-05,
      "loss": 1.0201,
      "step": 277
    },
    {
      "epoch": 0.0026705411928577435,
      "grad_norm": 1.1751242876052856,
      "learning_rate": 9.291810134014903e-05,
      "loss": 0.9516,
      "step": 278
    },
    {
      "epoch": 0.002680147456141404,
      "grad_norm": 1.0743447542190552,
      "learning_rate": 9.286739667911972e-05,
      "loss": 1.3508,
      "step": 279
    },
    {
      "epoch": 0.002689753719425065,
      "grad_norm": 0.9673576354980469,
      "learning_rate": 9.281652508570957e-05,
      "loss": 1.4203,
      "step": 280
    },
    {
      "epoch": 0.002699359982708726,
      "grad_norm": 1.3510057926177979,
      "learning_rate": 9.276548675802059e-05,
      "loss": 1.1467,
      "step": 281
    },
    {
      "epoch": 0.002708966245992387,
      "grad_norm": 1.318784236907959,
      "learning_rate": 9.271428189480404e-05,
      "loss": 1.0126,
      "step": 282
    },
    {
      "epoch": 0.002718572509276048,
      "grad_norm": 1.3473984003067017,
      "learning_rate": 9.266291069545972e-05,
      "loss": 0.9028,
      "step": 283
    },
    {
      "epoch": 0.002728178772559709,
      "grad_norm": 1.5032799243927002,
      "learning_rate": 9.261137336003512e-05,
      "loss": 1.6103,
      "step": 284
    },
    {
      "epoch": 0.00273778503584337,
      "grad_norm": 1.5539953708648682,
      "learning_rate": 9.255967008922474e-05,
      "loss": 1.6537,
      "step": 285
    },
    {
      "epoch": 0.002747391299127031,
      "grad_norm": 1.5699712038040161,
      "learning_rate": 9.250780108436925e-05,
      "loss": 0.9599,
      "step": 286
    },
    {
      "epoch": 0.002756997562410692,
      "grad_norm": 1.534355878829956,
      "learning_rate": 9.245576654745471e-05,
      "loss": 1.7713,
      "step": 287
    },
    {
      "epoch": 0.0027666038256943526,
      "grad_norm": 2.026632308959961,
      "learning_rate": 9.240356668111179e-05,
      "loss": 0.5741,
      "step": 288
    },
    {
      "epoch": 0.0027762100889780136,
      "grad_norm": 1.1314491033554077,
      "learning_rate": 9.235120168861496e-05,
      "loss": 0.9898,
      "step": 289
    },
    {
      "epoch": 0.0027858163522616747,
      "grad_norm": 2.2368247509002686,
      "learning_rate": 9.229867177388172e-05,
      "loss": 2.0135,
      "step": 290
    },
    {
      "epoch": 0.0027954226155453357,
      "grad_norm": 1.7465248107910156,
      "learning_rate": 9.224597714147186e-05,
      "loss": 1.1482,
      "step": 291
    },
    {
      "epoch": 0.0028050288788289963,
      "grad_norm": 1.8126107454299927,
      "learning_rate": 9.219311799658652e-05,
      "loss": 2.5589,
      "step": 292
    },
    {
      "epoch": 0.0028146351421126574,
      "grad_norm": 3.138742446899414,
      "learning_rate": 9.214009454506753e-05,
      "loss": 2.2879,
      "step": 293
    },
    {
      "epoch": 0.0028242414053963184,
      "grad_norm": 2.1528890132904053,
      "learning_rate": 9.208690699339657e-05,
      "loss": 1.5997,
      "step": 294
    },
    {
      "epoch": 0.0028338476686799795,
      "grad_norm": 2.031599283218384,
      "learning_rate": 9.203355554869428e-05,
      "loss": 2.6958,
      "step": 295
    },
    {
      "epoch": 0.0028434539319636405,
      "grad_norm": 4.82572603225708,
      "learning_rate": 9.198004041871961e-05,
      "loss": 1.5728,
      "step": 296
    },
    {
      "epoch": 0.002853060195247301,
      "grad_norm": 2.179384469985962,
      "learning_rate": 9.192636181186888e-05,
      "loss": 2.6932,
      "step": 297
    },
    {
      "epoch": 0.002862666458530962,
      "grad_norm": 2.6558616161346436,
      "learning_rate": 9.187251993717501e-05,
      "loss": 2.1433,
      "step": 298
    },
    {
      "epoch": 0.002872272721814623,
      "grad_norm": 2.6391353607177734,
      "learning_rate": 9.181851500430673e-05,
      "loss": 2.1276,
      "step": 299
    },
    {
      "epoch": 0.0028818789850982842,
      "grad_norm": 3.906266689300537,
      "learning_rate": 9.176434722356773e-05,
      "loss": 3.2244,
      "step": 300
    },
    {
      "epoch": 0.002891485248381945,
      "grad_norm": 0.22504565119743347,
      "learning_rate": 9.171001680589588e-05,
      "loss": 1.9156,
      "step": 301
    },
    {
      "epoch": 0.002901091511665606,
      "grad_norm": 0.3022899627685547,
      "learning_rate": 9.165552396286236e-05,
      "loss": 2.0682,
      "step": 302
    },
    {
      "epoch": 0.002910697774949267,
      "grad_norm": 0.4657721221446991,
      "learning_rate": 9.160086890667086e-05,
      "loss": 2.0263,
      "step": 303
    },
    {
      "epoch": 0.002920304038232928,
      "grad_norm": 0.46757808327674866,
      "learning_rate": 9.154605185015677e-05,
      "loss": 1.2894,
      "step": 304
    },
    {
      "epoch": 0.002929910301516589,
      "grad_norm": 0.5271124839782715,
      "learning_rate": 9.14910730067863e-05,
      "loss": 1.701,
      "step": 305
    },
    {
      "epoch": 0.0029395165648002496,
      "grad_norm": 0.8139910697937012,
      "learning_rate": 9.143593259065572e-05,
      "loss": 1.2558,
      "step": 306
    },
    {
      "epoch": 0.0029491228280839107,
      "grad_norm": 0.9099153280258179,
      "learning_rate": 9.138063081649051e-05,
      "loss": 1.9246,
      "step": 307
    },
    {
      "epoch": 0.0029587290913675717,
      "grad_norm": 0.5122951865196228,
      "learning_rate": 9.132516789964442e-05,
      "loss": 2.1474,
      "step": 308
    },
    {
      "epoch": 0.0029683353546512328,
      "grad_norm": 0.7223782539367676,
      "learning_rate": 9.126954405609882e-05,
      "loss": 1.3598,
      "step": 309
    },
    {
      "epoch": 0.0029779416179348934,
      "grad_norm": 0.9512868523597717,
      "learning_rate": 9.121375950246165e-05,
      "loss": 1.7712,
      "step": 310
    },
    {
      "epoch": 0.0029875478812185544,
      "grad_norm": 0.6798685193061829,
      "learning_rate": 9.115781445596676e-05,
      "loss": 2.4221,
      "step": 311
    },
    {
      "epoch": 0.0029971541445022155,
      "grad_norm": 0.8581318259239197,
      "learning_rate": 9.110170913447294e-05,
      "loss": 1.9201,
      "step": 312
    },
    {
      "epoch": 0.0030067604077858765,
      "grad_norm": 0.9510321021080017,
      "learning_rate": 9.104544375646313e-05,
      "loss": 1.7797,
      "step": 313
    },
    {
      "epoch": 0.0030163666710695375,
      "grad_norm": 1.0847291946411133,
      "learning_rate": 9.098901854104358e-05,
      "loss": 1.7707,
      "step": 314
    },
    {
      "epoch": 0.003025972934353198,
      "grad_norm": 1.4748034477233887,
      "learning_rate": 9.093243370794291e-05,
      "loss": 1.0256,
      "step": 315
    },
    {
      "epoch": 0.003035579197636859,
      "grad_norm": 0.7042948007583618,
      "learning_rate": 9.087568947751139e-05,
      "loss": 1.567,
      "step": 316
    },
    {
      "epoch": 0.0030451854609205202,
      "grad_norm": 1.0377893447875977,
      "learning_rate": 9.081878607071996e-05,
      "loss": 1.4352,
      "step": 317
    },
    {
      "epoch": 0.0030547917242041813,
      "grad_norm": 1.0769445896148682,
      "learning_rate": 9.076172370915943e-05,
      "loss": 0.8615,
      "step": 318
    },
    {
      "epoch": 0.003064397987487842,
      "grad_norm": 1.21920907497406,
      "learning_rate": 9.07045026150396e-05,
      "loss": 0.8682,
      "step": 319
    },
    {
      "epoch": 0.003074004250771503,
      "grad_norm": 0.8914228677749634,
      "learning_rate": 9.064712301118841e-05,
      "loss": 1.304,
      "step": 320
    },
    {
      "epoch": 0.003083610514055164,
      "grad_norm": 0.7449419498443604,
      "learning_rate": 9.058958512105104e-05,
      "loss": 0.3873,
      "step": 321
    },
    {
      "epoch": 0.003093216777338825,
      "grad_norm": 0.9614444971084595,
      "learning_rate": 9.053188916868911e-05,
      "loss": 0.8443,
      "step": 322
    },
    {
      "epoch": 0.003102823040622486,
      "grad_norm": 0.7646625638008118,
      "learning_rate": 9.047403537877971e-05,
      "loss": 1.1377,
      "step": 323
    },
    {
      "epoch": 0.0031124293039061467,
      "grad_norm": 0.711182713508606,
      "learning_rate": 9.04160239766146e-05,
      "loss": 1.337,
      "step": 324
    },
    {
      "epoch": 0.0031220355671898077,
      "grad_norm": 0.9347159266471863,
      "learning_rate": 9.035785518809927e-05,
      "loss": 1.6891,
      "step": 325
    },
    {
      "epoch": 0.0031316418304734688,
      "grad_norm": 1.1050485372543335,
      "learning_rate": 9.029952923975216e-05,
      "loss": 1.1436,
      "step": 326
    },
    {
      "epoch": 0.00314124809375713,
      "grad_norm": 0.8575931191444397,
      "learning_rate": 9.024104635870368e-05,
      "loss": 0.9083,
      "step": 327
    },
    {
      "epoch": 0.0031508543570407904,
      "grad_norm": 0.9506419897079468,
      "learning_rate": 9.018240677269533e-05,
      "loss": 1.6021,
      "step": 328
    },
    {
      "epoch": 0.0031604606203244514,
      "grad_norm": 0.8634719848632812,
      "learning_rate": 9.012361071007891e-05,
      "loss": 1.1789,
      "step": 329
    },
    {
      "epoch": 0.0031700668836081125,
      "grad_norm": 1.1321772336959839,
      "learning_rate": 9.00646583998155e-05,
      "loss": 1.4741,
      "step": 330
    },
    {
      "epoch": 0.0031796731468917735,
      "grad_norm": 1.0567991733551025,
      "learning_rate": 9.000555007147469e-05,
      "loss": 0.3774,
      "step": 331
    },
    {
      "epoch": 0.0031892794101754346,
      "grad_norm": 0.9422956109046936,
      "learning_rate": 8.994628595523358e-05,
      "loss": 1.5298,
      "step": 332
    },
    {
      "epoch": 0.003198885673459095,
      "grad_norm": 1.1020911931991577,
      "learning_rate": 8.988686628187597e-05,
      "loss": 0.916,
      "step": 333
    },
    {
      "epoch": 0.0032084919367427562,
      "grad_norm": 1.001184105873108,
      "learning_rate": 8.98272912827914e-05,
      "loss": 0.7681,
      "step": 334
    },
    {
      "epoch": 0.0032180982000264173,
      "grad_norm": 1.2963169813156128,
      "learning_rate": 8.976756118997427e-05,
      "loss": 0.8801,
      "step": 335
    },
    {
      "epoch": 0.0032277044633100783,
      "grad_norm": 1.318663239479065,
      "learning_rate": 8.970767623602299e-05,
      "loss": 1.0931,
      "step": 336
    },
    {
      "epoch": 0.003237310726593739,
      "grad_norm": 0.9375176429748535,
      "learning_rate": 8.964763665413893e-05,
      "loss": 0.434,
      "step": 337
    },
    {
      "epoch": 0.0032469169898774,
      "grad_norm": 1.4789881706237793,
      "learning_rate": 8.95874426781257e-05,
      "loss": 0.8414,
      "step": 338
    },
    {
      "epoch": 0.003256523253161061,
      "grad_norm": 2.0529563426971436,
      "learning_rate": 8.952709454238808e-05,
      "loss": 1.749,
      "step": 339
    },
    {
      "epoch": 0.003266129516444722,
      "grad_norm": 1.5810272693634033,
      "learning_rate": 8.94665924819312e-05,
      "loss": 0.7591,
      "step": 340
    },
    {
      "epoch": 0.003275735779728383,
      "grad_norm": 1.382327914237976,
      "learning_rate": 8.940593673235962e-05,
      "loss": 1.3951,
      "step": 341
    },
    {
      "epoch": 0.0032853420430120437,
      "grad_norm": 1.6018394231796265,
      "learning_rate": 8.934512752987634e-05,
      "loss": 1.3542,
      "step": 342
    },
    {
      "epoch": 0.0032949483062957047,
      "grad_norm": 2.640899896621704,
      "learning_rate": 8.928416511128195e-05,
      "loss": 2.0011,
      "step": 343
    },
    {
      "epoch": 0.003304554569579366,
      "grad_norm": 2.9989922046661377,
      "learning_rate": 8.922304971397368e-05,
      "loss": 1.8047,
      "step": 344
    },
    {
      "epoch": 0.003314160832863027,
      "grad_norm": 2.638798236846924,
      "learning_rate": 8.916178157594453e-05,
      "loss": 1.9521,
      "step": 345
    },
    {
      "epoch": 0.0033237670961466874,
      "grad_norm": 2.1246392726898193,
      "learning_rate": 8.910036093578222e-05,
      "loss": 1.373,
      "step": 346
    },
    {
      "epoch": 0.0033333733594303485,
      "grad_norm": 1.781572699546814,
      "learning_rate": 8.903878803266841e-05,
      "loss": 1.2986,
      "step": 347
    },
    {
      "epoch": 0.0033429796227140095,
      "grad_norm": 2.5545308589935303,
      "learning_rate": 8.897706310637765e-05,
      "loss": 2.5102,
      "step": 348
    },
    {
      "epoch": 0.0033525858859976706,
      "grad_norm": 3.0271899700164795,
      "learning_rate": 8.891518639727649e-05,
      "loss": 2.5167,
      "step": 349
    },
    {
      "epoch": 0.0033621921492813316,
      "grad_norm": 2.5531771183013916,
      "learning_rate": 8.88531581463226e-05,
      "loss": 2.0928,
      "step": 350
    },
    {
      "epoch": 0.0033717984125649922,
      "grad_norm": 0.22428272664546967,
      "learning_rate": 8.879097859506372e-05,
      "loss": 1.9347,
      "step": 351
    },
    {
      "epoch": 0.0033814046758486533,
      "grad_norm": 0.32076403498649597,
      "learning_rate": 8.872864798563675e-05,
      "loss": 2.3704,
      "step": 352
    },
    {
      "epoch": 0.0033910109391323143,
      "grad_norm": 0.5125539302825928,
      "learning_rate": 8.866616656076696e-05,
      "loss": 1.9072,
      "step": 353
    },
    {
      "epoch": 0.0034006172024159754,
      "grad_norm": 0.3426344394683838,
      "learning_rate": 8.860353456376679e-05,
      "loss": 1.7227,
      "step": 354
    },
    {
      "epoch": 0.003410223465699636,
      "grad_norm": 0.4843062460422516,
      "learning_rate": 8.854075223853508e-05,
      "loss": 1.8762,
      "step": 355
    },
    {
      "epoch": 0.003419829728983297,
      "grad_norm": 0.6052809357643127,
      "learning_rate": 8.847781982955612e-05,
      "loss": 1.8172,
      "step": 356
    },
    {
      "epoch": 0.003429435992266958,
      "grad_norm": 0.5863645672798157,
      "learning_rate": 8.841473758189854e-05,
      "loss": 1.9266,
      "step": 357
    },
    {
      "epoch": 0.003439042255550619,
      "grad_norm": 0.5478520393371582,
      "learning_rate": 8.835150574121455e-05,
      "loss": 1.9319,
      "step": 358
    },
    {
      "epoch": 0.00344864851883428,
      "grad_norm": 0.5814157128334045,
      "learning_rate": 8.828812455373891e-05,
      "loss": 1.0567,
      "step": 359
    },
    {
      "epoch": 0.0034582547821179407,
      "grad_norm": 0.5395415425300598,
      "learning_rate": 8.82245942662879e-05,
      "loss": 1.601,
      "step": 360
    },
    {
      "epoch": 0.0034678610454016018,
      "grad_norm": 1.6547329425811768,
      "learning_rate": 8.816091512625843e-05,
      "loss": 0.4541,
      "step": 361
    },
    {
      "epoch": 0.003477467308685263,
      "grad_norm": 1.026402235031128,
      "learning_rate": 8.80970873816271e-05,
      "loss": 1.5577,
      "step": 362
    },
    {
      "epoch": 0.003487073571968924,
      "grad_norm": 1.6169495582580566,
      "learning_rate": 8.803311128094918e-05,
      "loss": 0.5946,
      "step": 363
    },
    {
      "epoch": 0.0034966798352525845,
      "grad_norm": 0.655973494052887,
      "learning_rate": 8.796898707335766e-05,
      "loss": 1.24,
      "step": 364
    },
    {
      "epoch": 0.0035062860985362455,
      "grad_norm": 0.7976295351982117,
      "learning_rate": 8.790471500856228e-05,
      "loss": 0.9109,
      "step": 365
    },
    {
      "epoch": 0.0035158923618199066,
      "grad_norm": 0.831949770450592,
      "learning_rate": 8.784029533684856e-05,
      "loss": 1.5788,
      "step": 366
    },
    {
      "epoch": 0.0035254986251035676,
      "grad_norm": 0.8801969289779663,
      "learning_rate": 8.777572830907684e-05,
      "loss": 0.9746,
      "step": 367
    },
    {
      "epoch": 0.0035351048883872286,
      "grad_norm": 0.9921168088912964,
      "learning_rate": 8.771101417668127e-05,
      "loss": 1.5528,
      "step": 368
    },
    {
      "epoch": 0.0035447111516708893,
      "grad_norm": 0.8745017051696777,
      "learning_rate": 8.764615319166886e-05,
      "loss": 1.2441,
      "step": 369
    },
    {
      "epoch": 0.0035543174149545503,
      "grad_norm": 1.3670241832733154,
      "learning_rate": 8.758114560661846e-05,
      "loss": 1.512,
      "step": 370
    },
    {
      "epoch": 0.0035639236782382113,
      "grad_norm": 0.8068578243255615,
      "learning_rate": 8.751599167467985e-05,
      "loss": 1.1984,
      "step": 371
    },
    {
      "epoch": 0.0035735299415218724,
      "grad_norm": 0.8993633985519409,
      "learning_rate": 8.745069164957264e-05,
      "loss": 0.6093,
      "step": 372
    },
    {
      "epoch": 0.003583136204805533,
      "grad_norm": 0.9271231293678284,
      "learning_rate": 8.738524578558547e-05,
      "loss": 1.1912,
      "step": 373
    },
    {
      "epoch": 0.003592742468089194,
      "grad_norm": 0.7740107774734497,
      "learning_rate": 8.731965433757475e-05,
      "loss": 0.7562,
      "step": 374
    },
    {
      "epoch": 0.003602348731372855,
      "grad_norm": 1.6577304601669312,
      "learning_rate": 8.72539175609639e-05,
      "loss": 1.2857,
      "step": 375
    },
    {
      "epoch": 0.003611954994656516,
      "grad_norm": 0.8721552491188049,
      "learning_rate": 8.718803571174227e-05,
      "loss": 0.6391,
      "step": 376
    },
    {
      "epoch": 0.003621561257940177,
      "grad_norm": 1.2786345481872559,
      "learning_rate": 8.712200904646416e-05,
      "loss": 0.7984,
      "step": 377
    },
    {
      "epoch": 0.0036311675212238378,
      "grad_norm": 1.3661487102508545,
      "learning_rate": 8.705583782224775e-05,
      "loss": 1.3862,
      "step": 378
    },
    {
      "epoch": 0.003640773784507499,
      "grad_norm": 1.1460968255996704,
      "learning_rate": 8.698952229677422e-05,
      "loss": 1.6228,
      "step": 379
    },
    {
      "epoch": 0.00365038004779116,
      "grad_norm": 1.1570000648498535,
      "learning_rate": 8.69230627282866e-05,
      "loss": 1.2329,
      "step": 380
    },
    {
      "epoch": 0.003659986311074821,
      "grad_norm": 1.7973310947418213,
      "learning_rate": 8.685645937558896e-05,
      "loss": 1.5178,
      "step": 381
    },
    {
      "epoch": 0.0036695925743584815,
      "grad_norm": 1.2523235082626343,
      "learning_rate": 8.678971249804517e-05,
      "loss": 1.3784,
      "step": 382
    },
    {
      "epoch": 0.0036791988376421426,
      "grad_norm": 1.1840691566467285,
      "learning_rate": 8.67228223555781e-05,
      "loss": 1.2744,
      "step": 383
    },
    {
      "epoch": 0.0036888051009258036,
      "grad_norm": 1.7168068885803223,
      "learning_rate": 8.665578920866844e-05,
      "loss": 1.4796,
      "step": 384
    },
    {
      "epoch": 0.0036984113642094646,
      "grad_norm": 1.25692617893219,
      "learning_rate": 8.658861331835385e-05,
      "loss": 0.7745,
      "step": 385
    },
    {
      "epoch": 0.0037080176274931257,
      "grad_norm": 1.155740737915039,
      "learning_rate": 8.652129494622775e-05,
      "loss": 0.8555,
      "step": 386
    },
    {
      "epoch": 0.0037176238907767863,
      "grad_norm": 1.2726538181304932,
      "learning_rate": 8.645383435443852e-05,
      "loss": 2.3959,
      "step": 387
    },
    {
      "epoch": 0.0037272301540604473,
      "grad_norm": 1.4162434339523315,
      "learning_rate": 8.638623180568827e-05,
      "loss": 1.1419,
      "step": 388
    },
    {
      "epoch": 0.0037368364173441084,
      "grad_norm": 1.4636874198913574,
      "learning_rate": 8.631848756323197e-05,
      "loss": 1.9391,
      "step": 389
    },
    {
      "epoch": 0.0037464426806277694,
      "grad_norm": 1.2923591136932373,
      "learning_rate": 8.625060189087634e-05,
      "loss": 0.5568,
      "step": 390
    },
    {
      "epoch": 0.0037560489439114305,
      "grad_norm": 1.1599072217941284,
      "learning_rate": 8.618257505297886e-05,
      "loss": 0.7475,
      "step": 391
    },
    {
      "epoch": 0.003765655207195091,
      "grad_norm": 1.664733648300171,
      "learning_rate": 8.611440731444672e-05,
      "loss": 0.8963,
      "step": 392
    },
    {
      "epoch": 0.003775261470478752,
      "grad_norm": 2.7576544284820557,
      "learning_rate": 8.604609894073584e-05,
      "loss": 1.9104,
      "step": 393
    },
    {
      "epoch": 0.003784867733762413,
      "grad_norm": 2.1158580780029297,
      "learning_rate": 8.597765019784971e-05,
      "loss": 1.429,
      "step": 394
    },
    {
      "epoch": 0.003794473997046074,
      "grad_norm": 2.2090001106262207,
      "learning_rate": 8.590906135233854e-05,
      "loss": 1.5083,
      "step": 395
    },
    {
      "epoch": 0.003804080260329735,
      "grad_norm": 2.040402412414551,
      "learning_rate": 8.584033267129808e-05,
      "loss": 1.6655,
      "step": 396
    },
    {
      "epoch": 0.003813686523613396,
      "grad_norm": 2.177692174911499,
      "learning_rate": 8.577146442236857e-05,
      "loss": 2.1772,
      "step": 397
    },
    {
      "epoch": 0.003823292786897057,
      "grad_norm": 2.7620651721954346,
      "learning_rate": 8.570245687373383e-05,
      "loss": 1.8523,
      "step": 398
    },
    {
      "epoch": 0.003832899050180718,
      "grad_norm": 3.0491716861724854,
      "learning_rate": 8.563331029412012e-05,
      "loss": 2.6057,
      "step": 399
    },
    {
      "epoch": 0.003842505313464379,
      "grad_norm": 3.5101897716522217,
      "learning_rate": 8.556402495279505e-05,
      "loss": 3.3236,
      "step": 400
    },
    {
      "epoch": 0.0038521115767480396,
      "grad_norm": 0.24643929302692413,
      "learning_rate": 8.549460111956664e-05,
      "loss": 1.8803,
      "step": 401
    },
    {
      "epoch": 0.0038617178400317006,
      "grad_norm": 0.26961207389831543,
      "learning_rate": 8.542503906478223e-05,
      "loss": 2.259,
      "step": 402
    },
    {
      "epoch": 0.0038713241033153617,
      "grad_norm": 0.41167938709259033,
      "learning_rate": 8.535533905932738e-05,
      "loss": 2.4769,
      "step": 403
    },
    {
      "epoch": 0.0038809303665990227,
      "grad_norm": 0.4167620539665222,
      "learning_rate": 8.528550137462486e-05,
      "loss": 1.7652,
      "step": 404
    },
    {
      "epoch": 0.0038905366298826833,
      "grad_norm": 0.3676280081272125,
      "learning_rate": 8.521552628263362e-05,
      "loss": 1.7498,
      "step": 405
    },
    {
      "epoch": 0.0039001428931663444,
      "grad_norm": 1.3182330131530762,
      "learning_rate": 8.514541405584762e-05,
      "loss": 1.185,
      "step": 406
    },
    {
      "epoch": 0.003909749156450005,
      "grad_norm": 0.5758641958236694,
      "learning_rate": 8.507516496729495e-05,
      "loss": 1.7749,
      "step": 407
    },
    {
      "epoch": 0.0039193554197336665,
      "grad_norm": 1.0270899534225464,
      "learning_rate": 8.500477929053653e-05,
      "loss": 0.9116,
      "step": 408
    },
    {
      "epoch": 0.0039289616830173275,
      "grad_norm": 0.6413470506668091,
      "learning_rate": 8.493425729966534e-05,
      "loss": 1.3638,
      "step": 409
    },
    {
      "epoch": 0.0039385679463009885,
      "grad_norm": 1.5620859861373901,
      "learning_rate": 8.486359926930501e-05,
      "loss": 0.7964,
      "step": 410
    },
    {
      "epoch": 0.00394817420958465,
      "grad_norm": 0.6751886606216431,
      "learning_rate": 8.479280547460907e-05,
      "loss": 1.7879,
      "step": 411
    },
    {
      "epoch": 0.00395778047286831,
      "grad_norm": 0.8356006741523743,
      "learning_rate": 8.472187619125967e-05,
      "loss": 2.1369,
      "step": 412
    },
    {
      "epoch": 0.003967386736151971,
      "grad_norm": 0.8070880770683289,
      "learning_rate": 8.465081169546659e-05,
      "loss": 1.4631,
      "step": 413
    },
    {
      "epoch": 0.003976992999435632,
      "grad_norm": 0.919944703578949,
      "learning_rate": 8.457961226396615e-05,
      "loss": 2.0299,
      "step": 414
    },
    {
      "epoch": 0.003986599262719293,
      "grad_norm": 0.7714675068855286,
      "learning_rate": 8.450827817402011e-05,
      "loss": 1.4708,
      "step": 415
    },
    {
      "epoch": 0.003996205526002954,
      "grad_norm": 0.8762902617454529,
      "learning_rate": 8.443680970341466e-05,
      "loss": 1.5311,
      "step": 416
    },
    {
      "epoch": 0.004005811789286615,
      "grad_norm": 0.7733710408210754,
      "learning_rate": 8.436520713045922e-05,
      "loss": 0.9334,
      "step": 417
    },
    {
      "epoch": 0.004015418052570276,
      "grad_norm": 1.0407490730285645,
      "learning_rate": 8.42934707339855e-05,
      "loss": 1.0368,
      "step": 418
    },
    {
      "epoch": 0.004025024315853937,
      "grad_norm": 0.9425524473190308,
      "learning_rate": 8.422160079334628e-05,
      "loss": 0.9595,
      "step": 419
    },
    {
      "epoch": 0.004034630579137598,
      "grad_norm": 1.014255404472351,
      "learning_rate": 8.41495975884144e-05,
      "loss": 1.632,
      "step": 420
    },
    {
      "epoch": 0.004044236842421258,
      "grad_norm": 0.9300498962402344,
      "learning_rate": 8.40774613995817e-05,
      "loss": 1.0254,
      "step": 421
    },
    {
      "epoch": 0.004053843105704919,
      "grad_norm": 0.7729508876800537,
      "learning_rate": 8.400519250775779e-05,
      "loss": 1.588,
      "step": 422
    },
    {
      "epoch": 0.00406344936898858,
      "grad_norm": 0.849949300289154,
      "learning_rate": 8.393279119436912e-05,
      "loss": 1.1218,
      "step": 423
    },
    {
      "epoch": 0.004073055632272241,
      "grad_norm": 0.8489507436752319,
      "learning_rate": 8.386025774135778e-05,
      "loss": 1.1434,
      "step": 424
    },
    {
      "epoch": 0.0040826618955559025,
      "grad_norm": 1.2974132299423218,
      "learning_rate": 8.378759243118044e-05,
      "loss": 1.4761,
      "step": 425
    },
    {
      "epoch": 0.0040922681588395635,
      "grad_norm": 1.0576504468917847,
      "learning_rate": 8.371479554680723e-05,
      "loss": 1.8823,
      "step": 426
    },
    {
      "epoch": 0.0041018744221232245,
      "grad_norm": 1.4721527099609375,
      "learning_rate": 8.364186737172068e-05,
      "loss": 0.9306,
      "step": 427
    },
    {
      "epoch": 0.004111480685406886,
      "grad_norm": 1.2131175994873047,
      "learning_rate": 8.356880818991454e-05,
      "loss": 1.2291,
      "step": 428
    },
    {
      "epoch": 0.004121086948690547,
      "grad_norm": 1.180631399154663,
      "learning_rate": 8.349561828589277e-05,
      "loss": 0.9697,
      "step": 429
    },
    {
      "epoch": 0.004130693211974207,
      "grad_norm": 1.037341833114624,
      "learning_rate": 8.342229794466834e-05,
      "loss": 1.2839,
      "step": 430
    },
    {
      "epoch": 0.004140299475257868,
      "grad_norm": 1.3163753747940063,
      "learning_rate": 8.33488474517622e-05,
      "loss": 0.6599,
      "step": 431
    },
    {
      "epoch": 0.004149905738541529,
      "grad_norm": 1.0974161624908447,
      "learning_rate": 8.327526709320212e-05,
      "loss": 0.8106,
      "step": 432
    },
    {
      "epoch": 0.00415951200182519,
      "grad_norm": 1.6536469459533691,
      "learning_rate": 8.320155715552155e-05,
      "loss": 0.3985,
      "step": 433
    },
    {
      "epoch": 0.004169118265108851,
      "grad_norm": 1.5894252061843872,
      "learning_rate": 8.312771792575863e-05,
      "loss": 1.7108,
      "step": 434
    },
    {
      "epoch": 0.004178724528392512,
      "grad_norm": 1.1886699199676514,
      "learning_rate": 8.305374969145488e-05,
      "loss": 1.3607,
      "step": 435
    },
    {
      "epoch": 0.004188330791676173,
      "grad_norm": 1.1035550832748413,
      "learning_rate": 8.297965274065425e-05,
      "loss": 1.2168,
      "step": 436
    },
    {
      "epoch": 0.004197937054959834,
      "grad_norm": 1.2974339723587036,
      "learning_rate": 8.290542736190188e-05,
      "loss": 1.0366,
      "step": 437
    },
    {
      "epoch": 0.004207543318243495,
      "grad_norm": 1.8107496500015259,
      "learning_rate": 8.283107384424312e-05,
      "loss": 1.5904,
      "step": 438
    },
    {
      "epoch": 0.004217149581527155,
      "grad_norm": 1.7278616428375244,
      "learning_rate": 8.275659247722222e-05,
      "loss": 1.7065,
      "step": 439
    },
    {
      "epoch": 0.004226755844810816,
      "grad_norm": 2.5479753017425537,
      "learning_rate": 8.268198355088133e-05,
      "loss": 1.3555,
      "step": 440
    },
    {
      "epoch": 0.004236362108094477,
      "grad_norm": 1.4928597211837769,
      "learning_rate": 8.260724735575933e-05,
      "loss": 0.692,
      "step": 441
    },
    {
      "epoch": 0.0042459683713781384,
      "grad_norm": 4.546663761138916,
      "learning_rate": 8.253238418289073e-05,
      "loss": 1.6267,
      "step": 442
    },
    {
      "epoch": 0.0042555746346617995,
      "grad_norm": 1.6526551246643066,
      "learning_rate": 8.24573943238045e-05,
      "loss": 1.4062,
      "step": 443
    },
    {
      "epoch": 0.0042651808979454605,
      "grad_norm": 2.7270216941833496,
      "learning_rate": 8.238227807052292e-05,
      "loss": 2.0834,
      "step": 444
    },
    {
      "epoch": 0.004274787161229122,
      "grad_norm": 2.437709093093872,
      "learning_rate": 8.230703571556048e-05,
      "loss": 1.3957,
      "step": 445
    },
    {
      "epoch": 0.004284393424512783,
      "grad_norm": 2.273916244506836,
      "learning_rate": 8.223166755192278e-05,
      "loss": 1.3965,
      "step": 446
    },
    {
      "epoch": 0.004293999687796444,
      "grad_norm": 3.0870871543884277,
      "learning_rate": 8.215617387310524e-05,
      "loss": 2.0259,
      "step": 447
    },
    {
      "epoch": 0.004303605951080104,
      "grad_norm": 2.2421727180480957,
      "learning_rate": 8.208055497309215e-05,
      "loss": 1.9555,
      "step": 448
    },
    {
      "epoch": 0.004313212214363765,
      "grad_norm": 4.066991806030273,
      "learning_rate": 8.200481114635536e-05,
      "loss": 2.2371,
      "step": 449
    },
    {
      "epoch": 0.004322818477647426,
      "grad_norm": 2.891526460647583,
      "learning_rate": 8.192894268785327e-05,
      "loss": 2.8821,
      "step": 450
    },
    {
      "epoch": 0.004332424740931087,
      "grad_norm": 0.2004936784505844,
      "learning_rate": 8.185294989302958e-05,
      "loss": 1.9357,
      "step": 451
    },
    {
      "epoch": 0.004342031004214748,
      "grad_norm": 0.40916886925697327,
      "learning_rate": 8.177683305781216e-05,
      "loss": 2.1703,
      "step": 452
    },
    {
      "epoch": 0.004351637267498409,
      "grad_norm": 0.7315785884857178,
      "learning_rate": 8.170059247861194e-05,
      "loss": 0.8347,
      "step": 453
    },
    {
      "epoch": 0.00436124353078207,
      "grad_norm": 0.6368415951728821,
      "learning_rate": 8.162422845232172e-05,
      "loss": 1.2616,
      "step": 454
    },
    {
      "epoch": 0.004370849794065731,
      "grad_norm": 0.5154627561569214,
      "learning_rate": 8.154774127631501e-05,
      "loss": 1.8042,
      "step": 455
    },
    {
      "epoch": 0.004380456057349392,
      "grad_norm": 0.4894868731498718,
      "learning_rate": 8.147113124844493e-05,
      "loss": 1.5353,
      "step": 456
    },
    {
      "epoch": 0.004390062320633052,
      "grad_norm": 0.5612379908561707,
      "learning_rate": 8.139439866704293e-05,
      "loss": 1.7016,
      "step": 457
    },
    {
      "epoch": 0.004399668583916713,
      "grad_norm": 0.9150093197822571,
      "learning_rate": 8.131754383091777e-05,
      "loss": 0.2741,
      "step": 458
    },
    {
      "epoch": 0.0044092748472003744,
      "grad_norm": 0.6039626598358154,
      "learning_rate": 8.124056703935423e-05,
      "loss": 2.4897,
      "step": 459
    },
    {
      "epoch": 0.0044188811104840355,
      "grad_norm": 0.791835606098175,
      "learning_rate": 8.116346859211208e-05,
      "loss": 2.4056,
      "step": 460
    },
    {
      "epoch": 0.0044284873737676965,
      "grad_norm": 1.0985621213912964,
      "learning_rate": 8.108624878942477e-05,
      "loss": 1.6009,
      "step": 461
    },
    {
      "epoch": 0.004438093637051358,
      "grad_norm": 0.8068703413009644,
      "learning_rate": 8.100890793199834e-05,
      "loss": 0.8474,
      "step": 462
    },
    {
      "epoch": 0.004447699900335019,
      "grad_norm": 0.8546838760375977,
      "learning_rate": 8.093144632101026e-05,
      "loss": 0.6256,
      "step": 463
    },
    {
      "epoch": 0.00445730616361868,
      "grad_norm": 0.94398033618927,
      "learning_rate": 8.085386425810823e-05,
      "loss": 0.8968,
      "step": 464
    },
    {
      "epoch": 0.004466912426902341,
      "grad_norm": 0.7965465188026428,
      "learning_rate": 8.077616204540897e-05,
      "loss": 0.7744,
      "step": 465
    },
    {
      "epoch": 0.004476518690186001,
      "grad_norm": 0.9627684950828552,
      "learning_rate": 8.069833998549712e-05,
      "loss": 1.4823,
      "step": 466
    },
    {
      "epoch": 0.004486124953469662,
      "grad_norm": 1.4884891510009766,
      "learning_rate": 8.062039838142402e-05,
      "loss": 1.554,
      "step": 467
    },
    {
      "epoch": 0.004495731216753323,
      "grad_norm": 0.912140965461731,
      "learning_rate": 8.054233753670653e-05,
      "loss": 0.5019,
      "step": 468
    },
    {
      "epoch": 0.004505337480036984,
      "grad_norm": 0.8282574415206909,
      "learning_rate": 8.046415775532585e-05,
      "loss": 1.7091,
      "step": 469
    },
    {
      "epoch": 0.004514943743320645,
      "grad_norm": 0.7616280317306519,
      "learning_rate": 8.038585934172632e-05,
      "loss": 1.0579,
      "step": 470
    },
    {
      "epoch": 0.004524550006604306,
      "grad_norm": 0.8210064768791199,
      "learning_rate": 8.030744260081426e-05,
      "loss": 1.0293,
      "step": 471
    },
    {
      "epoch": 0.004534156269887967,
      "grad_norm": 1.368720531463623,
      "learning_rate": 8.022890783795685e-05,
      "loss": 1.3994,
      "step": 472
    },
    {
      "epoch": 0.004543762533171628,
      "grad_norm": 0.9457901120185852,
      "learning_rate": 8.015025535898073e-05,
      "loss": 1.3559,
      "step": 473
    },
    {
      "epoch": 0.004553368796455289,
      "grad_norm": 0.9419399499893188,
      "learning_rate": 8.007148547017105e-05,
      "loss": 1.3564,
      "step": 474
    },
    {
      "epoch": 0.004562975059738949,
      "grad_norm": 1.3647840023040771,
      "learning_rate": 7.999259847827015e-05,
      "loss": 1.3006,
      "step": 475
    },
    {
      "epoch": 0.0045725813230226104,
      "grad_norm": 1.2996370792388916,
      "learning_rate": 7.991359469047633e-05,
      "loss": 1.3202,
      "step": 476
    },
    {
      "epoch": 0.0045821875863062715,
      "grad_norm": 1.1049540042877197,
      "learning_rate": 7.983447441444281e-05,
      "loss": 1.0557,
      "step": 477
    },
    {
      "epoch": 0.0045917938495899325,
      "grad_norm": 1.3034887313842773,
      "learning_rate": 7.975523795827635e-05,
      "loss": 1.2773,
      "step": 478
    },
    {
      "epoch": 0.0046014001128735936,
      "grad_norm": 1.549209475517273,
      "learning_rate": 7.967588563053616e-05,
      "loss": 1.8935,
      "step": 479
    },
    {
      "epoch": 0.004611006376157255,
      "grad_norm": 1.4847763776779175,
      "learning_rate": 7.95964177402327e-05,
      "loss": 1.1378,
      "step": 480
    },
    {
      "epoch": 0.004620612639440916,
      "grad_norm": 1.0118752717971802,
      "learning_rate": 7.951683459682641e-05,
      "loss": 1.1059,
      "step": 481
    },
    {
      "epoch": 0.004630218902724577,
      "grad_norm": 1.3798691034317017,
      "learning_rate": 7.943713651022659e-05,
      "loss": 1.4514,
      "step": 482
    },
    {
      "epoch": 0.004639825166008238,
      "grad_norm": 1.4011057615280151,
      "learning_rate": 7.935732379079008e-05,
      "loss": 0.5678,
      "step": 483
    },
    {
      "epoch": 0.004649431429291898,
      "grad_norm": 1.6054455041885376,
      "learning_rate": 7.927739674932017e-05,
      "loss": 1.8404,
      "step": 484
    },
    {
      "epoch": 0.004659037692575559,
      "grad_norm": 2.3447251319885254,
      "learning_rate": 7.919735569706533e-05,
      "loss": 2.021,
      "step": 485
    },
    {
      "epoch": 0.00466864395585922,
      "grad_norm": 1.919410228729248,
      "learning_rate": 7.911720094571796e-05,
      "loss": 1.3371,
      "step": 486
    },
    {
      "epoch": 0.004678250219142881,
      "grad_norm": 1.3991084098815918,
      "learning_rate": 7.903693280741331e-05,
      "loss": 0.7526,
      "step": 487
    },
    {
      "epoch": 0.004687856482426542,
      "grad_norm": 1.5650057792663574,
      "learning_rate": 7.895655159472811e-05,
      "loss": 1.2052,
      "step": 488
    },
    {
      "epoch": 0.004697462745710203,
      "grad_norm": 1.887589931488037,
      "learning_rate": 7.887605762067945e-05,
      "loss": 1.7086,
      "step": 489
    },
    {
      "epoch": 0.004707069008993864,
      "grad_norm": 1.8097609281539917,
      "learning_rate": 7.879545119872346e-05,
      "loss": 1.3661,
      "step": 490
    },
    {
      "epoch": 0.004716675272277525,
      "grad_norm": 2.0076053142547607,
      "learning_rate": 7.871473264275429e-05,
      "loss": 2.4296,
      "step": 491
    },
    {
      "epoch": 0.004726281535561186,
      "grad_norm": 1.6763304471969604,
      "learning_rate": 7.863390226710266e-05,
      "loss": 1.856,
      "step": 492
    },
    {
      "epoch": 0.004735887798844846,
      "grad_norm": 2.710705041885376,
      "learning_rate": 7.855296038653475e-05,
      "loss": 1.4541,
      "step": 493
    },
    {
      "epoch": 0.0047454940621285075,
      "grad_norm": 2.2385811805725098,
      "learning_rate": 7.847190731625095e-05,
      "loss": 2.1358,
      "step": 494
    },
    {
      "epoch": 0.0047551003254121685,
      "grad_norm": 1.8508498668670654,
      "learning_rate": 7.83907433718847e-05,
      "loss": 1.0959,
      "step": 495
    },
    {
      "epoch": 0.0047647065886958296,
      "grad_norm": 2.1279263496398926,
      "learning_rate": 7.830946886950113e-05,
      "loss": 1.6881,
      "step": 496
    },
    {
      "epoch": 0.004774312851979491,
      "grad_norm": 2.9290308952331543,
      "learning_rate": 7.82280841255959e-05,
      "loss": 2.0353,
      "step": 497
    },
    {
      "epoch": 0.004783919115263152,
      "grad_norm": 2.8512561321258545,
      "learning_rate": 7.814658945709405e-05,
      "loss": 2.6172,
      "step": 498
    },
    {
      "epoch": 0.004793525378546813,
      "grad_norm": 4.732594013214111,
      "learning_rate": 7.80649851813486e-05,
      "loss": 2.1789,
      "step": 499
    },
    {
      "epoch": 0.004803131641830474,
      "grad_norm": 3.0503485202789307,
      "learning_rate": 7.798327161613943e-05,
      "loss": 2.2864,
      "step": 500
    },
    {
      "epoch": 0.004812737905114135,
      "grad_norm": 0.2729131579399109,
      "learning_rate": 7.790144907967201e-05,
      "loss": 1.7868,
      "step": 501
    },
    {
      "epoch": 0.004822344168397795,
      "grad_norm": 0.4615051746368408,
      "learning_rate": 7.781951789057615e-05,
      "loss": 1.7581,
      "step": 502
    },
    {
      "epoch": 0.004831950431681456,
      "grad_norm": 0.3877827525138855,
      "learning_rate": 7.773747836790481e-05,
      "loss": 1.9096,
      "step": 503
    },
    {
      "epoch": 0.004841556694965117,
      "grad_norm": 0.4477629065513611,
      "learning_rate": 7.765533083113274e-05,
      "loss": 1.8809,
      "step": 504
    },
    {
      "epoch": 0.004851162958248778,
      "grad_norm": 0.5718364715576172,
      "learning_rate": 7.757307560015538e-05,
      "loss": 1.417,
      "step": 505
    },
    {
      "epoch": 0.004860769221532439,
      "grad_norm": 0.3858019709587097,
      "learning_rate": 7.749071299528754e-05,
      "loss": 1.7606,
      "step": 506
    },
    {
      "epoch": 0.0048703754848161,
      "grad_norm": 0.6225813627243042,
      "learning_rate": 7.740824333726213e-05,
      "loss": 2.3465,
      "step": 507
    },
    {
      "epoch": 0.004879981748099761,
      "grad_norm": 0.742667555809021,
      "learning_rate": 7.732566694722898e-05,
      "loss": 0.1085,
      "step": 508
    },
    {
      "epoch": 0.004889588011383422,
      "grad_norm": 0.9599735736846924,
      "learning_rate": 7.724298414675353e-05,
      "loss": 1.1597,
      "step": 509
    },
    {
      "epoch": 0.004899194274667083,
      "grad_norm": 0.5957375764846802,
      "learning_rate": 7.71601952578156e-05,
      "loss": 1.71,
      "step": 510
    },
    {
      "epoch": 0.0049088005379507435,
      "grad_norm": 0.8335545063018799,
      "learning_rate": 7.707730060280812e-05,
      "loss": 0.68,
      "step": 511
    },
    {
      "epoch": 0.0049184068012344045,
      "grad_norm": 0.6904851198196411,
      "learning_rate": 7.699430050453592e-05,
      "loss": 1.7136,
      "step": 512
    },
    {
      "epoch": 0.0049280130645180656,
      "grad_norm": 0.9278645515441895,
      "learning_rate": 7.691119528621444e-05,
      "loss": 1.6292,
      "step": 513
    },
    {
      "epoch": 0.004937619327801727,
      "grad_norm": 1.2602430582046509,
      "learning_rate": 7.682798527146844e-05,
      "loss": 1.576,
      "step": 514
    },
    {
      "epoch": 0.004947225591085388,
      "grad_norm": 1.5897618532180786,
      "learning_rate": 7.674467078433081e-05,
      "loss": 1.0402,
      "step": 515
    },
    {
      "epoch": 0.004956831854369049,
      "grad_norm": 1.2256033420562744,
      "learning_rate": 7.666125214924127e-05,
      "loss": 2.037,
      "step": 516
    },
    {
      "epoch": 0.00496643811765271,
      "grad_norm": 0.9116320013999939,
      "learning_rate": 7.657772969104508e-05,
      "loss": 1.0748,
      "step": 517
    },
    {
      "epoch": 0.004976044380936371,
      "grad_norm": 0.7539246082305908,
      "learning_rate": 7.649410373499183e-05,
      "loss": 0.9674,
      "step": 518
    },
    {
      "epoch": 0.004985650644220032,
      "grad_norm": 1.2198669910430908,
      "learning_rate": 7.641037460673412e-05,
      "loss": 1.3576,
      "step": 519
    },
    {
      "epoch": 0.004995256907503692,
      "grad_norm": 0.9954208731651306,
      "learning_rate": 7.632654263232635e-05,
      "loss": 1.102,
      "step": 520
    },
    {
      "epoch": 0.005004863170787353,
      "grad_norm": 0.9288671612739563,
      "learning_rate": 7.624260813822342e-05,
      "loss": 0.9487,
      "step": 521
    },
    {
      "epoch": 0.005014469434071014,
      "grad_norm": 0.9206609129905701,
      "learning_rate": 7.615857145127943e-05,
      "loss": 1.1272,
      "step": 522
    },
    {
      "epoch": 0.005024075697354675,
      "grad_norm": 1.1357004642486572,
      "learning_rate": 7.607443289874642e-05,
      "loss": 1.1946,
      "step": 523
    },
    {
      "epoch": 0.005033681960638336,
      "grad_norm": 0.7786014676094055,
      "learning_rate": 7.59901928082732e-05,
      "loss": 1.4639,
      "step": 524
    },
    {
      "epoch": 0.005043288223921997,
      "grad_norm": 1.064112663269043,
      "learning_rate": 7.590585150790389e-05,
      "loss": 0.9497,
      "step": 525
    },
    {
      "epoch": 0.005052894487205658,
      "grad_norm": 0.8393730521202087,
      "learning_rate": 7.582140932607674e-05,
      "loss": 0.856,
      "step": 526
    },
    {
      "epoch": 0.005062500750489319,
      "grad_norm": 0.9430371522903442,
      "learning_rate": 7.573686659162293e-05,
      "loss": 1.3609,
      "step": 527
    },
    {
      "epoch": 0.00507210701377298,
      "grad_norm": 1.0690443515777588,
      "learning_rate": 7.565222363376514e-05,
      "loss": 1.1736,
      "step": 528
    },
    {
      "epoch": 0.0050817132770566405,
      "grad_norm": 1.1118578910827637,
      "learning_rate": 7.556748078211635e-05,
      "loss": 0.7785,
      "step": 529
    },
    {
      "epoch": 0.0050913195403403015,
      "grad_norm": 1.1453135013580322,
      "learning_rate": 7.548263836667852e-05,
      "loss": 0.9409,
      "step": 530
    },
    {
      "epoch": 0.005100925803623963,
      "grad_norm": 1.0411078929901123,
      "learning_rate": 7.53976967178414e-05,
      "loss": 0.6912,
      "step": 531
    },
    {
      "epoch": 0.005110532066907624,
      "grad_norm": 1.48271644115448,
      "learning_rate": 7.531265616638109e-05,
      "loss": 0.9797,
      "step": 532
    },
    {
      "epoch": 0.005120138330191285,
      "grad_norm": 1.2526296377182007,
      "learning_rate": 7.522751704345887e-05,
      "loss": 1.3984,
      "step": 533
    },
    {
      "epoch": 0.005129744593474946,
      "grad_norm": 1.0007219314575195,
      "learning_rate": 7.514227968061988e-05,
      "loss": 1.2419,
      "step": 534
    },
    {
      "epoch": 0.005139350856758607,
      "grad_norm": 0.9133110046386719,
      "learning_rate": 7.505694440979178e-05,
      "loss": 1.3175,
      "step": 535
    },
    {
      "epoch": 0.005148957120042268,
      "grad_norm": 1.1066343784332275,
      "learning_rate": 7.497151156328358e-05,
      "loss": 1.5721,
      "step": 536
    },
    {
      "epoch": 0.005158563383325929,
      "grad_norm": 1.1402653455734253,
      "learning_rate": 7.488598147378416e-05,
      "loss": 0.5471,
      "step": 537
    },
    {
      "epoch": 0.005168169646609589,
      "grad_norm": 1.7259846925735474,
      "learning_rate": 7.480035447436116e-05,
      "loss": 1.2847,
      "step": 538
    },
    {
      "epoch": 0.00517777590989325,
      "grad_norm": 2.174537420272827,
      "learning_rate": 7.471463089845956e-05,
      "loss": 1.4439,
      "step": 539
    },
    {
      "epoch": 0.005187382173176911,
      "grad_norm": 1.8676972389221191,
      "learning_rate": 7.462881107990044e-05,
      "loss": 1.9367,
      "step": 540
    },
    {
      "epoch": 0.005196988436460572,
      "grad_norm": 1.848473310470581,
      "learning_rate": 7.454289535287968e-05,
      "loss": 1.6324,
      "step": 541
    },
    {
      "epoch": 0.005206594699744233,
      "grad_norm": 2.1099581718444824,
      "learning_rate": 7.445688405196662e-05,
      "loss": 1.2173,
      "step": 542
    },
    {
      "epoch": 0.005216200963027894,
      "grad_norm": 2.3113853931427,
      "learning_rate": 7.437077751210279e-05,
      "loss": 1.9867,
      "step": 543
    },
    {
      "epoch": 0.005225807226311555,
      "grad_norm": 2.1420352458953857,
      "learning_rate": 7.428457606860057e-05,
      "loss": 1.977,
      "step": 544
    },
    {
      "epoch": 0.005235413489595216,
      "grad_norm": 1.932133436203003,
      "learning_rate": 7.419828005714194e-05,
      "loss": 1.0403,
      "step": 545
    },
    {
      "epoch": 0.005245019752878877,
      "grad_norm": 2.0039331912994385,
      "learning_rate": 7.411188981377714e-05,
      "loss": 1.8694,
      "step": 546
    },
    {
      "epoch": 0.0052546260161625375,
      "grad_norm": 1.9787790775299072,
      "learning_rate": 7.402540567492337e-05,
      "loss": 1.5328,
      "step": 547
    },
    {
      "epoch": 0.005264232279446199,
      "grad_norm": 2.918614625930786,
      "learning_rate": 7.393882797736342e-05,
      "loss": 2.3182,
      "step": 548
    },
    {
      "epoch": 0.00527383854272986,
      "grad_norm": 3.125279664993286,
      "learning_rate": 7.385215705824449e-05,
      "loss": 2.398,
      "step": 549
    },
    {
      "epoch": 0.005283444806013521,
      "grad_norm": 2.4191298484802246,
      "learning_rate": 7.376539325507673e-05,
      "loss": 2.9557,
      "step": 550
    },
    {
      "epoch": 0.005293051069297182,
      "grad_norm": 0.23384812474250793,
      "learning_rate": 7.367853690573208e-05,
      "loss": 2.015,
      "step": 551
    },
    {
      "epoch": 0.005302657332580843,
      "grad_norm": 0.2290044128894806,
      "learning_rate": 7.359158834844277e-05,
      "loss": 2.3179,
      "step": 552
    },
    {
      "epoch": 0.005312263595864504,
      "grad_norm": 0.33055487275123596,
      "learning_rate": 7.350454792180016e-05,
      "loss": 1.8315,
      "step": 553
    },
    {
      "epoch": 0.005321869859148165,
      "grad_norm": 0.3259883522987366,
      "learning_rate": 7.341741596475336e-05,
      "loss": 1.892,
      "step": 554
    },
    {
      "epoch": 0.005331476122431826,
      "grad_norm": 1.1160476207733154,
      "learning_rate": 7.333019281660789e-05,
      "loss": 1.1411,
      "step": 555
    },
    {
      "epoch": 0.005341082385715487,
      "grad_norm": 0.5873980522155762,
      "learning_rate": 7.324287881702443e-05,
      "loss": 1.5318,
      "step": 556
    },
    {
      "epoch": 0.005350688648999147,
      "grad_norm": 0.776999831199646,
      "learning_rate": 7.31554743060174e-05,
      "loss": 2.1125,
      "step": 557
    },
    {
      "epoch": 0.005360294912282808,
      "grad_norm": 0.5524105429649353,
      "learning_rate": 7.30679796239537e-05,
      "loss": 1.4147,
      "step": 558
    },
    {
      "epoch": 0.005369901175566469,
      "grad_norm": 0.7826858162879944,
      "learning_rate": 7.298039511155138e-05,
      "loss": 0.1666,
      "step": 559
    },
    {
      "epoch": 0.00537950743885013,
      "grad_norm": 1.0273869037628174,
      "learning_rate": 7.28927211098783e-05,
      "loss": 1.4367,
      "step": 560
    },
    {
      "epoch": 0.005389113702133791,
      "grad_norm": 0.5444875359535217,
      "learning_rate": 7.280495796035079e-05,
      "loss": 1.266,
      "step": 561
    },
    {
      "epoch": 0.005398719965417452,
      "grad_norm": 0.9350177049636841,
      "learning_rate": 7.271710600473239e-05,
      "loss": 0.685,
      "step": 562
    },
    {
      "epoch": 0.005408326228701113,
      "grad_norm": 0.7619469165802002,
      "learning_rate": 7.262916558513237e-05,
      "loss": 2.428,
      "step": 563
    },
    {
      "epoch": 0.005417932491984774,
      "grad_norm": 0.9650070667266846,
      "learning_rate": 7.25411370440046e-05,
      "loss": 2.0582,
      "step": 564
    },
    {
      "epoch": 0.0054275387552684354,
      "grad_norm": 0.720902144908905,
      "learning_rate": 7.245302072414601e-05,
      "loss": 0.6057,
      "step": 565
    },
    {
      "epoch": 0.005437145018552096,
      "grad_norm": 0.7611209154129028,
      "learning_rate": 7.236481696869543e-05,
      "loss": 0.7498,
      "step": 566
    },
    {
      "epoch": 0.005446751281835757,
      "grad_norm": 1.0854814052581787,
      "learning_rate": 7.227652612113213e-05,
      "loss": 1.056,
      "step": 567
    },
    {
      "epoch": 0.005456357545119418,
      "grad_norm": 0.9895567893981934,
      "learning_rate": 7.218814852527455e-05,
      "loss": 1.4203,
      "step": 568
    },
    {
      "epoch": 0.005465963808403079,
      "grad_norm": 1.0300164222717285,
      "learning_rate": 7.209968452527896e-05,
      "loss": 1.5143,
      "step": 569
    },
    {
      "epoch": 0.00547557007168674,
      "grad_norm": 1.4882161617279053,
      "learning_rate": 7.201113446563806e-05,
      "loss": 1.4328,
      "step": 570
    },
    {
      "epoch": 0.005485176334970401,
      "grad_norm": 0.6258545517921448,
      "learning_rate": 7.192249869117971e-05,
      "loss": 0.3928,
      "step": 571
    },
    {
      "epoch": 0.005494782598254062,
      "grad_norm": 1.8529062271118164,
      "learning_rate": 7.183377754706555e-05,
      "loss": 1.8288,
      "step": 572
    },
    {
      "epoch": 0.005504388861537723,
      "grad_norm": 0.8300051093101501,
      "learning_rate": 7.174497137878966e-05,
      "loss": 1.4176,
      "step": 573
    },
    {
      "epoch": 0.005513995124821384,
      "grad_norm": 1.2373651266098022,
      "learning_rate": 7.16560805321772e-05,
      "loss": 0.728,
      "step": 574
    },
    {
      "epoch": 0.005523601388105044,
      "grad_norm": 0.97736656665802,
      "learning_rate": 7.156710535338312e-05,
      "loss": 0.8017,
      "step": 575
    },
    {
      "epoch": 0.005533207651388705,
      "grad_norm": 1.2627793550491333,
      "learning_rate": 7.147804618889076e-05,
      "loss": 1.1722,
      "step": 576
    },
    {
      "epoch": 0.005542813914672366,
      "grad_norm": 0.9401075839996338,
      "learning_rate": 7.138890338551048e-05,
      "loss": 0.8307,
      "step": 577
    },
    {
      "epoch": 0.005552420177956027,
      "grad_norm": 0.9534733891487122,
      "learning_rate": 7.12996772903784e-05,
      "loss": 1.1464,
      "step": 578
    },
    {
      "epoch": 0.005562026441239688,
      "grad_norm": 1.3306607007980347,
      "learning_rate": 7.121036825095492e-05,
      "loss": 1.5269,
      "step": 579
    },
    {
      "epoch": 0.005571632704523349,
      "grad_norm": 1.5687206983566284,
      "learning_rate": 7.11209766150235e-05,
      "loss": 0.9284,
      "step": 580
    },
    {
      "epoch": 0.00558123896780701,
      "grad_norm": 1.4719488620758057,
      "learning_rate": 7.103150273068921e-05,
      "loss": 1.5219,
      "step": 581
    },
    {
      "epoch": 0.0055908452310906714,
      "grad_norm": 1.3736610412597656,
      "learning_rate": 7.094194694637744e-05,
      "loss": 1.1801,
      "step": 582
    },
    {
      "epoch": 0.0056004514943743325,
      "grad_norm": 1.2721003293991089,
      "learning_rate": 7.085230961083249e-05,
      "loss": 1.2748,
      "step": 583
    },
    {
      "epoch": 0.005610057757657993,
      "grad_norm": 1.2059344053268433,
      "learning_rate": 7.076259107311622e-05,
      "loss": 0.5336,
      "step": 584
    },
    {
      "epoch": 0.005619664020941654,
      "grad_norm": 1.2052173614501953,
      "learning_rate": 7.067279168260671e-05,
      "loss": 0.6967,
      "step": 585
    },
    {
      "epoch": 0.005629270284225315,
      "grad_norm": 1.636717677116394,
      "learning_rate": 7.058291178899692e-05,
      "loss": 1.081,
      "step": 586
    },
    {
      "epoch": 0.005638876547508976,
      "grad_norm": 1.1783536672592163,
      "learning_rate": 7.04929517422933e-05,
      "loss": 0.9709,
      "step": 587
    },
    {
      "epoch": 0.005648482810792637,
      "grad_norm": 1.521591067314148,
      "learning_rate": 7.040291189281435e-05,
      "loss": 1.0148,
      "step": 588
    },
    {
      "epoch": 0.005658089074076298,
      "grad_norm": 1.2916276454925537,
      "learning_rate": 7.031279259118946e-05,
      "loss": 0.7099,
      "step": 589
    },
    {
      "epoch": 0.005667695337359959,
      "grad_norm": 1.2751636505126953,
      "learning_rate": 7.022259418835731e-05,
      "loss": 1.3847,
      "step": 590
    },
    {
      "epoch": 0.00567730160064362,
      "grad_norm": 1.5614402294158936,
      "learning_rate": 7.013231703556471e-05,
      "loss": 1.1997,
      "step": 591
    },
    {
      "epoch": 0.005686907863927281,
      "grad_norm": 1.6436371803283691,
      "learning_rate": 7.004196148436506e-05,
      "loss": 1.4766,
      "step": 592
    },
    {
      "epoch": 0.005696514127210941,
      "grad_norm": 2.1577930450439453,
      "learning_rate": 6.995152788661705e-05,
      "loss": 1.0113,
      "step": 593
    },
    {
      "epoch": 0.005706120390494602,
      "grad_norm": 2.4957237243652344,
      "learning_rate": 6.986101659448337e-05,
      "loss": 1.9189,
      "step": 594
    },
    {
      "epoch": 0.005715726653778263,
      "grad_norm": 3.0496058464050293,
      "learning_rate": 6.977042796042917e-05,
      "loss": 1.0358,
      "step": 595
    },
    {
      "epoch": 0.005725332917061924,
      "grad_norm": 3.5942752361297607,
      "learning_rate": 6.967976233722089e-05,
      "loss": 1.9449,
      "step": 596
    },
    {
      "epoch": 0.005734939180345585,
      "grad_norm": 2.297795534133911,
      "learning_rate": 6.958902007792466e-05,
      "loss": 2.2546,
      "step": 597
    },
    {
      "epoch": 0.005744545443629246,
      "grad_norm": 2.280122756958008,
      "learning_rate": 6.949820153590514e-05,
      "loss": 2.4985,
      "step": 598
    },
    {
      "epoch": 0.0057541517069129074,
      "grad_norm": 2.695183753967285,
      "learning_rate": 6.940730706482399e-05,
      "loss": 2.0181,
      "step": 599
    },
    {
      "epoch": 0.0057637579701965685,
      "grad_norm": 3.334068536758423,
      "learning_rate": 6.931633701863857e-05,
      "loss": 2.9684,
      "step": 600
    },
    {
      "epoch": 0.0057733642334802295,
      "grad_norm": 0.24161791801452637,
      "learning_rate": 6.922529175160054e-05,
      "loss": 1.8707,
      "step": 601
    },
    {
      "epoch": 0.00578297049676389,
      "grad_norm": 0.39161163568496704,
      "learning_rate": 6.91341716182545e-05,
      "loss": 1.614,
      "step": 602
    },
    {
      "epoch": 0.005792576760047551,
      "grad_norm": 0.8574214577674866,
      "learning_rate": 6.904297697343655e-05,
      "loss": 1.6897,
      "step": 603
    },
    {
      "epoch": 0.005802183023331212,
      "grad_norm": 0.47848549485206604,
      "learning_rate": 6.895170817227295e-05,
      "loss": 1.6673,
      "step": 604
    },
    {
      "epoch": 0.005811789286614873,
      "grad_norm": 0.5340243577957153,
      "learning_rate": 6.886036557017881e-05,
      "loss": 2.0483,
      "step": 605
    },
    {
      "epoch": 0.005821395549898534,
      "grad_norm": 0.8916251063346863,
      "learning_rate": 6.876894952285654e-05,
      "loss": 0.9101,
      "step": 606
    },
    {
      "epoch": 0.005831001813182195,
      "grad_norm": 0.6121930480003357,
      "learning_rate": 6.867746038629462e-05,
      "loss": 1.6485,
      "step": 607
    },
    {
      "epoch": 0.005840608076465856,
      "grad_norm": 0.725107729434967,
      "learning_rate": 6.858589851676611e-05,
      "loss": 2.0477,
      "step": 608
    },
    {
      "epoch": 0.005850214339749517,
      "grad_norm": 0.7420092821121216,
      "learning_rate": 6.849426427082735e-05,
      "loss": 1.5726,
      "step": 609
    },
    {
      "epoch": 0.005859820603033178,
      "grad_norm": 0.9063351154327393,
      "learning_rate": 6.840255800531647e-05,
      "loss": 1.0145,
      "step": 610
    },
    {
      "epoch": 0.005869426866316838,
      "grad_norm": 1.0234450101852417,
      "learning_rate": 6.83107800773521e-05,
      "loss": 1.0682,
      "step": 611
    },
    {
      "epoch": 0.005879033129600499,
      "grad_norm": 1.0494650602340698,
      "learning_rate": 6.82189308443319e-05,
      "loss": 0.8531,
      "step": 612
    },
    {
      "epoch": 0.00588863939288416,
      "grad_norm": 0.9903169274330139,
      "learning_rate": 6.812701066393124e-05,
      "loss": 0.9627,
      "step": 613
    },
    {
      "epoch": 0.005898245656167821,
      "grad_norm": 1.1083897352218628,
      "learning_rate": 6.803501989410174e-05,
      "loss": 1.8463,
      "step": 614
    },
    {
      "epoch": 0.005907851919451482,
      "grad_norm": 0.6791064739227295,
      "learning_rate": 6.79429588930699e-05,
      "loss": 0.5716,
      "step": 615
    },
    {
      "epoch": 0.005917458182735143,
      "grad_norm": 0.8364579081535339,
      "learning_rate": 6.785082801933576e-05,
      "loss": 0.6466,
      "step": 616
    },
    {
      "epoch": 0.0059270644460188045,
      "grad_norm": 1.5560290813446045,
      "learning_rate": 6.775862763167142e-05,
      "loss": 1.5328,
      "step": 617
    },
    {
      "epoch": 0.0059366707093024655,
      "grad_norm": 1.0115303993225098,
      "learning_rate": 6.766635808911968e-05,
      "loss": 1.3922,
      "step": 618
    },
    {
      "epoch": 0.0059462769725861266,
      "grad_norm": 0.9655584692955017,
      "learning_rate": 6.757401975099262e-05,
      "loss": 1.8814,
      "step": 619
    },
    {
      "epoch": 0.005955883235869787,
      "grad_norm": 0.7838444113731384,
      "learning_rate": 6.748161297687027e-05,
      "loss": 0.8698,
      "step": 620
    },
    {
      "epoch": 0.005965489499153448,
      "grad_norm": 1.0263564586639404,
      "learning_rate": 6.738913812659912e-05,
      "loss": 0.7311,
      "step": 621
    },
    {
      "epoch": 0.005975095762437109,
      "grad_norm": 0.9038912653923035,
      "learning_rate": 6.729659556029078e-05,
      "loss": 0.8694,
      "step": 622
    },
    {
      "epoch": 0.00598470202572077,
      "grad_norm": 0.9689344167709351,
      "learning_rate": 6.720398563832055e-05,
      "loss": 1.413,
      "step": 623
    },
    {
      "epoch": 0.005994308289004431,
      "grad_norm": 0.8128175139427185,
      "learning_rate": 6.711130872132601e-05,
      "loss": 0.7708,
      "step": 624
    },
    {
      "epoch": 0.006003914552288092,
      "grad_norm": 0.7718427777290344,
      "learning_rate": 6.701856517020565e-05,
      "loss": 0.7669,
      "step": 625
    },
    {
      "epoch": 0.006013520815571753,
      "grad_norm": 1.487931251525879,
      "learning_rate": 6.692575534611744e-05,
      "loss": 1.4143,
      "step": 626
    },
    {
      "epoch": 0.006023127078855414,
      "grad_norm": 1.1424639225006104,
      "learning_rate": 6.683287961047742e-05,
      "loss": 0.6997,
      "step": 627
    },
    {
      "epoch": 0.006032733342139075,
      "grad_norm": 1.2063740491867065,
      "learning_rate": 6.67399383249583e-05,
      "loss": 0.8804,
      "step": 628
    },
    {
      "epoch": 0.006042339605422735,
      "grad_norm": 2.2495484352111816,
      "learning_rate": 6.664693185148807e-05,
      "loss": 0.7845,
      "step": 629
    },
    {
      "epoch": 0.006051945868706396,
      "grad_norm": 1.5282721519470215,
      "learning_rate": 6.655386055224853e-05,
      "loss": 0.8536,
      "step": 630
    },
    {
      "epoch": 0.006061552131990057,
      "grad_norm": 1.1136977672576904,
      "learning_rate": 6.646072478967397e-05,
      "loss": 0.7699,
      "step": 631
    },
    {
      "epoch": 0.006071158395273718,
      "grad_norm": 1.1592923402786255,
      "learning_rate": 6.636752492644969e-05,
      "loss": 1.0147,
      "step": 632
    },
    {
      "epoch": 0.006080764658557379,
      "grad_norm": 1.208528995513916,
      "learning_rate": 6.627426132551058e-05,
      "loss": 0.984,
      "step": 633
    },
    {
      "epoch": 0.0060903709218410405,
      "grad_norm": 2.10524845123291,
      "learning_rate": 6.618093435003981e-05,
      "loss": 1.0386,
      "step": 634
    },
    {
      "epoch": 0.0060999771851247015,
      "grad_norm": 0.8809115290641785,
      "learning_rate": 6.608754436346725e-05,
      "loss": 0.7055,
      "step": 635
    },
    {
      "epoch": 0.0061095834484083626,
      "grad_norm": 1.5567530393600464,
      "learning_rate": 6.59940917294682e-05,
      "loss": 1.3524,
      "step": 636
    },
    {
      "epoch": 0.006119189711692024,
      "grad_norm": 1.2569174766540527,
      "learning_rate": 6.590057681196191e-05,
      "loss": 0.749,
      "step": 637
    },
    {
      "epoch": 0.006128795974975684,
      "grad_norm": 1.2567017078399658,
      "learning_rate": 6.580699997511018e-05,
      "loss": 0.8428,
      "step": 638
    },
    {
      "epoch": 0.006138402238259345,
      "grad_norm": 2.0362513065338135,
      "learning_rate": 6.571336158331589e-05,
      "loss": 0.7708,
      "step": 639
    },
    {
      "epoch": 0.006148008501543006,
      "grad_norm": 1.9483751058578491,
      "learning_rate": 6.561966200122169e-05,
      "loss": 0.9227,
      "step": 640
    },
    {
      "epoch": 0.006157614764826667,
      "grad_norm": 1.3305878639221191,
      "learning_rate": 6.552590159370844e-05,
      "loss": 1.7061,
      "step": 641
    },
    {
      "epoch": 0.006167221028110328,
      "grad_norm": 1.4485077857971191,
      "learning_rate": 6.543208072589395e-05,
      "loss": 2.8786,
      "step": 642
    },
    {
      "epoch": 0.006176827291393989,
      "grad_norm": 1.8780678510665894,
      "learning_rate": 6.53381997631314e-05,
      "loss": 0.9801,
      "step": 643
    },
    {
      "epoch": 0.00618643355467765,
      "grad_norm": 2.370722532272339,
      "learning_rate": 6.524425907100803e-05,
      "loss": 1.2922,
      "step": 644
    },
    {
      "epoch": 0.006196039817961311,
      "grad_norm": 3.2226974964141846,
      "learning_rate": 6.515025901534364e-05,
      "loss": 2.2276,
      "step": 645
    },
    {
      "epoch": 0.006205646081244972,
      "grad_norm": 2.434577703475952,
      "learning_rate": 6.505619996218923e-05,
      "loss": 2.2825,
      "step": 646
    },
    {
      "epoch": 0.006215252344528632,
      "grad_norm": 3.01188588142395,
      "learning_rate": 6.496208227782556e-05,
      "loss": 1.3767,
      "step": 647
    },
    {
      "epoch": 0.006224858607812293,
      "grad_norm": 3.6666319370269775,
      "learning_rate": 6.486790632876168e-05,
      "loss": 1.7043,
      "step": 648
    },
    {
      "epoch": 0.006234464871095954,
      "grad_norm": 5.564884662628174,
      "learning_rate": 6.477367248173352e-05,
      "loss": 2.0076,
      "step": 649
    },
    {
      "epoch": 0.006244071134379615,
      "grad_norm": 2.5864717960357666,
      "learning_rate": 6.467938110370249e-05,
      "loss": 3.1291,
      "step": 650
    },
    {
      "epoch": 0.0062536773976632765,
      "grad_norm": 0.21204684674739838,
      "learning_rate": 6.458503256185404e-05,
      "loss": 1.8723,
      "step": 651
    },
    {
      "epoch": 0.0062632836609469375,
      "grad_norm": 0.3201037645339966,
      "learning_rate": 6.449062722359625e-05,
      "loss": 2.0093,
      "step": 652
    },
    {
      "epoch": 0.0062728899242305985,
      "grad_norm": 0.5894408226013184,
      "learning_rate": 6.439616545655834e-05,
      "loss": 0.8569,
      "step": 653
    },
    {
      "epoch": 0.00628249618751426,
      "grad_norm": 0.45742398500442505,
      "learning_rate": 6.430164762858925e-05,
      "loss": 1.9254,
      "step": 654
    },
    {
      "epoch": 0.006292102450797921,
      "grad_norm": 0.6730548143386841,
      "learning_rate": 6.420707410775626e-05,
      "loss": 1.9982,
      "step": 655
    },
    {
      "epoch": 0.006301708714081581,
      "grad_norm": 0.6552911996841431,
      "learning_rate": 6.411244526234356e-05,
      "loss": 1.3847,
      "step": 656
    },
    {
      "epoch": 0.006311314977365242,
      "grad_norm": 0.8640971183776855,
      "learning_rate": 6.401776146085072e-05,
      "loss": 1.2179,
      "step": 657
    },
    {
      "epoch": 0.006320921240648903,
      "grad_norm": 0.658015787601471,
      "learning_rate": 6.392302307199137e-05,
      "loss": 2.4458,
      "step": 658
    },
    {
      "epoch": 0.006330527503932564,
      "grad_norm": 0.6875593066215515,
      "learning_rate": 6.382823046469167e-05,
      "loss": 1.757,
      "step": 659
    },
    {
      "epoch": 0.006340133767216225,
      "grad_norm": 1.2372474670410156,
      "learning_rate": 6.373338400808891e-05,
      "loss": 0.4271,
      "step": 660
    },
    {
      "epoch": 0.006349740030499886,
      "grad_norm": 0.7517385482788086,
      "learning_rate": 6.363848407153016e-05,
      "loss": 1.4436,
      "step": 661
    },
    {
      "epoch": 0.006359346293783547,
      "grad_norm": 1.5335979461669922,
      "learning_rate": 6.354353102457064e-05,
      "loss": 0.6825,
      "step": 662
    },
    {
      "epoch": 0.006368952557067208,
      "grad_norm": 0.8045147657394409,
      "learning_rate": 6.344852523697247e-05,
      "loss": 1.6608,
      "step": 663
    },
    {
      "epoch": 0.006378558820350869,
      "grad_norm": 0.8012584447860718,
      "learning_rate": 6.33534670787031e-05,
      "loss": 0.8046,
      "step": 664
    },
    {
      "epoch": 0.006388165083634529,
      "grad_norm": 0.9083331823348999,
      "learning_rate": 6.325835691993394e-05,
      "loss": 0.5685,
      "step": 665
    },
    {
      "epoch": 0.00639777134691819,
      "grad_norm": 0.939195454120636,
      "learning_rate": 6.31631951310389e-05,
      "loss": 1.4135,
      "step": 666
    },
    {
      "epoch": 0.006407377610201851,
      "grad_norm": 1.007326364517212,
      "learning_rate": 6.306798208259297e-05,
      "loss": 0.472,
      "step": 667
    },
    {
      "epoch": 0.0064169838734855125,
      "grad_norm": 1.0333832502365112,
      "learning_rate": 6.297271814537069e-05,
      "loss": 0.7612,
      "step": 668
    },
    {
      "epoch": 0.0064265901367691735,
      "grad_norm": 1.0427626371383667,
      "learning_rate": 6.287740369034485e-05,
      "loss": 0.7435,
      "step": 669
    },
    {
      "epoch": 0.0064361964000528345,
      "grad_norm": 0.852563202381134,
      "learning_rate": 6.278203908868487e-05,
      "loss": 1.2503,
      "step": 670
    },
    {
      "epoch": 0.006445802663336496,
      "grad_norm": 0.7133233547210693,
      "learning_rate": 6.26866247117555e-05,
      "loss": 0.7333,
      "step": 671
    },
    {
      "epoch": 0.006455408926620157,
      "grad_norm": 0.7750645875930786,
      "learning_rate": 6.259116093111538e-05,
      "loss": 0.755,
      "step": 672
    },
    {
      "epoch": 0.006465015189903818,
      "grad_norm": 0.7328550815582275,
      "learning_rate": 6.249564811851543e-05,
      "loss": 0.7444,
      "step": 673
    },
    {
      "epoch": 0.006474621453187478,
      "grad_norm": 1.0068961381912231,
      "learning_rate": 6.240008664589755e-05,
      "loss": 1.3012,
      "step": 674
    },
    {
      "epoch": 0.006484227716471139,
      "grad_norm": 1.217362880706787,
      "learning_rate": 6.230447688539316e-05,
      "loss": 1.3448,
      "step": 675
    },
    {
      "epoch": 0.0064938339797548,
      "grad_norm": 0.9604043364524841,
      "learning_rate": 6.220881920932169e-05,
      "loss": 1.7038,
      "step": 676
    },
    {
      "epoch": 0.006503440243038461,
      "grad_norm": 1.4450401067733765,
      "learning_rate": 6.211311399018916e-05,
      "loss": 1.2399,
      "step": 677
    },
    {
      "epoch": 0.006513046506322122,
      "grad_norm": 1.0792911052703857,
      "learning_rate": 6.201736160068676e-05,
      "loss": 1.1052,
      "step": 678
    },
    {
      "epoch": 0.006522652769605783,
      "grad_norm": 0.6653390526771545,
      "learning_rate": 6.192156241368929e-05,
      "loss": 0.4138,
      "step": 679
    },
    {
      "epoch": 0.006532259032889444,
      "grad_norm": 1.2464731931686401,
      "learning_rate": 6.182571680225389e-05,
      "loss": 1.8063,
      "step": 680
    },
    {
      "epoch": 0.006541865296173105,
      "grad_norm": 3.5973472595214844,
      "learning_rate": 6.172982513961845e-05,
      "loss": 1.3962,
      "step": 681
    },
    {
      "epoch": 0.006551471559456766,
      "grad_norm": 1.5316755771636963,
      "learning_rate": 6.163388779920012e-05,
      "loss": 1.085,
      "step": 682
    },
    {
      "epoch": 0.006561077822740426,
      "grad_norm": 1.0810480117797852,
      "learning_rate": 6.153790515459404e-05,
      "loss": 1.214,
      "step": 683
    },
    {
      "epoch": 0.006570684086024087,
      "grad_norm": 1.1533911228179932,
      "learning_rate": 6.144187757957167e-05,
      "loss": 0.8485,
      "step": 684
    },
    {
      "epoch": 0.0065802903493077484,
      "grad_norm": 1.2479926347732544,
      "learning_rate": 6.13458054480795e-05,
      "loss": 0.9956,
      "step": 685
    },
    {
      "epoch": 0.0065898966125914095,
      "grad_norm": 1.7469778060913086,
      "learning_rate": 6.124968913423753e-05,
      "loss": 1.5352,
      "step": 686
    },
    {
      "epoch": 0.0065995028758750705,
      "grad_norm": 0.9868939518928528,
      "learning_rate": 6.115352901233779e-05,
      "loss": 0.4239,
      "step": 687
    },
    {
      "epoch": 0.006609109139158732,
      "grad_norm": 1.5375529527664185,
      "learning_rate": 6.105732545684287e-05,
      "loss": 0.659,
      "step": 688
    },
    {
      "epoch": 0.006618715402442393,
      "grad_norm": 1.36130952835083,
      "learning_rate": 6.096107884238458e-05,
      "loss": 0.9391,
      "step": 689
    },
    {
      "epoch": 0.006628321665726054,
      "grad_norm": 1.2588740587234497,
      "learning_rate": 6.086478954376233e-05,
      "loss": 0.9372,
      "step": 690
    },
    {
      "epoch": 0.006637927929009715,
      "grad_norm": 1.4325975179672241,
      "learning_rate": 6.0768457935941817e-05,
      "loss": 1.5429,
      "step": 691
    },
    {
      "epoch": 0.006647534192293375,
      "grad_norm": 1.8158403635025024,
      "learning_rate": 6.0672084394053464e-05,
      "loss": 2.1542,
      "step": 692
    },
    {
      "epoch": 0.006657140455577036,
      "grad_norm": 1.7152471542358398,
      "learning_rate": 6.0575669293390954e-05,
      "loss": 2.6676,
      "step": 693
    },
    {
      "epoch": 0.006666746718860697,
      "grad_norm": 1.6411482095718384,
      "learning_rate": 6.047921300940988e-05,
      "loss": 1.2703,
      "step": 694
    },
    {
      "epoch": 0.006676352982144358,
      "grad_norm": 1.5534995794296265,
      "learning_rate": 6.038271591772615e-05,
      "loss": 2.5827,
      "step": 695
    },
    {
      "epoch": 0.006685959245428019,
      "grad_norm": 4.166032314300537,
      "learning_rate": 6.0286178394114644e-05,
      "loss": 1.82,
      "step": 696
    },
    {
      "epoch": 0.00669556550871168,
      "grad_norm": 4.059316158294678,
      "learning_rate": 6.0189600814507604e-05,
      "loss": 1.5249,
      "step": 697
    },
    {
      "epoch": 0.006705171771995341,
      "grad_norm": 2.2713623046875,
      "learning_rate": 6.009298355499336e-05,
      "loss": 1.3637,
      "step": 698
    },
    {
      "epoch": 0.006714778035279002,
      "grad_norm": 2.4380080699920654,
      "learning_rate": 5.9996326991814654e-05,
      "loss": 2.1375,
      "step": 699
    },
    {
      "epoch": 0.006724384298562663,
      "grad_norm": 2.346661329269409,
      "learning_rate": 5.989963150136735e-05,
      "loss": 2.3015,
      "step": 700
    },
    {
      "epoch": 0.006733990561846323,
      "grad_norm": 0.23651187121868134,
      "learning_rate": 5.980289746019892e-05,
      "loss": 1.9188,
      "step": 701
    },
    {
      "epoch": 0.0067435968251299844,
      "grad_norm": 0.23148967325687408,
      "learning_rate": 5.970612524500687e-05,
      "loss": 2.0968,
      "step": 702
    },
    {
      "epoch": 0.0067532030884136455,
      "grad_norm": 0.30047598481178284,
      "learning_rate": 5.9609315232637483e-05,
      "loss": 2.0637,
      "step": 703
    },
    {
      "epoch": 0.0067628093516973065,
      "grad_norm": 0.4141835868358612,
      "learning_rate": 5.951246780008409e-05,
      "loss": 1.5716,
      "step": 704
    },
    {
      "epoch": 0.006772415614980968,
      "grad_norm": 0.6243100166320801,
      "learning_rate": 5.941558332448589e-05,
      "loss": 2.0394,
      "step": 705
    },
    {
      "epoch": 0.006782021878264629,
      "grad_norm": 0.7601946592330933,
      "learning_rate": 5.931866218312623e-05,
      "loss": 1.8996,
      "step": 706
    },
    {
      "epoch": 0.00679162814154829,
      "grad_norm": 0.6117839813232422,
      "learning_rate": 5.922170475343125e-05,
      "loss": 2.6391,
      "step": 707
    },
    {
      "epoch": 0.006801234404831951,
      "grad_norm": 1.049946904182434,
      "learning_rate": 5.912471141296845e-05,
      "loss": 0.4379,
      "step": 708
    },
    {
      "epoch": 0.006810840668115612,
      "grad_norm": 0.738462507724762,
      "learning_rate": 5.9027682539445104e-05,
      "loss": 2.2825,
      "step": 709
    },
    {
      "epoch": 0.006820446931399272,
      "grad_norm": 0.6180376410484314,
      "learning_rate": 5.893061851070696e-05,
      "loss": 2.1615,
      "step": 710
    },
    {
      "epoch": 0.006830053194682933,
      "grad_norm": 1.3661633729934692,
      "learning_rate": 5.883351970473654e-05,
      "loss": 1.0772,
      "step": 711
    },
    {
      "epoch": 0.006839659457966594,
      "grad_norm": 0.7220520973205566,
      "learning_rate": 5.87363864996519e-05,
      "loss": 2.1969,
      "step": 712
    },
    {
      "epoch": 0.006849265721250255,
      "grad_norm": 0.8649919033050537,
      "learning_rate": 5.863921927370498e-05,
      "loss": 1.2892,
      "step": 713
    },
    {
      "epoch": 0.006858871984533916,
      "grad_norm": 0.8611554503440857,
      "learning_rate": 5.854201840528025e-05,
      "loss": 2.0199,
      "step": 714
    },
    {
      "epoch": 0.006868478247817577,
      "grad_norm": 0.7961388826370239,
      "learning_rate": 5.8444784272893175e-05,
      "loss": 1.9217,
      "step": 715
    },
    {
      "epoch": 0.006878084511101238,
      "grad_norm": 0.8721778988838196,
      "learning_rate": 5.834751725518872e-05,
      "loss": 0.788,
      "step": 716
    },
    {
      "epoch": 0.006887690774384899,
      "grad_norm": 1.1339527368545532,
      "learning_rate": 5.8250217730939973e-05,
      "loss": 1.61,
      "step": 717
    },
    {
      "epoch": 0.00689729703766856,
      "grad_norm": 1.4477065801620483,
      "learning_rate": 5.8152886079046557e-05,
      "loss": 1.4427,
      "step": 718
    },
    {
      "epoch": 0.0069069033009522204,
      "grad_norm": 1.6265654563903809,
      "learning_rate": 5.8055522678533225e-05,
      "loss": 1.0501,
      "step": 719
    },
    {
      "epoch": 0.0069165095642358815,
      "grad_norm": 1.0987688302993774,
      "learning_rate": 5.795812790854837e-05,
      "loss": 0.7332,
      "step": 720
    },
    {
      "epoch": 0.0069261158275195425,
      "grad_norm": 1.2305853366851807,
      "learning_rate": 5.786070214836254e-05,
      "loss": 1.5255,
      "step": 721
    },
    {
      "epoch": 0.0069357220908032036,
      "grad_norm": 0.9146164655685425,
      "learning_rate": 5.7763245777366946e-05,
      "loss": 1.3126,
      "step": 722
    },
    {
      "epoch": 0.006945328354086865,
      "grad_norm": 0.9069737792015076,
      "learning_rate": 5.7665759175072034e-05,
      "loss": 0.6871,
      "step": 723
    },
    {
      "epoch": 0.006954934617370526,
      "grad_norm": 0.9631674289703369,
      "learning_rate": 5.756824272110594e-05,
      "loss": 1.9215,
      "step": 724
    },
    {
      "epoch": 0.006964540880654187,
      "grad_norm": 1.0983726978302002,
      "learning_rate": 5.747069679521305e-05,
      "loss": 1.5532,
      "step": 725
    },
    {
      "epoch": 0.006974147143937848,
      "grad_norm": 1.024429440498352,
      "learning_rate": 5.737312177725257e-05,
      "loss": 0.9326,
      "step": 726
    },
    {
      "epoch": 0.006983753407221509,
      "grad_norm": 1.4739354848861694,
      "learning_rate": 5.727551804719693e-05,
      "loss": 1.1918,
      "step": 727
    },
    {
      "epoch": 0.006993359670505169,
      "grad_norm": 0.7881227135658264,
      "learning_rate": 5.7177885985130406e-05,
      "loss": 1.221,
      "step": 728
    },
    {
      "epoch": 0.00700296593378883,
      "grad_norm": 1.2614957094192505,
      "learning_rate": 5.708022597124758e-05,
      "loss": 1.3183,
      "step": 729
    },
    {
      "epoch": 0.007012572197072491,
      "grad_norm": 1.5614076852798462,
      "learning_rate": 5.698253838585192e-05,
      "loss": 0.9673,
      "step": 730
    },
    {
      "epoch": 0.007022178460356152,
      "grad_norm": 1.3219282627105713,
      "learning_rate": 5.688482360935423e-05,
      "loss": 1.091,
      "step": 731
    },
    {
      "epoch": 0.007031784723639813,
      "grad_norm": 1.3982640504837036,
      "learning_rate": 5.678708202227122e-05,
      "loss": 1.1743,
      "step": 732
    },
    {
      "epoch": 0.007041390986923474,
      "grad_norm": 0.9826419949531555,
      "learning_rate": 5.668931400522396e-05,
      "loss": 2.0959,
      "step": 733
    },
    {
      "epoch": 0.007050997250207135,
      "grad_norm": 0.9945687651634216,
      "learning_rate": 5.65915199389365e-05,
      "loss": 0.922,
      "step": 734
    },
    {
      "epoch": 0.007060603513490796,
      "grad_norm": 1.2570598125457764,
      "learning_rate": 5.649370020423431e-05,
      "loss": 0.5763,
      "step": 735
    },
    {
      "epoch": 0.007070209776774457,
      "grad_norm": 1.1600147485733032,
      "learning_rate": 5.63958551820428e-05,
      "loss": 1.7147,
      "step": 736
    },
    {
      "epoch": 0.0070798160400581175,
      "grad_norm": 1.7402187585830688,
      "learning_rate": 5.629798525338589e-05,
      "loss": 1.3233,
      "step": 737
    },
    {
      "epoch": 0.0070894223033417785,
      "grad_norm": 1.9408538341522217,
      "learning_rate": 5.6200090799384434e-05,
      "loss": 1.0154,
      "step": 738
    },
    {
      "epoch": 0.0070990285666254396,
      "grad_norm": 1.5093344449996948,
      "learning_rate": 5.6102172201254835e-05,
      "loss": 1.3302,
      "step": 739
    },
    {
      "epoch": 0.007108634829909101,
      "grad_norm": 1.5210307836532593,
      "learning_rate": 5.60042298403075e-05,
      "loss": 0.6474,
      "step": 740
    },
    {
      "epoch": 0.007118241093192762,
      "grad_norm": 1.397374153137207,
      "learning_rate": 5.5906264097945407e-05,
      "loss": 2.123,
      "step": 741
    },
    {
      "epoch": 0.007127847356476423,
      "grad_norm": 2.007950782775879,
      "learning_rate": 5.580827535566252e-05,
      "loss": 1.1553,
      "step": 742
    },
    {
      "epoch": 0.007137453619760084,
      "grad_norm": 1.709997296333313,
      "learning_rate": 5.5710263995042434e-05,
      "loss": 1.8799,
      "step": 743
    },
    {
      "epoch": 0.007147059883043745,
      "grad_norm": 1.440117597579956,
      "learning_rate": 5.5612230397756756e-05,
      "loss": 2.2835,
      "step": 744
    },
    {
      "epoch": 0.007156666146327406,
      "grad_norm": 2.815105438232422,
      "learning_rate": 5.551417494556376e-05,
      "loss": 1.916,
      "step": 745
    },
    {
      "epoch": 0.007166272409611066,
      "grad_norm": 1.952278733253479,
      "learning_rate": 5.5416098020306775e-05,
      "loss": 1.6018,
      "step": 746
    },
    {
      "epoch": 0.007175878672894727,
      "grad_norm": 2.495955467224121,
      "learning_rate": 5.531800000391275e-05,
      "loss": 2.1768,
      "step": 747
    },
    {
      "epoch": 0.007185484936178388,
      "grad_norm": 2.365595817565918,
      "learning_rate": 5.521988127839082e-05,
      "loss": 1.4999,
      "step": 748
    },
    {
      "epoch": 0.007195091199462049,
      "grad_norm": 2.582688093185425,
      "learning_rate": 5.5121742225830665e-05,
      "loss": 3.2851,
      "step": 749
    },
    {
      "epoch": 0.00720469746274571,
      "grad_norm": 2.7943549156188965,
      "learning_rate": 5.502358322840124e-05,
      "loss": 2.7174,
      "step": 750
    },
    {
      "epoch": 0.007214303726029371,
      "grad_norm": 0.2175443470478058,
      "learning_rate": 5.4925404668349076e-05,
      "loss": 1.8653,
      "step": 751
    },
    {
      "epoch": 0.007223909989313032,
      "grad_norm": 0.27280619740486145,
      "learning_rate": 5.482720692799694e-05,
      "loss": 2.0198,
      "step": 752
    },
    {
      "epoch": 0.007233516252596693,
      "grad_norm": 0.6662306189537048,
      "learning_rate": 5.472899038974225e-05,
      "loss": 1.6132,
      "step": 753
    },
    {
      "epoch": 0.007243122515880354,
      "grad_norm": 0.40286341309547424,
      "learning_rate": 5.463075543605565e-05,
      "loss": 1.9216,
      "step": 754
    },
    {
      "epoch": 0.0072527287791640145,
      "grad_norm": 0.39617720246315,
      "learning_rate": 5.45325024494795e-05,
      "loss": 1.7202,
      "step": 755
    },
    {
      "epoch": 0.0072623350424476756,
      "grad_norm": 0.4564102292060852,
      "learning_rate": 5.443423181262636e-05,
      "loss": 1.9006,
      "step": 756
    },
    {
      "epoch": 0.007271941305731337,
      "grad_norm": 0.4254985749721527,
      "learning_rate": 5.433594390817756e-05,
      "loss": 1.3095,
      "step": 757
    },
    {
      "epoch": 0.007281547569014998,
      "grad_norm": 0.5473141074180603,
      "learning_rate": 5.42376391188816e-05,
      "loss": 0.2916,
      "step": 758
    },
    {
      "epoch": 0.007291153832298659,
      "grad_norm": 2.2755699157714844,
      "learning_rate": 5.413931782755283e-05,
      "loss": 2.161,
      "step": 759
    },
    {
      "epoch": 0.00730076009558232,
      "grad_norm": 1.0997384786605835,
      "learning_rate": 5.4040980417069786e-05,
      "loss": 0.6705,
      "step": 760
    },
    {
      "epoch": 0.007310366358865981,
      "grad_norm": 0.559664785861969,
      "learning_rate": 5.3942627270373826e-05,
      "loss": 1.6167,
      "step": 761
    },
    {
      "epoch": 0.007319972622149642,
      "grad_norm": 1.585973858833313,
      "learning_rate": 5.3844258770467546e-05,
      "loss": 0.5012,
      "step": 762
    },
    {
      "epoch": 0.007329578885433303,
      "grad_norm": 0.79909348487854,
      "learning_rate": 5.374587530041335e-05,
      "loss": 1.263,
      "step": 763
    },
    {
      "epoch": 0.007339185148716963,
      "grad_norm": 0.6739315390586853,
      "learning_rate": 5.364747724333193e-05,
      "loss": 0.8633,
      "step": 764
    },
    {
      "epoch": 0.007348791412000624,
      "grad_norm": 0.7611742615699768,
      "learning_rate": 5.35490649824008e-05,
      "loss": 1.6013,
      "step": 765
    },
    {
      "epoch": 0.007358397675284285,
      "grad_norm": 0.9830788373947144,
      "learning_rate": 5.345063890085277e-05,
      "loss": 1.0599,
      "step": 766
    },
    {
      "epoch": 0.007368003938567946,
      "grad_norm": 0.7386084198951721,
      "learning_rate": 5.335219938197445e-05,
      "loss": 0.688,
      "step": 767
    },
    {
      "epoch": 0.007377610201851607,
      "grad_norm": 0.8981940150260925,
      "learning_rate": 5.325374680910483e-05,
      "loss": 1.4739,
      "step": 768
    },
    {
      "epoch": 0.007387216465135268,
      "grad_norm": 0.9368958473205566,
      "learning_rate": 5.315528156563367e-05,
      "loss": 1.5773,
      "step": 769
    },
    {
      "epoch": 0.007396822728418929,
      "grad_norm": 0.8272035717964172,
      "learning_rate": 5.305680403500016e-05,
      "loss": 0.9121,
      "step": 770
    },
    {
      "epoch": 0.00740642899170259,
      "grad_norm": 0.9068160057067871,
      "learning_rate": 5.295831460069124e-05,
      "loss": 1.0445,
      "step": 771
    },
    {
      "epoch": 0.007416035254986251,
      "grad_norm": 0.7859509587287903,
      "learning_rate": 5.2859813646240264e-05,
      "loss": 1.6082,
      "step": 772
    },
    {
      "epoch": 0.007425641518269912,
      "grad_norm": 1.319761872291565,
      "learning_rate": 5.276130155522541e-05,
      "loss": 1.6231,
      "step": 773
    },
    {
      "epoch": 0.007435247781553573,
      "grad_norm": 1.1460787057876587,
      "learning_rate": 5.266277871126827e-05,
      "loss": 1.3581,
      "step": 774
    },
    {
      "epoch": 0.007444854044837234,
      "grad_norm": 0.9503245949745178,
      "learning_rate": 5.256424549803228e-05,
      "loss": 1.3949,
      "step": 775
    },
    {
      "epoch": 0.007454460308120895,
      "grad_norm": 1.178879976272583,
      "learning_rate": 5.246570229922125e-05,
      "loss": 0.9441,
      "step": 776
    },
    {
      "epoch": 0.007464066571404556,
      "grad_norm": 0.8095961809158325,
      "learning_rate": 5.236714949857791e-05,
      "loss": 1.0463,
      "step": 777
    },
    {
      "epoch": 0.007473672834688217,
      "grad_norm": 1.2184892892837524,
      "learning_rate": 5.226858747988232e-05,
      "loss": 1.741,
      "step": 778
    },
    {
      "epoch": 0.007483279097971878,
      "grad_norm": 1.4450145959854126,
      "learning_rate": 5.2170016626950505e-05,
      "loss": 1.5171,
      "step": 779
    },
    {
      "epoch": 0.007492885361255539,
      "grad_norm": 1.5643839836120605,
      "learning_rate": 5.207143732363283e-05,
      "loss": 1.0711,
      "step": 780
    },
    {
      "epoch": 0.0075024916245392,
      "grad_norm": 1.181209683418274,
      "learning_rate": 5.1972849953812644e-05,
      "loss": 1.2028,
      "step": 781
    },
    {
      "epoch": 0.007512097887822861,
      "grad_norm": 1.0909786224365234,
      "learning_rate": 5.1874254901404615e-05,
      "loss": 1.1279,
      "step": 782
    },
    {
      "epoch": 0.007521704151106521,
      "grad_norm": 0.9027730226516724,
      "learning_rate": 5.1775652550353405e-05,
      "loss": 0.6074,
      "step": 783
    },
    {
      "epoch": 0.007531310414390182,
      "grad_norm": 1.2985749244689941,
      "learning_rate": 5.167704328463204e-05,
      "loss": 0.8716,
      "step": 784
    },
    {
      "epoch": 0.007540916677673843,
      "grad_norm": 1.5541952848434448,
      "learning_rate": 5.157842748824053e-05,
      "loss": 1.7836,
      "step": 785
    },
    {
      "epoch": 0.007550522940957504,
      "grad_norm": 1.4488773345947266,
      "learning_rate": 5.147980554520427e-05,
      "loss": 1.6139,
      "step": 786
    },
    {
      "epoch": 0.007560129204241165,
      "grad_norm": 1.253610610961914,
      "learning_rate": 5.138117783957261e-05,
      "loss": 0.686,
      "step": 787
    },
    {
      "epoch": 0.007569735467524826,
      "grad_norm": 1.4648478031158447,
      "learning_rate": 5.128254475541735e-05,
      "loss": 1.5117,
      "step": 788
    },
    {
      "epoch": 0.007579341730808487,
      "grad_norm": 1.3933899402618408,
      "learning_rate": 5.1183906676831197e-05,
      "loss": 1.2329,
      "step": 789
    },
    {
      "epoch": 0.007588947994092148,
      "grad_norm": 1.9739011526107788,
      "learning_rate": 5.108526398792637e-05,
      "loss": 2.153,
      "step": 790
    },
    {
      "epoch": 0.0075985542573758095,
      "grad_norm": 1.6183604001998901,
      "learning_rate": 5.098661707283298e-05,
      "loss": 1.7236,
      "step": 791
    },
    {
      "epoch": 0.00760816052065947,
      "grad_norm": 2.1352550983428955,
      "learning_rate": 5.088796631569762e-05,
      "loss": 2.4682,
      "step": 792
    },
    {
      "epoch": 0.007617766783943131,
      "grad_norm": 1.5372040271759033,
      "learning_rate": 5.078931210068185e-05,
      "loss": 2.1808,
      "step": 793
    },
    {
      "epoch": 0.007627373047226792,
      "grad_norm": 1.695050597190857,
      "learning_rate": 5.0690654811960694e-05,
      "loss": 1.7808,
      "step": 794
    },
    {
      "epoch": 0.007636979310510453,
      "grad_norm": 1.7952525615692139,
      "learning_rate": 5.059199483372114e-05,
      "loss": 2.244,
      "step": 795
    },
    {
      "epoch": 0.007646585573794114,
      "grad_norm": 2.353778123855591,
      "learning_rate": 5.049333255016065e-05,
      "loss": 1.4523,
      "step": 796
    },
    {
      "epoch": 0.007656191837077775,
      "grad_norm": 2.1845998764038086,
      "learning_rate": 5.039466834548568e-05,
      "loss": 1.9827,
      "step": 797
    },
    {
      "epoch": 0.007665798100361436,
      "grad_norm": 2.4176957607269287,
      "learning_rate": 5.029600260391011e-05,
      "loss": 2.1736,
      "step": 798
    },
    {
      "epoch": 0.007675404363645097,
      "grad_norm": 4.842705249786377,
      "learning_rate": 5.0197335709653883e-05,
      "loss": 3.2633,
      "step": 799
    },
    {
      "epoch": 0.007685010626928758,
      "grad_norm": 4.012521743774414,
      "learning_rate": 5.0098668046941374e-05,
      "loss": 2.6827,
      "step": 800
    },
    {
      "epoch": 0.007694616890212418,
      "grad_norm": 0.28005409240722656,
      "learning_rate": 5e-05,
      "loss": 1.9109,
      "step": 801
    },
    {
      "epoch": 0.007704223153496079,
      "grad_norm": 0.34262511134147644,
      "learning_rate": 4.990133195305863e-05,
      "loss": 1.9908,
      "step": 802
    },
    {
      "epoch": 0.00771382941677974,
      "grad_norm": 0.47594109177589417,
      "learning_rate": 4.980266429034613e-05,
      "loss": 1.2814,
      "step": 803
    },
    {
      "epoch": 0.007723435680063401,
      "grad_norm": 0.5651593804359436,
      "learning_rate": 4.9703997396089915e-05,
      "loss": 1.676,
      "step": 804
    },
    {
      "epoch": 0.007733041943347062,
      "grad_norm": 0.4979057013988495,
      "learning_rate": 4.960533165451435e-05,
      "loss": 2.0841,
      "step": 805
    },
    {
      "epoch": 0.007742648206630723,
      "grad_norm": 0.5786827802658081,
      "learning_rate": 4.950666744983935e-05,
      "loss": 1.3463,
      "step": 806
    },
    {
      "epoch": 0.007752254469914384,
      "grad_norm": 0.8138961791992188,
      "learning_rate": 4.9408005166278855e-05,
      "loss": 1.3522,
      "step": 807
    },
    {
      "epoch": 0.0077618607331980454,
      "grad_norm": 0.8986212611198425,
      "learning_rate": 4.930934518803931e-05,
      "loss": 1.942,
      "step": 808
    },
    {
      "epoch": 0.0077714669964817065,
      "grad_norm": 0.8701353669166565,
      "learning_rate": 4.921068789931816e-05,
      "loss": 1.2113,
      "step": 809
    },
    {
      "epoch": 0.007781073259765367,
      "grad_norm": 0.5777512192726135,
      "learning_rate": 4.911203368430239e-05,
      "loss": 1.5071,
      "step": 810
    },
    {
      "epoch": 0.007790679523049028,
      "grad_norm": 0.8532912731170654,
      "learning_rate": 4.901338292716704e-05,
      "loss": 1.6725,
      "step": 811
    },
    {
      "epoch": 0.007800285786332689,
      "grad_norm": 1.083898663520813,
      "learning_rate": 4.891473601207364e-05,
      "loss": 0.1608,
      "step": 812
    },
    {
      "epoch": 0.00780989204961635,
      "grad_norm": 0.8599928021430969,
      "learning_rate": 4.8816093323168815e-05,
      "loss": 1.5031,
      "step": 813
    },
    {
      "epoch": 0.00781949831290001,
      "grad_norm": 0.8121448755264282,
      "learning_rate": 4.871745524458267e-05,
      "loss": 1.637,
      "step": 814
    },
    {
      "epoch": 0.007829104576183672,
      "grad_norm": 0.9152563214302063,
      "learning_rate": 4.8618822160427406e-05,
      "loss": 1.7807,
      "step": 815
    },
    {
      "epoch": 0.007838710839467333,
      "grad_norm": 0.7985622882843018,
      "learning_rate": 4.8520194454795734e-05,
      "loss": 0.9186,
      "step": 816
    },
    {
      "epoch": 0.007848317102750994,
      "grad_norm": 0.8245862722396851,
      "learning_rate": 4.842157251175947e-05,
      "loss": 0.5464,
      "step": 817
    },
    {
      "epoch": 0.007857923366034655,
      "grad_norm": 0.8594374060630798,
      "learning_rate": 4.832295671536797e-05,
      "loss": 1.5298,
      "step": 818
    },
    {
      "epoch": 0.007867529629318316,
      "grad_norm": 0.8866499662399292,
      "learning_rate": 4.822434744964661e-05,
      "loss": 1.0295,
      "step": 819
    },
    {
      "epoch": 0.007877135892601977,
      "grad_norm": 0.684601366519928,
      "learning_rate": 4.8125745098595396e-05,
      "loss": 1.4636,
      "step": 820
    },
    {
      "epoch": 0.007886742155885638,
      "grad_norm": 1.0914888381958008,
      "learning_rate": 4.802715004618737e-05,
      "loss": 0.9231,
      "step": 821
    },
    {
      "epoch": 0.0078963484191693,
      "grad_norm": 0.933941662311554,
      "learning_rate": 4.7928562676367175e-05,
      "loss": 1.1464,
      "step": 822
    },
    {
      "epoch": 0.007905954682452958,
      "grad_norm": 0.8920226693153381,
      "learning_rate": 4.7829983373049507e-05,
      "loss": 1.1701,
      "step": 823
    },
    {
      "epoch": 0.00791556094573662,
      "grad_norm": 0.9995471239089966,
      "learning_rate": 4.7731412520117696e-05,
      "loss": 0.8185,
      "step": 824
    },
    {
      "epoch": 0.00792516720902028,
      "grad_norm": 0.8008311986923218,
      "learning_rate": 4.763285050142211e-05,
      "loss": 1.2535,
      "step": 825
    },
    {
      "epoch": 0.007934773472303942,
      "grad_norm": 1.2564903497695923,
      "learning_rate": 4.7534297700778745e-05,
      "loss": 1.1128,
      "step": 826
    },
    {
      "epoch": 0.007944379735587603,
      "grad_norm": 1.232520580291748,
      "learning_rate": 4.743575450196773e-05,
      "loss": 1.387,
      "step": 827
    },
    {
      "epoch": 0.007953985998871264,
      "grad_norm": 0.9317668676376343,
      "learning_rate": 4.733722128873173e-05,
      "loss": 1.381,
      "step": 828
    },
    {
      "epoch": 0.007963592262154925,
      "grad_norm": 0.9363964200019836,
      "learning_rate": 4.7238698444774595e-05,
      "loss": 1.4447,
      "step": 829
    },
    {
      "epoch": 0.007973198525438586,
      "grad_norm": 1.0789803266525269,
      "learning_rate": 4.7140186353759754e-05,
      "loss": 1.1348,
      "step": 830
    },
    {
      "epoch": 0.007982804788722247,
      "grad_norm": 0.9478041529655457,
      "learning_rate": 4.704168539930878e-05,
      "loss": 0.9745,
      "step": 831
    },
    {
      "epoch": 0.007992411052005908,
      "grad_norm": 1.2430824041366577,
      "learning_rate": 4.694319596499985e-05,
      "loss": 1.1838,
      "step": 832
    },
    {
      "epoch": 0.008002017315289569,
      "grad_norm": 1.0949147939682007,
      "learning_rate": 4.6844718434366334e-05,
      "loss": 1.1484,
      "step": 833
    },
    {
      "epoch": 0.00801162357857323,
      "grad_norm": 1.1800644397735596,
      "learning_rate": 4.6746253190895187e-05,
      "loss": 1.1536,
      "step": 834
    },
    {
      "epoch": 0.008021229841856891,
      "grad_norm": 1.0418188571929932,
      "learning_rate": 4.664780061802557e-05,
      "loss": 1.0326,
      "step": 835
    },
    {
      "epoch": 0.008030836105140552,
      "grad_norm": 1.0652717351913452,
      "learning_rate": 4.6549361099147245e-05,
      "loss": 0.8009,
      "step": 836
    },
    {
      "epoch": 0.008040442368424213,
      "grad_norm": 1.2421663999557495,
      "learning_rate": 4.64509350175992e-05,
      "loss": 0.7165,
      "step": 837
    },
    {
      "epoch": 0.008050048631707874,
      "grad_norm": 1.3865346908569336,
      "learning_rate": 4.635252275666808e-05,
      "loss": 1.0376,
      "step": 838
    },
    {
      "epoch": 0.008059654894991535,
      "grad_norm": 1.5849058628082275,
      "learning_rate": 4.6254124699586656e-05,
      "loss": 0.9685,
      "step": 839
    },
    {
      "epoch": 0.008069261158275196,
      "grad_norm": 1.0971784591674805,
      "learning_rate": 4.6155741229532466e-05,
      "loss": 0.3945,
      "step": 840
    },
    {
      "epoch": 0.008078867421558856,
      "grad_norm": 1.1344660520553589,
      "learning_rate": 4.605737272962618e-05,
      "loss": 1.3123,
      "step": 841
    },
    {
      "epoch": 0.008088473684842517,
      "grad_norm": 1.4826970100402832,
      "learning_rate": 4.5959019582930226e-05,
      "loss": 0.9792,
      "step": 842
    },
    {
      "epoch": 0.008098079948126178,
      "grad_norm": 1.6331377029418945,
      "learning_rate": 4.5860682172447184e-05,
      "loss": 1.9312,
      "step": 843
    },
    {
      "epoch": 0.008107686211409839,
      "grad_norm": 1.9050483703613281,
      "learning_rate": 4.576236088111842e-05,
      "loss": 1.4512,
      "step": 844
    },
    {
      "epoch": 0.0081172924746935,
      "grad_norm": 2.199693441390991,
      "learning_rate": 4.566405609182247e-05,
      "loss": 1.4704,
      "step": 845
    },
    {
      "epoch": 0.00812689873797716,
      "grad_norm": 1.617666244506836,
      "learning_rate": 4.5565768187373645e-05,
      "loss": 0.6612,
      "step": 846
    },
    {
      "epoch": 0.008136505001260822,
      "grad_norm": 1.9102528095245361,
      "learning_rate": 4.546749755052051e-05,
      "loss": 1.8447,
      "step": 847
    },
    {
      "epoch": 0.008146111264544483,
      "grad_norm": 4.163633346557617,
      "learning_rate": 4.5369244563944344e-05,
      "loss": 1.3268,
      "step": 848
    },
    {
      "epoch": 0.008155717527828144,
      "grad_norm": 2.9465367794036865,
      "learning_rate": 4.527100961025776e-05,
      "loss": 1.9061,
      "step": 849
    },
    {
      "epoch": 0.008165323791111805,
      "grad_norm": 3.069312572479248,
      "learning_rate": 4.517279307200307e-05,
      "loss": 1.6854,
      "step": 850
    },
    {
      "epoch": 0.008174930054395466,
      "grad_norm": 0.21755695343017578,
      "learning_rate": 4.507459533165093e-05,
      "loss": 2.1128,
      "step": 851
    },
    {
      "epoch": 0.008184536317679127,
      "grad_norm": 0.2629605531692505,
      "learning_rate": 4.497641677159877e-05,
      "loss": 2.0752,
      "step": 852
    },
    {
      "epoch": 0.008194142580962788,
      "grad_norm": 0.40802887082099915,
      "learning_rate": 4.4878257774169346e-05,
      "loss": 2.0199,
      "step": 853
    },
    {
      "epoch": 0.008203748844246449,
      "grad_norm": 0.5108217000961304,
      "learning_rate": 4.478011872160921e-05,
      "loss": 1.4964,
      "step": 854
    },
    {
      "epoch": 0.00821335510753011,
      "grad_norm": 0.5092230439186096,
      "learning_rate": 4.4681999996087274e-05,
      "loss": 2.0481,
      "step": 855
    },
    {
      "epoch": 0.008222961370813771,
      "grad_norm": 0.97291499376297,
      "learning_rate": 4.4583901979693236e-05,
      "loss": 1.8207,
      "step": 856
    },
    {
      "epoch": 0.008232567634097432,
      "grad_norm": 0.6948288679122925,
      "learning_rate": 4.448582505443625e-05,
      "loss": 2.0944,
      "step": 857
    },
    {
      "epoch": 0.008242173897381093,
      "grad_norm": 0.7920825481414795,
      "learning_rate": 4.438776960224325e-05,
      "loss": 1.0083,
      "step": 858
    },
    {
      "epoch": 0.008251780160664753,
      "grad_norm": 0.5737664699554443,
      "learning_rate": 4.4289736004957585e-05,
      "loss": 1.3958,
      "step": 859
    },
    {
      "epoch": 0.008261386423948414,
      "grad_norm": 0.7429783344268799,
      "learning_rate": 4.419172464433749e-05,
      "loss": 1.298,
      "step": 860
    },
    {
      "epoch": 0.008270992687232075,
      "grad_norm": 0.5883791446685791,
      "learning_rate": 4.4093735902054605e-05,
      "loss": 1.2556,
      "step": 861
    },
    {
      "epoch": 0.008280598950515736,
      "grad_norm": 0.7453375458717346,
      "learning_rate": 4.399577015969251e-05,
      "loss": 1.2325,
      "step": 862
    },
    {
      "epoch": 0.008290205213799397,
      "grad_norm": 0.8359687924385071,
      "learning_rate": 4.3897827798745183e-05,
      "loss": 1.6855,
      "step": 863
    },
    {
      "epoch": 0.008299811477083058,
      "grad_norm": 0.8295079469680786,
      "learning_rate": 4.379990920061559e-05,
      "loss": 1.5158,
      "step": 864
    },
    {
      "epoch": 0.008309417740366719,
      "grad_norm": 0.7614023685455322,
      "learning_rate": 4.3702014746614136e-05,
      "loss": 1.3692,
      "step": 865
    },
    {
      "epoch": 0.00831902400365038,
      "grad_norm": 0.697290301322937,
      "learning_rate": 4.360414481795719e-05,
      "loss": 1.4369,
      "step": 866
    },
    {
      "epoch": 0.008328630266934041,
      "grad_norm": 1.2065218687057495,
      "learning_rate": 4.350629979576569e-05,
      "loss": 0.8997,
      "step": 867
    },
    {
      "epoch": 0.008338236530217702,
      "grad_norm": 0.7741699814796448,
      "learning_rate": 4.34084800610635e-05,
      "loss": 1.6777,
      "step": 868
    },
    {
      "epoch": 0.008347842793501363,
      "grad_norm": 0.9095771908760071,
      "learning_rate": 4.331068599477605e-05,
      "loss": 1.0961,
      "step": 869
    },
    {
      "epoch": 0.008357449056785024,
      "grad_norm": 1.1288762092590332,
      "learning_rate": 4.32129179777288e-05,
      "loss": 1.083,
      "step": 870
    },
    {
      "epoch": 0.008367055320068685,
      "grad_norm": 0.9637406468391418,
      "learning_rate": 4.311517639064578e-05,
      "loss": 0.7931,
      "step": 871
    },
    {
      "epoch": 0.008376661583352346,
      "grad_norm": 1.5801156759262085,
      "learning_rate": 4.301746161414808e-05,
      "loss": 1.8733,
      "step": 872
    },
    {
      "epoch": 0.008386267846636007,
      "grad_norm": 1.188769817352295,
      "learning_rate": 4.2919774028752436e-05,
      "loss": 0.9238,
      "step": 873
    },
    {
      "epoch": 0.008395874109919668,
      "grad_norm": 1.186320185661316,
      "learning_rate": 4.282211401486961e-05,
      "loss": 0.6842,
      "step": 874
    },
    {
      "epoch": 0.00840548037320333,
      "grad_norm": 1.1875613927841187,
      "learning_rate": 4.27244819528031e-05,
      "loss": 1.8859,
      "step": 875
    },
    {
      "epoch": 0.00841508663648699,
      "grad_norm": 1.0114357471466064,
      "learning_rate": 4.262687822274744e-05,
      "loss": 0.6381,
      "step": 876
    },
    {
      "epoch": 0.00842469289977065,
      "grad_norm": 0.8018882870674133,
      "learning_rate": 4.2529303204786953e-05,
      "loss": 1.1794,
      "step": 877
    },
    {
      "epoch": 0.00843429916305431,
      "grad_norm": 1.1977322101593018,
      "learning_rate": 4.243175727889408e-05,
      "loss": 1.3599,
      "step": 878
    },
    {
      "epoch": 0.008443905426337972,
      "grad_norm": 0.9608428478240967,
      "learning_rate": 4.233424082492797e-05,
      "loss": 1.3254,
      "step": 879
    },
    {
      "epoch": 0.008453511689621633,
      "grad_norm": 2.5006957054138184,
      "learning_rate": 4.223675422263306e-05,
      "loss": 1.3162,
      "step": 880
    },
    {
      "epoch": 0.008463117952905294,
      "grad_norm": 1.2491915225982666,
      "learning_rate": 4.213929785163747e-05,
      "loss": 1.1267,
      "step": 881
    },
    {
      "epoch": 0.008472724216188955,
      "grad_norm": 1.2751023769378662,
      "learning_rate": 4.204187209145164e-05,
      "loss": 0.7234,
      "step": 882
    },
    {
      "epoch": 0.008482330479472616,
      "grad_norm": 0.996956467628479,
      "learning_rate": 4.1944477321466786e-05,
      "loss": 0.547,
      "step": 883
    },
    {
      "epoch": 0.008491936742756277,
      "grad_norm": 1.9938204288482666,
      "learning_rate": 4.184711392095346e-05,
      "loss": 0.9268,
      "step": 884
    },
    {
      "epoch": 0.008501543006039938,
      "grad_norm": 1.4488263130187988,
      "learning_rate": 4.1749782269060045e-05,
      "loss": 1.3632,
      "step": 885
    },
    {
      "epoch": 0.008511149269323599,
      "grad_norm": 0.8692067861557007,
      "learning_rate": 4.165248274481128e-05,
      "loss": 1.0008,
      "step": 886
    },
    {
      "epoch": 0.00852075553260726,
      "grad_norm": 1.0646692514419556,
      "learning_rate": 4.1555215727106844e-05,
      "loss": 0.5483,
      "step": 887
    },
    {
      "epoch": 0.008530361795890921,
      "grad_norm": 2.2823381423950195,
      "learning_rate": 4.145798159471975e-05,
      "loss": 1.187,
      "step": 888
    },
    {
      "epoch": 0.008539968059174582,
      "grad_norm": 1.5813469886779785,
      "learning_rate": 4.136078072629503e-05,
      "loss": 0.678,
      "step": 889
    },
    {
      "epoch": 0.008549574322458243,
      "grad_norm": 1.3979244232177734,
      "learning_rate": 4.1263613500348106e-05,
      "loss": 1.4461,
      "step": 890
    },
    {
      "epoch": 0.008559180585741904,
      "grad_norm": 1.895545244216919,
      "learning_rate": 4.116648029526347e-05,
      "loss": 2.1331,
      "step": 891
    },
    {
      "epoch": 0.008568786849025565,
      "grad_norm": 1.9366155862808228,
      "learning_rate": 4.106938148929306e-05,
      "loss": 1.1776,
      "step": 892
    },
    {
      "epoch": 0.008578393112309226,
      "grad_norm": 1.934545874595642,
      "learning_rate": 4.097231746055491e-05,
      "loss": 2.151,
      "step": 893
    },
    {
      "epoch": 0.008587999375592887,
      "grad_norm": 2.7551753520965576,
      "learning_rate": 4.087528858703157e-05,
      "loss": 1.5815,
      "step": 894
    },
    {
      "epoch": 0.008597605638876548,
      "grad_norm": 2.1389689445495605,
      "learning_rate": 4.077829524656877e-05,
      "loss": 1.8332,
      "step": 895
    },
    {
      "epoch": 0.008607211902160208,
      "grad_norm": 2.1170918941497803,
      "learning_rate": 4.068133781687378e-05,
      "loss": 2.0668,
      "step": 896
    },
    {
      "epoch": 0.008616818165443869,
      "grad_norm": 2.2046422958374023,
      "learning_rate": 4.05844166755141e-05,
      "loss": 1.7716,
      "step": 897
    },
    {
      "epoch": 0.00862642442872753,
      "grad_norm": 2.4212512969970703,
      "learning_rate": 4.04875321999159e-05,
      "loss": 2.3655,
      "step": 898
    },
    {
      "epoch": 0.00863603069201119,
      "grad_norm": 2.662921667098999,
      "learning_rate": 4.039068476736253e-05,
      "loss": 2.2696,
      "step": 899
    },
    {
      "epoch": 0.008645636955294852,
      "grad_norm": 2.7001564502716064,
      "learning_rate": 4.0293874754993134e-05,
      "loss": 2.1271,
      "step": 900
    },
    {
      "epoch": 0.008655243218578513,
      "grad_norm": 0.22573977708816528,
      "learning_rate": 4.01971025398011e-05,
      "loss": 1.9665,
      "step": 901
    },
    {
      "epoch": 0.008664849481862174,
      "grad_norm": 0.30016207695007324,
      "learning_rate": 4.010036849863267e-05,
      "loss": 2.307,
      "step": 902
    },
    {
      "epoch": 0.008674455745145835,
      "grad_norm": 0.5028771758079529,
      "learning_rate": 4.000367300818537e-05,
      "loss": 1.7471,
      "step": 903
    },
    {
      "epoch": 0.008684062008429496,
      "grad_norm": 0.7436515092849731,
      "learning_rate": 3.9907016445006675e-05,
      "loss": 0.908,
      "step": 904
    },
    {
      "epoch": 0.008693668271713157,
      "grad_norm": 0.5793784260749817,
      "learning_rate": 3.98103991854924e-05,
      "loss": 1.3202,
      "step": 905
    },
    {
      "epoch": 0.008703274534996818,
      "grad_norm": 0.5660339593887329,
      "learning_rate": 3.971382160588536e-05,
      "loss": 2.4011,
      "step": 906
    },
    {
      "epoch": 0.00871288079828048,
      "grad_norm": 0.9589015245437622,
      "learning_rate": 3.961728408227384e-05,
      "loss": 0.7023,
      "step": 907
    },
    {
      "epoch": 0.00872248706156414,
      "grad_norm": 1.1704084873199463,
      "learning_rate": 3.952078699059012e-05,
      "loss": 1.1041,
      "step": 908
    },
    {
      "epoch": 0.008732093324847801,
      "grad_norm": 0.583778977394104,
      "learning_rate": 3.942433070660905e-05,
      "loss": 1.0679,
      "step": 909
    },
    {
      "epoch": 0.008741699588131462,
      "grad_norm": 0.8043504953384399,
      "learning_rate": 3.9327915605946555e-05,
      "loss": 0.2447,
      "step": 910
    },
    {
      "epoch": 0.008751305851415123,
      "grad_norm": 0.8582820296287537,
      "learning_rate": 3.923154206405819e-05,
      "loss": 0.2773,
      "step": 911
    },
    {
      "epoch": 0.008760912114698784,
      "grad_norm": 0.8017820715904236,
      "learning_rate": 3.913521045623767e-05,
      "loss": 2.9081,
      "step": 912
    },
    {
      "epoch": 0.008770518377982445,
      "grad_norm": 0.9629442691802979,
      "learning_rate": 3.9038921157615444e-05,
      "loss": 1.3954,
      "step": 913
    },
    {
      "epoch": 0.008780124641266105,
      "grad_norm": 0.9008824825286865,
      "learning_rate": 3.8942674543157144e-05,
      "loss": 1.7669,
      "step": 914
    },
    {
      "epoch": 0.008789730904549766,
      "grad_norm": 0.8641378879547119,
      "learning_rate": 3.884647098766224e-05,
      "loss": 1.0665,
      "step": 915
    },
    {
      "epoch": 0.008799337167833427,
      "grad_norm": 1.2315315008163452,
      "learning_rate": 3.8750310865762467e-05,
      "loss": 1.0329,
      "step": 916
    },
    {
      "epoch": 0.008808943431117088,
      "grad_norm": 0.9756618142127991,
      "learning_rate": 3.8654194551920485e-05,
      "loss": 1.4467,
      "step": 917
    },
    {
      "epoch": 0.008818549694400749,
      "grad_norm": 0.7158561944961548,
      "learning_rate": 3.855812242042834e-05,
      "loss": 1.3531,
      "step": 918
    },
    {
      "epoch": 0.00882815595768441,
      "grad_norm": 1.2072174549102783,
      "learning_rate": 3.846209484540597e-05,
      "loss": 1.1226,
      "step": 919
    },
    {
      "epoch": 0.008837762220968071,
      "grad_norm": 0.7302363514900208,
      "learning_rate": 3.8366112200799885e-05,
      "loss": 1.0807,
      "step": 920
    },
    {
      "epoch": 0.008847368484251732,
      "grad_norm": 0.8558917045593262,
      "learning_rate": 3.827017486038157e-05,
      "loss": 1.2806,
      "step": 921
    },
    {
      "epoch": 0.008856974747535393,
      "grad_norm": 0.8099687099456787,
      "learning_rate": 3.8174283197746115e-05,
      "loss": 1.636,
      "step": 922
    },
    {
      "epoch": 0.008866581010819054,
      "grad_norm": 0.9366759061813354,
      "learning_rate": 3.8078437586310716e-05,
      "loss": 1.4779,
      "step": 923
    },
    {
      "epoch": 0.008876187274102715,
      "grad_norm": 0.869033694267273,
      "learning_rate": 3.7982638399313275e-05,
      "loss": 1.2049,
      "step": 924
    },
    {
      "epoch": 0.008885793537386376,
      "grad_norm": 0.8336467146873474,
      "learning_rate": 3.788688600981085e-05,
      "loss": 0.9452,
      "step": 925
    },
    {
      "epoch": 0.008895399800670037,
      "grad_norm": 2.6784579753875732,
      "learning_rate": 3.7791180790678304e-05,
      "loss": 1.15,
      "step": 926
    },
    {
      "epoch": 0.008905006063953698,
      "grad_norm": 0.9354357719421387,
      "learning_rate": 3.769552311460684e-05,
      "loss": 1.5219,
      "step": 927
    },
    {
      "epoch": 0.00891461232723736,
      "grad_norm": 0.8974885940551758,
      "learning_rate": 3.759991335410245e-05,
      "loss": 0.5291,
      "step": 928
    },
    {
      "epoch": 0.00892421859052102,
      "grad_norm": 1.2430256605148315,
      "learning_rate": 3.750435188148459e-05,
      "loss": 0.9052,
      "step": 929
    },
    {
      "epoch": 0.008933824853804681,
      "grad_norm": 0.8521973490715027,
      "learning_rate": 3.740883906888463e-05,
      "loss": 0.819,
      "step": 930
    },
    {
      "epoch": 0.008943431117088342,
      "grad_norm": 1.6005194187164307,
      "learning_rate": 3.73133752882445e-05,
      "loss": 1.3707,
      "step": 931
    },
    {
      "epoch": 0.008953037380372002,
      "grad_norm": 0.9994732141494751,
      "learning_rate": 3.721796091131515e-05,
      "loss": 1.015,
      "step": 932
    },
    {
      "epoch": 0.008962643643655663,
      "grad_norm": 1.673446774482727,
      "learning_rate": 3.712259630965518e-05,
      "loss": 1.2889,
      "step": 933
    },
    {
      "epoch": 0.008972249906939324,
      "grad_norm": 1.7021641731262207,
      "learning_rate": 3.702728185462931e-05,
      "loss": 1.7331,
      "step": 934
    },
    {
      "epoch": 0.008981856170222985,
      "grad_norm": 1.5104621648788452,
      "learning_rate": 3.6932017917407045e-05,
      "loss": 1.2059,
      "step": 935
    },
    {
      "epoch": 0.008991462433506646,
      "grad_norm": 1.31712007522583,
      "learning_rate": 3.6836804868961096e-05,
      "loss": 0.9306,
      "step": 936
    },
    {
      "epoch": 0.009001068696790307,
      "grad_norm": 1.2793023586273193,
      "learning_rate": 3.6741643080066065e-05,
      "loss": 0.752,
      "step": 937
    },
    {
      "epoch": 0.009010674960073968,
      "grad_norm": 1.8931103944778442,
      "learning_rate": 3.6646532921296916e-05,
      "loss": 0.7365,
      "step": 938
    },
    {
      "epoch": 0.009020281223357629,
      "grad_norm": 2.0308194160461426,
      "learning_rate": 3.655147476302754e-05,
      "loss": 1.1353,
      "step": 939
    },
    {
      "epoch": 0.00902988748664129,
      "grad_norm": 1.1300160884857178,
      "learning_rate": 3.645646897542937e-05,
      "loss": 0.6225,
      "step": 940
    },
    {
      "epoch": 0.009039493749924951,
      "grad_norm": 1.6924431324005127,
      "learning_rate": 3.636151592846985e-05,
      "loss": 1.7395,
      "step": 941
    },
    {
      "epoch": 0.009049100013208612,
      "grad_norm": 1.5833849906921387,
      "learning_rate": 3.626661599191109e-05,
      "loss": 1.965,
      "step": 942
    },
    {
      "epoch": 0.009058706276492273,
      "grad_norm": 1.2627257108688354,
      "learning_rate": 3.617176953530835e-05,
      "loss": 2.1956,
      "step": 943
    },
    {
      "epoch": 0.009068312539775934,
      "grad_norm": 3.2384164333343506,
      "learning_rate": 3.6076976928008655e-05,
      "loss": 2.3885,
      "step": 944
    },
    {
      "epoch": 0.009077918803059595,
      "grad_norm": 2.144325017929077,
      "learning_rate": 3.5982238539149285e-05,
      "loss": 2.6086,
      "step": 945
    },
    {
      "epoch": 0.009087525066343256,
      "grad_norm": 1.9335722923278809,
      "learning_rate": 3.5887554737656436e-05,
      "loss": 2.2584,
      "step": 946
    },
    {
      "epoch": 0.009097131329626917,
      "grad_norm": 1.953365683555603,
      "learning_rate": 3.579292589224375e-05,
      "loss": 1.9991,
      "step": 947
    },
    {
      "epoch": 0.009106737592910578,
      "grad_norm": 2.2731497287750244,
      "learning_rate": 3.5698352371410766e-05,
      "loss": 2.629,
      "step": 948
    },
    {
      "epoch": 0.00911634385619424,
      "grad_norm": 2.468738079071045,
      "learning_rate": 3.560383454344168e-05,
      "loss": 1.5646,
      "step": 949
    },
    {
      "epoch": 0.009125950119477899,
      "grad_norm": 3.085813283920288,
      "learning_rate": 3.550937277640375e-05,
      "loss": 2.7305,
      "step": 950
    },
    {
      "epoch": 0.00913555638276156,
      "grad_norm": 0.22140689194202423,
      "learning_rate": 3.541496743814596e-05,
      "loss": 1.7153,
      "step": 951
    },
    {
      "epoch": 0.009145162646045221,
      "grad_norm": 0.3282793164253235,
      "learning_rate": 3.5320618896297526e-05,
      "loss": 1.9968,
      "step": 952
    },
    {
      "epoch": 0.009154768909328882,
      "grad_norm": 0.5756869316101074,
      "learning_rate": 3.522632751826651e-05,
      "loss": 1.2241,
      "step": 953
    },
    {
      "epoch": 0.009164375172612543,
      "grad_norm": 0.8841428160667419,
      "learning_rate": 3.513209367123834e-05,
      "loss": 0.823,
      "step": 954
    },
    {
      "epoch": 0.009173981435896204,
      "grad_norm": 0.9375598430633545,
      "learning_rate": 3.503791772217445e-05,
      "loss": 0.9061,
      "step": 955
    },
    {
      "epoch": 0.009183587699179865,
      "grad_norm": 0.691542387008667,
      "learning_rate": 3.494380003781076e-05,
      "loss": 1.6351,
      "step": 956
    },
    {
      "epoch": 0.009193193962463526,
      "grad_norm": 1.019049882888794,
      "learning_rate": 3.484974098465636e-05,
      "loss": 1.4461,
      "step": 957
    },
    {
      "epoch": 0.009202800225747187,
      "grad_norm": 0.5466054677963257,
      "learning_rate": 3.475574092899198e-05,
      "loss": 2.1113,
      "step": 958
    },
    {
      "epoch": 0.009212406489030848,
      "grad_norm": 0.819279134273529,
      "learning_rate": 3.4661800236868604e-05,
      "loss": 1.0559,
      "step": 959
    },
    {
      "epoch": 0.00922201275231451,
      "grad_norm": 0.7879970669746399,
      "learning_rate": 3.456791927410606e-05,
      "loss": 2.239,
      "step": 960
    },
    {
      "epoch": 0.00923161901559817,
      "grad_norm": 0.5917340517044067,
      "learning_rate": 3.447409840629156e-05,
      "loss": 1.8615,
      "step": 961
    },
    {
      "epoch": 0.009241225278881831,
      "grad_norm": 0.8712078332901001,
      "learning_rate": 3.438033799877833e-05,
      "loss": 1.6709,
      "step": 962
    },
    {
      "epoch": 0.009250831542165492,
      "grad_norm": 0.7379903197288513,
      "learning_rate": 3.428663841668412e-05,
      "loss": 0.7869,
      "step": 963
    },
    {
      "epoch": 0.009260437805449153,
      "grad_norm": 1.1096373796463013,
      "learning_rate": 3.419300002488984e-05,
      "loss": 0.7667,
      "step": 964
    },
    {
      "epoch": 0.009270044068732814,
      "grad_norm": 0.8927579522132874,
      "learning_rate": 3.409942318803809e-05,
      "loss": 0.9665,
      "step": 965
    },
    {
      "epoch": 0.009279650332016475,
      "grad_norm": 1.2291507720947266,
      "learning_rate": 3.4005908270531794e-05,
      "loss": 0.3514,
      "step": 966
    },
    {
      "epoch": 0.009289256595300137,
      "grad_norm": 0.9622271060943604,
      "learning_rate": 3.391245563653276e-05,
      "loss": 1.7473,
      "step": 967
    },
    {
      "epoch": 0.009298862858583796,
      "grad_norm": 1.3838717937469482,
      "learning_rate": 3.381906564996019e-05,
      "loss": 1.3694,
      "step": 968
    },
    {
      "epoch": 0.009308469121867457,
      "grad_norm": 0.812837541103363,
      "learning_rate": 3.3725738674489414e-05,
      "loss": 1.3865,
      "step": 969
    },
    {
      "epoch": 0.009318075385151118,
      "grad_norm": 1.1338518857955933,
      "learning_rate": 3.3632475073550316e-05,
      "loss": 0.9999,
      "step": 970
    },
    {
      "epoch": 0.009327681648434779,
      "grad_norm": 0.9546958208084106,
      "learning_rate": 3.3539275210326044e-05,
      "loss": 1.4315,
      "step": 971
    },
    {
      "epoch": 0.00933728791171844,
      "grad_norm": 0.8147903680801392,
      "learning_rate": 3.344613944775148e-05,
      "loss": 1.3554,
      "step": 972
    },
    {
      "epoch": 0.009346894175002101,
      "grad_norm": 0.9234085083007812,
      "learning_rate": 3.335306814851196e-05,
      "loss": 0.8272,
      "step": 973
    },
    {
      "epoch": 0.009356500438285762,
      "grad_norm": 1.1460888385772705,
      "learning_rate": 3.326006167504171e-05,
      "loss": 0.6755,
      "step": 974
    },
    {
      "epoch": 0.009366106701569423,
      "grad_norm": 1.057664155960083,
      "learning_rate": 3.31671203895226e-05,
      "loss": 1.5541,
      "step": 975
    },
    {
      "epoch": 0.009375712964853084,
      "grad_norm": 1.058992862701416,
      "learning_rate": 3.307424465388257e-05,
      "loss": 1.5063,
      "step": 976
    },
    {
      "epoch": 0.009385319228136745,
      "grad_norm": 1.2181916236877441,
      "learning_rate": 3.298143482979436e-05,
      "loss": 1.4608,
      "step": 977
    },
    {
      "epoch": 0.009394925491420406,
      "grad_norm": 1.1339645385742188,
      "learning_rate": 3.2888691278674e-05,
      "loss": 1.5576,
      "step": 978
    },
    {
      "epoch": 0.009404531754704067,
      "grad_norm": 0.9855846762657166,
      "learning_rate": 3.2796014361679464e-05,
      "loss": 0.7722,
      "step": 979
    },
    {
      "epoch": 0.009414138017987728,
      "grad_norm": 0.9740868806838989,
      "learning_rate": 3.270340443970923e-05,
      "loss": 1.0831,
      "step": 980
    },
    {
      "epoch": 0.00942374428127139,
      "grad_norm": 1.1198534965515137,
      "learning_rate": 3.261086187340088e-05,
      "loss": 1.2708,
      "step": 981
    },
    {
      "epoch": 0.00943335054455505,
      "grad_norm": 1.076177954673767,
      "learning_rate": 3.251838702312974e-05,
      "loss": 0.479,
      "step": 982
    },
    {
      "epoch": 0.009442956807838711,
      "grad_norm": 1.3318387269973755,
      "learning_rate": 3.242598024900738e-05,
      "loss": 0.8363,
      "step": 983
    },
    {
      "epoch": 0.009452563071122373,
      "grad_norm": 1.1832640171051025,
      "learning_rate": 3.233364191088034e-05,
      "loss": 1.3174,
      "step": 984
    },
    {
      "epoch": 0.009462169334406034,
      "grad_norm": 1.369909405708313,
      "learning_rate": 3.224137236832859e-05,
      "loss": 0.7257,
      "step": 985
    },
    {
      "epoch": 0.009471775597689693,
      "grad_norm": 1.5975923538208008,
      "learning_rate": 3.214917198066423e-05,
      "loss": 1.0138,
      "step": 986
    },
    {
      "epoch": 0.009481381860973354,
      "grad_norm": 1.3214428424835205,
      "learning_rate": 3.2057041106930104e-05,
      "loss": 1.2329,
      "step": 987
    },
    {
      "epoch": 0.009490988124257015,
      "grad_norm": 1.8858808279037476,
      "learning_rate": 3.196498010589828e-05,
      "loss": 1.6627,
      "step": 988
    },
    {
      "epoch": 0.009500594387540676,
      "grad_norm": 1.3149359226226807,
      "learning_rate": 3.187298933606878e-05,
      "loss": 0.9466,
      "step": 989
    },
    {
      "epoch": 0.009510200650824337,
      "grad_norm": 1.6815400123596191,
      "learning_rate": 3.178106915566811e-05,
      "loss": 2.0278,
      "step": 990
    },
    {
      "epoch": 0.009519806914107998,
      "grad_norm": 1.6646859645843506,
      "learning_rate": 3.1689219922647924e-05,
      "loss": 1.0195,
      "step": 991
    },
    {
      "epoch": 0.009529413177391659,
      "grad_norm": 1.338820457458496,
      "learning_rate": 3.159744199468354e-05,
      "loss": 1.9606,
      "step": 992
    },
    {
      "epoch": 0.00953901944067532,
      "grad_norm": 1.9522191286087036,
      "learning_rate": 3.150573572917267e-05,
      "loss": 2.5081,
      "step": 993
    },
    {
      "epoch": 0.009548625703958981,
      "grad_norm": 1.9978740215301514,
      "learning_rate": 3.1414101483233906e-05,
      "loss": 2.0999,
      "step": 994
    },
    {
      "epoch": 0.009558231967242642,
      "grad_norm": 1.9656665325164795,
      "learning_rate": 3.13225396137054e-05,
      "loss": 1.7777,
      "step": 995
    },
    {
      "epoch": 0.009567838230526303,
      "grad_norm": 2.2645909786224365,
      "learning_rate": 3.123105047714347e-05,
      "loss": 1.7336,
      "step": 996
    },
    {
      "epoch": 0.009577444493809964,
      "grad_norm": 2.343191146850586,
      "learning_rate": 3.11396344298212e-05,
      "loss": 1.5694,
      "step": 997
    },
    {
      "epoch": 0.009587050757093625,
      "grad_norm": 3.799161195755005,
      "learning_rate": 3.1048291827727055e-05,
      "loss": 2.3011,
      "step": 998
    },
    {
      "epoch": 0.009596657020377286,
      "grad_norm": 3.1082608699798584,
      "learning_rate": 3.095702302656347e-05,
      "loss": 1.4972,
      "step": 999
    },
    {
      "epoch": 0.009606263283660947,
      "grad_norm": 2.619208574295044,
      "learning_rate": 3.086582838174551e-05,
      "loss": 2.8263,
      "step": 1000
    },
    {
      "epoch": 0.009615869546944609,
      "grad_norm": 0.2590996026992798,
      "learning_rate": 3.077470824839947e-05,
      "loss": 1.8199,
      "step": 1001
    },
    {
      "epoch": 0.00962547581022827,
      "grad_norm": 0.2559146583080292,
      "learning_rate": 3.068366298136144e-05,
      "loss": 1.8772,
      "step": 1002
    },
    {
      "epoch": 0.00963508207351193,
      "grad_norm": 0.44040748476982117,
      "learning_rate": 3.059269293517603e-05,
      "loss": 1.9291,
      "step": 1003
    },
    {
      "epoch": 0.00964468833679559,
      "grad_norm": 0.49574318528175354,
      "learning_rate": 3.050179846409488e-05,
      "loss": 2.1069,
      "step": 1004
    },
    {
      "epoch": 0.009654294600079251,
      "grad_norm": 0.7528969049453735,
      "learning_rate": 3.0410979922075343e-05,
      "loss": 1.6598,
      "step": 1005
    },
    {
      "epoch": 0.009663900863362912,
      "grad_norm": 0.5101423859596252,
      "learning_rate": 3.0320237662779116e-05,
      "loss": 2.4249,
      "step": 1006
    },
    {
      "epoch": 0.009673507126646573,
      "grad_norm": 0.8318787813186646,
      "learning_rate": 3.022957203957083e-05,
      "loss": 1.9748,
      "step": 1007
    },
    {
      "epoch": 0.009683113389930234,
      "grad_norm": 0.582283079624176,
      "learning_rate": 3.013898340551665e-05,
      "loss": 1.8147,
      "step": 1008
    },
    {
      "epoch": 0.009692719653213895,
      "grad_norm": 0.6493712663650513,
      "learning_rate": 3.004847211338295e-05,
      "loss": 1.635,
      "step": 1009
    },
    {
      "epoch": 0.009702325916497556,
      "grad_norm": 0.7176229357719421,
      "learning_rate": 2.9958038515634963e-05,
      "loss": 1.0172,
      "step": 1010
    },
    {
      "epoch": 0.009711932179781217,
      "grad_norm": 1.1313773393630981,
      "learning_rate": 2.9867682964435294e-05,
      "loss": 1.5006,
      "step": 1011
    },
    {
      "epoch": 0.009721538443064878,
      "grad_norm": 1.1590285301208496,
      "learning_rate": 2.977740581164269e-05,
      "loss": 0.9992,
      "step": 1012
    },
    {
      "epoch": 0.00973114470634854,
      "grad_norm": 0.8442860245704651,
      "learning_rate": 2.9687207408810557e-05,
      "loss": 1.5836,
      "step": 1013
    },
    {
      "epoch": 0.0097407509696322,
      "grad_norm": 1.013481855392456,
      "learning_rate": 2.9597088107185672e-05,
      "loss": 1.0372,
      "step": 1014
    },
    {
      "epoch": 0.009750357232915861,
      "grad_norm": 0.8225231766700745,
      "learning_rate": 2.9507048257706727e-05,
      "loss": 1.5832,
      "step": 1015
    },
    {
      "epoch": 0.009759963496199522,
      "grad_norm": 1.6369093656539917,
      "learning_rate": 2.9417088211003087e-05,
      "loss": 1.1977,
      "step": 1016
    },
    {
      "epoch": 0.009769569759483183,
      "grad_norm": 1.0204166173934937,
      "learning_rate": 2.9327208317393303e-05,
      "loss": 1.2537,
      "step": 1017
    },
    {
      "epoch": 0.009779176022766845,
      "grad_norm": 1.0008682012557983,
      "learning_rate": 2.923740892688379e-05,
      "loss": 0.8269,
      "step": 1018
    },
    {
      "epoch": 0.009788782286050506,
      "grad_norm": 0.7066851258277893,
      "learning_rate": 2.9147690389167514e-05,
      "loss": 0.9974,
      "step": 1019
    },
    {
      "epoch": 0.009798388549334167,
      "grad_norm": 1.306208610534668,
      "learning_rate": 2.905805305362257e-05,
      "loss": 1.5424,
      "step": 1020
    },
    {
      "epoch": 0.009807994812617828,
      "grad_norm": 1.0213159322738647,
      "learning_rate": 2.8968497269310803e-05,
      "loss": 1.5015,
      "step": 1021
    },
    {
      "epoch": 0.009817601075901487,
      "grad_norm": 1.1977981328964233,
      "learning_rate": 2.8879023384976517e-05,
      "loss": 1.5252,
      "step": 1022
    },
    {
      "epoch": 0.009827207339185148,
      "grad_norm": 0.8805446624755859,
      "learning_rate": 2.8789631749045097e-05,
      "loss": 1.0103,
      "step": 1023
    },
    {
      "epoch": 0.009836813602468809,
      "grad_norm": 0.9490370750427246,
      "learning_rate": 2.8700322709621635e-05,
      "loss": 1.4349,
      "step": 1024
    },
    {
      "epoch": 0.00984641986575247,
      "grad_norm": 1.1463714838027954,
      "learning_rate": 2.8611096614489518e-05,
      "loss": 1.699,
      "step": 1025
    },
    {
      "epoch": 0.009856026129036131,
      "grad_norm": 1.0782697200775146,
      "learning_rate": 2.8521953811109237e-05,
      "loss": 0.7948,
      "step": 1026
    },
    {
      "epoch": 0.009865632392319792,
      "grad_norm": 1.1975711584091187,
      "learning_rate": 2.8432894646616885e-05,
      "loss": 1.7773,
      "step": 1027
    },
    {
      "epoch": 0.009875238655603453,
      "grad_norm": 1.154541254043579,
      "learning_rate": 2.834391946782281e-05,
      "loss": 1.1594,
      "step": 1028
    },
    {
      "epoch": 0.009884844918887114,
      "grad_norm": 1.5485776662826538,
      "learning_rate": 2.8255028621210355e-05,
      "loss": 1.8537,
      "step": 1029
    },
    {
      "epoch": 0.009894451182170775,
      "grad_norm": 0.917025089263916,
      "learning_rate": 2.8166222452934455e-05,
      "loss": 0.5871,
      "step": 1030
    },
    {
      "epoch": 0.009904057445454436,
      "grad_norm": 1.6280449628829956,
      "learning_rate": 2.8077501308820308e-05,
      "loss": 0.6936,
      "step": 1031
    },
    {
      "epoch": 0.009913663708738097,
      "grad_norm": 1.1050609350204468,
      "learning_rate": 2.7988865534361953e-05,
      "loss": 1.7737,
      "step": 1032
    },
    {
      "epoch": 0.009923269972021758,
      "grad_norm": 1.8401347398757935,
      "learning_rate": 2.790031547472105e-05,
      "loss": 1.9293,
      "step": 1033
    },
    {
      "epoch": 0.00993287623530542,
      "grad_norm": 1.3945201635360718,
      "learning_rate": 2.7811851474725453e-05,
      "loss": 1.2302,
      "step": 1034
    },
    {
      "epoch": 0.00994248249858908,
      "grad_norm": 1.2234818935394287,
      "learning_rate": 2.7723473878867877e-05,
      "loss": 1.1554,
      "step": 1035
    },
    {
      "epoch": 0.009952088761872742,
      "grad_norm": 1.6884944438934326,
      "learning_rate": 2.763518303130458e-05,
      "loss": 0.7462,
      "step": 1036
    },
    {
      "epoch": 0.009961695025156403,
      "grad_norm": 1.3270468711853027,
      "learning_rate": 2.754697927585399e-05,
      "loss": 1.4613,
      "step": 1037
    },
    {
      "epoch": 0.009971301288440064,
      "grad_norm": 1.3611489534378052,
      "learning_rate": 2.7458862955995423e-05,
      "loss": 1.4007,
      "step": 1038
    },
    {
      "epoch": 0.009980907551723725,
      "grad_norm": 1.178431510925293,
      "learning_rate": 2.737083441486763e-05,
      "loss": 0.7878,
      "step": 1039
    },
    {
      "epoch": 0.009990513815007384,
      "grad_norm": 1.6255234479904175,
      "learning_rate": 2.7282893995267623e-05,
      "loss": 0.5396,
      "step": 1040
    },
    {
      "epoch": 0.010000120078291045,
      "grad_norm": 1.9463783502578735,
      "learning_rate": 2.71950420396492e-05,
      "loss": 1.104,
      "step": 1041
    },
    {
      "epoch": 0.010009726341574706,
      "grad_norm": 1.7179642915725708,
      "learning_rate": 2.7107278890121724e-05,
      "loss": 1.2293,
      "step": 1042
    },
    {
      "epoch": 0.010019332604858367,
      "grad_norm": 1.482115626335144,
      "learning_rate": 2.7019604888448642e-05,
      "loss": 2.0738,
      "step": 1043
    },
    {
      "epoch": 0.010028938868142028,
      "grad_norm": 2.2849106788635254,
      "learning_rate": 2.693202037604632e-05,
      "loss": 1.5818,
      "step": 1044
    },
    {
      "epoch": 0.01003854513142569,
      "grad_norm": 1.7904789447784424,
      "learning_rate": 2.6844525693982613e-05,
      "loss": 2.0463,
      "step": 1045
    },
    {
      "epoch": 0.01004815139470935,
      "grad_norm": 2.5114529132843018,
      "learning_rate": 2.6757121182975575e-05,
      "loss": 2.096,
      "step": 1046
    },
    {
      "epoch": 0.010057757657993011,
      "grad_norm": 1.6390053033828735,
      "learning_rate": 2.666980718339211e-05,
      "loss": 2.1709,
      "step": 1047
    },
    {
      "epoch": 0.010067363921276672,
      "grad_norm": 3.0869405269622803,
      "learning_rate": 2.658258403524665e-05,
      "loss": 2.8105,
      "step": 1048
    },
    {
      "epoch": 0.010076970184560333,
      "grad_norm": 2.684781789779663,
      "learning_rate": 2.6495452078199863e-05,
      "loss": 2.849,
      "step": 1049
    },
    {
      "epoch": 0.010086576447843994,
      "grad_norm": 3.2524240016937256,
      "learning_rate": 2.640841165155725e-05,
      "loss": 3.1607,
      "step": 1050
    },
    {
      "epoch": 0.010096182711127655,
      "grad_norm": 0.2134387493133545,
      "learning_rate": 2.6321463094267934e-05,
      "loss": 1.9414,
      "step": 1051
    },
    {
      "epoch": 0.010105788974411316,
      "grad_norm": 0.2765384316444397,
      "learning_rate": 2.6234606744923262e-05,
      "loss": 2.1654,
      "step": 1052
    },
    {
      "epoch": 0.010115395237694978,
      "grad_norm": 0.30473747849464417,
      "learning_rate": 2.614784294175554e-05,
      "loss": 1.9738,
      "step": 1053
    },
    {
      "epoch": 0.010125001500978639,
      "grad_norm": 0.7567789554595947,
      "learning_rate": 2.6061172022636604e-05,
      "loss": 0.7363,
      "step": 1054
    },
    {
      "epoch": 0.0101346077642623,
      "grad_norm": 0.9735413193702698,
      "learning_rate": 2.597459432507664e-05,
      "loss": 1.1967,
      "step": 1055
    },
    {
      "epoch": 0.01014421402754596,
      "grad_norm": 0.7056718468666077,
      "learning_rate": 2.5888110186222868e-05,
      "loss": 1.2954,
      "step": 1056
    },
    {
      "epoch": 0.010153820290829622,
      "grad_norm": 0.6763874888420105,
      "learning_rate": 2.5801719942858065e-05,
      "loss": 2.2343,
      "step": 1057
    },
    {
      "epoch": 0.010163426554113281,
      "grad_norm": 0.7575677633285522,
      "learning_rate": 2.5715423931399442e-05,
      "loss": 2.2128,
      "step": 1058
    },
    {
      "epoch": 0.010173032817396942,
      "grad_norm": 0.7311117649078369,
      "learning_rate": 2.562922248789722e-05,
      "loss": 0.3624,
      "step": 1059
    },
    {
      "epoch": 0.010182639080680603,
      "grad_norm": 0.8437654972076416,
      "learning_rate": 2.554311594803339e-05,
      "loss": 0.3796,
      "step": 1060
    },
    {
      "epoch": 0.010192245343964264,
      "grad_norm": 1.143589973449707,
      "learning_rate": 2.5457104647120322e-05,
      "loss": 0.7522,
      "step": 1061
    },
    {
      "epoch": 0.010201851607247925,
      "grad_norm": 0.9767776131629944,
      "learning_rate": 2.5371188920099564e-05,
      "loss": 1.5349,
      "step": 1062
    },
    {
      "epoch": 0.010211457870531586,
      "grad_norm": 0.7877697348594666,
      "learning_rate": 2.5285369101540445e-05,
      "loss": 0.7316,
      "step": 1063
    },
    {
      "epoch": 0.010221064133815247,
      "grad_norm": 0.7976076602935791,
      "learning_rate": 2.519964552563887e-05,
      "loss": 1.4591,
      "step": 1064
    },
    {
      "epoch": 0.010230670397098908,
      "grad_norm": 0.9339377880096436,
      "learning_rate": 2.5114018526215844e-05,
      "loss": 1.2164,
      "step": 1065
    },
    {
      "epoch": 0.01024027666038257,
      "grad_norm": 0.9645799994468689,
      "learning_rate": 2.5028488436716418e-05,
      "loss": 0.688,
      "step": 1066
    },
    {
      "epoch": 0.01024988292366623,
      "grad_norm": 0.879385769367218,
      "learning_rate": 2.494305559020822e-05,
      "loss": 1.4033,
      "step": 1067
    },
    {
      "epoch": 0.010259489186949891,
      "grad_norm": 0.6166214942932129,
      "learning_rate": 2.4857720319380128e-05,
      "loss": 0.8206,
      "step": 1068
    },
    {
      "epoch": 0.010269095450233552,
      "grad_norm": 0.8537705540657043,
      "learning_rate": 2.4772482956541132e-05,
      "loss": 1.405,
      "step": 1069
    },
    {
      "epoch": 0.010278701713517214,
      "grad_norm": 1.4637665748596191,
      "learning_rate": 2.4687343833618908e-05,
      "loss": 1.6084,
      "step": 1070
    },
    {
      "epoch": 0.010288307976800875,
      "grad_norm": 1.2369686365127563,
      "learning_rate": 2.4602303282158616e-05,
      "loss": 1.5067,
      "step": 1071
    },
    {
      "epoch": 0.010297914240084536,
      "grad_norm": 1.106581449508667,
      "learning_rate": 2.451736163332149e-05,
      "loss": 1.5697,
      "step": 1072
    },
    {
      "epoch": 0.010307520503368197,
      "grad_norm": 1.1647673845291138,
      "learning_rate": 2.4432519217883676e-05,
      "loss": 1.353,
      "step": 1073
    },
    {
      "epoch": 0.010317126766651858,
      "grad_norm": 0.8058618307113647,
      "learning_rate": 2.4347776366234876e-05,
      "loss": 1.0683,
      "step": 1074
    },
    {
      "epoch": 0.010326733029935519,
      "grad_norm": 1.0358927249908447,
      "learning_rate": 2.4263133408377076e-05,
      "loss": 0.7075,
      "step": 1075
    },
    {
      "epoch": 0.010336339293219178,
      "grad_norm": 0.9629650115966797,
      "learning_rate": 2.4178590673923264e-05,
      "loss": 1.4257,
      "step": 1076
    },
    {
      "epoch": 0.010345945556502839,
      "grad_norm": 1.0257420539855957,
      "learning_rate": 2.4094148492096125e-05,
      "loss": 0.4294,
      "step": 1077
    },
    {
      "epoch": 0.0103555518197865,
      "grad_norm": 1.0663120746612549,
      "learning_rate": 2.4009807191726813e-05,
      "loss": 1.9218,
      "step": 1078
    },
    {
      "epoch": 0.010365158083070161,
      "grad_norm": 1.414450764656067,
      "learning_rate": 2.3925567101253576e-05,
      "loss": 0.7072,
      "step": 1079
    },
    {
      "epoch": 0.010374764346353822,
      "grad_norm": 1.1448527574539185,
      "learning_rate": 2.3841428548720586e-05,
      "loss": 0.4485,
      "step": 1080
    },
    {
      "epoch": 0.010384370609637483,
      "grad_norm": 0.9150617718696594,
      "learning_rate": 2.3757391861776585e-05,
      "loss": 0.7941,
      "step": 1081
    },
    {
      "epoch": 0.010393976872921144,
      "grad_norm": 1.2475779056549072,
      "learning_rate": 2.3673457367673664e-05,
      "loss": 1.2234,
      "step": 1082
    },
    {
      "epoch": 0.010403583136204805,
      "grad_norm": 1.059650182723999,
      "learning_rate": 2.3589625393265895e-05,
      "loss": 0.8523,
      "step": 1083
    },
    {
      "epoch": 0.010413189399488466,
      "grad_norm": 1.0351049900054932,
      "learning_rate": 2.3505896265008193e-05,
      "loss": 0.6191,
      "step": 1084
    },
    {
      "epoch": 0.010422795662772127,
      "grad_norm": 1.3045110702514648,
      "learning_rate": 2.3422270308954934e-05,
      "loss": 0.644,
      "step": 1085
    },
    {
      "epoch": 0.010432401926055788,
      "grad_norm": 1.0564193725585938,
      "learning_rate": 2.333874785075874e-05,
      "loss": 0.657,
      "step": 1086
    },
    {
      "epoch": 0.01044200818933945,
      "grad_norm": 1.9321848154067993,
      "learning_rate": 2.3255329215669185e-05,
      "loss": 1.655,
      "step": 1087
    },
    {
      "epoch": 0.01045161445262311,
      "grad_norm": 1.5764094591140747,
      "learning_rate": 2.317201472853156e-05,
      "loss": 0.5873,
      "step": 1088
    },
    {
      "epoch": 0.010461220715906772,
      "grad_norm": 1.291933536529541,
      "learning_rate": 2.3088804713785584e-05,
      "loss": 1.2509,
      "step": 1089
    },
    {
      "epoch": 0.010470826979190433,
      "grad_norm": 2.386507511138916,
      "learning_rate": 2.3005699495464095e-05,
      "loss": 1.8315,
      "step": 1090
    },
    {
      "epoch": 0.010480433242474094,
      "grad_norm": 1.3089796304702759,
      "learning_rate": 2.2922699397191893e-05,
      "loss": 1.5758,
      "step": 1091
    },
    {
      "epoch": 0.010490039505757755,
      "grad_norm": 2.028481960296631,
      "learning_rate": 2.2839804742184416e-05,
      "loss": 1.8212,
      "step": 1092
    },
    {
      "epoch": 0.010499645769041416,
      "grad_norm": 1.8130245208740234,
      "learning_rate": 2.2757015853246493e-05,
      "loss": 2.0159,
      "step": 1093
    },
    {
      "epoch": 0.010509252032325075,
      "grad_norm": 2.3461854457855225,
      "learning_rate": 2.2674333052771035e-05,
      "loss": 1.7399,
      "step": 1094
    },
    {
      "epoch": 0.010518858295608736,
      "grad_norm": 1.5513070821762085,
      "learning_rate": 2.2591756662737862e-05,
      "loss": 1.9366,
      "step": 1095
    },
    {
      "epoch": 0.010528464558892397,
      "grad_norm": 4.964890003204346,
      "learning_rate": 2.2509287004712477e-05,
      "loss": 1.7922,
      "step": 1096
    },
    {
      "epoch": 0.010538070822176058,
      "grad_norm": 2.3452582359313965,
      "learning_rate": 2.242692439984463e-05,
      "loss": 1.8275,
      "step": 1097
    },
    {
      "epoch": 0.01054767708545972,
      "grad_norm": 2.8225247859954834,
      "learning_rate": 2.234466916886727e-05,
      "loss": 3.046,
      "step": 1098
    },
    {
      "epoch": 0.01055728334874338,
      "grad_norm": 4.457972526550293,
      "learning_rate": 2.2262521632095203e-05,
      "loss": 2.0443,
      "step": 1099
    },
    {
      "epoch": 0.010566889612027041,
      "grad_norm": 2.7998135089874268,
      "learning_rate": 2.2180482109423856e-05,
      "loss": 2.1427,
      "step": 1100
    },
    {
      "epoch": 0.010576495875310702,
      "grad_norm": 0.23421290516853333,
      "learning_rate": 2.2098550920327998e-05,
      "loss": 1.8788,
      "step": 1101
    },
    {
      "epoch": 0.010586102138594363,
      "grad_norm": 0.3744014501571655,
      "learning_rate": 2.2016728383860575e-05,
      "loss": 2.0246,
      "step": 1102
    },
    {
      "epoch": 0.010595708401878024,
      "grad_norm": 0.6559910178184509,
      "learning_rate": 2.1935014818651405e-05,
      "loss": 1.3571,
      "step": 1103
    },
    {
      "epoch": 0.010605314665161686,
      "grad_norm": 0.7594718337059021,
      "learning_rate": 2.1853410542905968e-05,
      "loss": 1.5635,
      "step": 1104
    },
    {
      "epoch": 0.010614920928445347,
      "grad_norm": 0.6861061453819275,
      "learning_rate": 2.177191587440409e-05,
      "loss": 2.112,
      "step": 1105
    },
    {
      "epoch": 0.010624527191729008,
      "grad_norm": 2.1050045490264893,
      "learning_rate": 2.169053113049888e-05,
      "loss": 0.6081,
      "step": 1106
    },
    {
      "epoch": 0.010634133455012669,
      "grad_norm": 1.1872990131378174,
      "learning_rate": 2.1609256628115316e-05,
      "loss": 0.8834,
      "step": 1107
    },
    {
      "epoch": 0.01064373971829633,
      "grad_norm": 0.7430525422096252,
      "learning_rate": 2.152809268374905e-05,
      "loss": 1.6775,
      "step": 1108
    },
    {
      "epoch": 0.01065334598157999,
      "grad_norm": 0.7202814817428589,
      "learning_rate": 2.1447039613465265e-05,
      "loss": 1.3221,
      "step": 1109
    },
    {
      "epoch": 0.010662952244863652,
      "grad_norm": 0.6342955231666565,
      "learning_rate": 2.1366097732897345e-05,
      "loss": 2.0599,
      "step": 1110
    },
    {
      "epoch": 0.010672558508147313,
      "grad_norm": 1.234000325202942,
      "learning_rate": 2.128526735724572e-05,
      "loss": 2.1801,
      "step": 1111
    },
    {
      "epoch": 0.010682164771430974,
      "grad_norm": 0.7882181406021118,
      "learning_rate": 2.120454880127654e-05,
      "loss": 1.5362,
      "step": 1112
    },
    {
      "epoch": 0.010691771034714633,
      "grad_norm": 1.1634896993637085,
      "learning_rate": 2.1123942379320576e-05,
      "loss": 0.9388,
      "step": 1113
    },
    {
      "epoch": 0.010701377297998294,
      "grad_norm": 0.764276385307312,
      "learning_rate": 2.10434484052719e-05,
      "loss": 1.0732,
      "step": 1114
    },
    {
      "epoch": 0.010710983561281955,
      "grad_norm": 1.1624335050582886,
      "learning_rate": 2.096306719258669e-05,
      "loss": 0.1751,
      "step": 1115
    },
    {
      "epoch": 0.010720589824565616,
      "grad_norm": 0.9240299463272095,
      "learning_rate": 2.088279905428204e-05,
      "loss": 1.6425,
      "step": 1116
    },
    {
      "epoch": 0.010730196087849277,
      "grad_norm": 0.8713756203651428,
      "learning_rate": 2.0802644302934683e-05,
      "loss": 1.5494,
      "step": 1117
    },
    {
      "epoch": 0.010739802351132938,
      "grad_norm": 1.4305771589279175,
      "learning_rate": 2.0722603250679847e-05,
      "loss": 1.3601,
      "step": 1118
    },
    {
      "epoch": 0.0107494086144166,
      "grad_norm": 0.8065025210380554,
      "learning_rate": 2.0642676209209934e-05,
      "loss": 1.0655,
      "step": 1119
    },
    {
      "epoch": 0.01075901487770026,
      "grad_norm": 1.0280262231826782,
      "learning_rate": 2.0562863489773415e-05,
      "loss": 0.9791,
      "step": 1120
    },
    {
      "epoch": 0.010768621140983922,
      "grad_norm": 0.8497940897941589,
      "learning_rate": 2.0483165403173583e-05,
      "loss": 0.9443,
      "step": 1121
    },
    {
      "epoch": 0.010778227404267583,
      "grad_norm": 0.9765475988388062,
      "learning_rate": 2.0403582259767312e-05,
      "loss": 0.7712,
      "step": 1122
    },
    {
      "epoch": 0.010787833667551244,
      "grad_norm": 0.779600203037262,
      "learning_rate": 2.0324114369463855e-05,
      "loss": 0.5175,
      "step": 1123
    },
    {
      "epoch": 0.010797439930834905,
      "grad_norm": 1.2606279850006104,
      "learning_rate": 2.024476204172367e-05,
      "loss": 1.6264,
      "step": 1124
    },
    {
      "epoch": 0.010807046194118566,
      "grad_norm": 0.9941396117210388,
      "learning_rate": 2.0165525585557204e-05,
      "loss": 1.111,
      "step": 1125
    },
    {
      "epoch": 0.010816652457402227,
      "grad_norm": 1.5044811964035034,
      "learning_rate": 2.0086405309523676e-05,
      "loss": 1.3411,
      "step": 1126
    },
    {
      "epoch": 0.010826258720685888,
      "grad_norm": 1.2421293258666992,
      "learning_rate": 2.0007401521729863e-05,
      "loss": 1.9771,
      "step": 1127
    },
    {
      "epoch": 0.010835864983969549,
      "grad_norm": 0.9801291823387146,
      "learning_rate": 1.9928514529828947e-05,
      "loss": 1.1597,
      "step": 1128
    },
    {
      "epoch": 0.01084547124725321,
      "grad_norm": 1.084414005279541,
      "learning_rate": 1.984974464101928e-05,
      "loss": 1.1055,
      "step": 1129
    },
    {
      "epoch": 0.010855077510536871,
      "grad_norm": 1.1225916147232056,
      "learning_rate": 1.9771092162043165e-05,
      "loss": 0.8676,
      "step": 1130
    },
    {
      "epoch": 0.01086468377382053,
      "grad_norm": 1.0418264865875244,
      "learning_rate": 1.9692557399185734e-05,
      "loss": 1.2746,
      "step": 1131
    },
    {
      "epoch": 0.010874290037104191,
      "grad_norm": 1.1687397956848145,
      "learning_rate": 1.9614140658273693e-05,
      "loss": 0.7436,
      "step": 1132
    },
    {
      "epoch": 0.010883896300387852,
      "grad_norm": 1.0795031785964966,
      "learning_rate": 1.953584224467418e-05,
      "loss": 0.8601,
      "step": 1133
    },
    {
      "epoch": 0.010893502563671513,
      "grad_norm": 1.367310643196106,
      "learning_rate": 1.9457662463293484e-05,
      "loss": 1.6607,
      "step": 1134
    },
    {
      "epoch": 0.010903108826955174,
      "grad_norm": 1.0754454135894775,
      "learning_rate": 1.9379601618575977e-05,
      "loss": 0.883,
      "step": 1135
    },
    {
      "epoch": 0.010912715090238835,
      "grad_norm": 1.1320902109146118,
      "learning_rate": 1.9301660014502894e-05,
      "loss": 1.0108,
      "step": 1136
    },
    {
      "epoch": 0.010922321353522496,
      "grad_norm": 1.2725812196731567,
      "learning_rate": 1.9223837954591046e-05,
      "loss": 1.7963,
      "step": 1137
    },
    {
      "epoch": 0.010931927616806158,
      "grad_norm": 1.8460112810134888,
      "learning_rate": 1.914613574189178e-05,
      "loss": 0.8088,
      "step": 1138
    },
    {
      "epoch": 0.010941533880089819,
      "grad_norm": 1.0875414609909058,
      "learning_rate": 1.9068553678989736e-05,
      "loss": 1.222,
      "step": 1139
    },
    {
      "epoch": 0.01095114014337348,
      "grad_norm": 1.4948869943618774,
      "learning_rate": 1.8991092068001675e-05,
      "loss": 1.7867,
      "step": 1140
    },
    {
      "epoch": 0.01096074640665714,
      "grad_norm": 1.8290828466415405,
      "learning_rate": 1.8913751210575248e-05,
      "loss": 1.0475,
      "step": 1141
    },
    {
      "epoch": 0.010970352669940802,
      "grad_norm": 2.107839345932007,
      "learning_rate": 1.883653140788793e-05,
      "loss": 1.1523,
      "step": 1142
    },
    {
      "epoch": 0.010979958933224463,
      "grad_norm": 2.2905049324035645,
      "learning_rate": 1.8759432960645774e-05,
      "loss": 1.7803,
      "step": 1143
    },
    {
      "epoch": 0.010989565196508124,
      "grad_norm": 1.8264007568359375,
      "learning_rate": 1.8682456169082263e-05,
      "loss": 2.4772,
      "step": 1144
    },
    {
      "epoch": 0.010999171459791785,
      "grad_norm": 2.193277359008789,
      "learning_rate": 1.8605601332957077e-05,
      "loss": 2.2549,
      "step": 1145
    },
    {
      "epoch": 0.011008777723075446,
      "grad_norm": 2.192891836166382,
      "learning_rate": 1.852886875155508e-05,
      "loss": 1.8016,
      "step": 1146
    },
    {
      "epoch": 0.011018383986359107,
      "grad_norm": 3.2925832271575928,
      "learning_rate": 1.8452258723684995e-05,
      "loss": 1.4206,
      "step": 1147
    },
    {
      "epoch": 0.011027990249642768,
      "grad_norm": 2.7194485664367676,
      "learning_rate": 1.8375771547678295e-05,
      "loss": 1.9565,
      "step": 1148
    },
    {
      "epoch": 0.011037596512926427,
      "grad_norm": 2.9804797172546387,
      "learning_rate": 1.8299407521388067e-05,
      "loss": 2.0869,
      "step": 1149
    },
    {
      "epoch": 0.011047202776210088,
      "grad_norm": 2.8504514694213867,
      "learning_rate": 1.8223166942187848e-05,
      "loss": 2.4472,
      "step": 1150
    },
    {
      "epoch": 0.01105680903949375,
      "grad_norm": 0.21416255831718445,
      "learning_rate": 1.8147050106970437e-05,
      "loss": 2.0324,
      "step": 1151
    },
    {
      "epoch": 0.01106641530277741,
      "grad_norm": 0.2835964262485504,
      "learning_rate": 1.8071057312146738e-05,
      "loss": 1.9876,
      "step": 1152
    },
    {
      "epoch": 0.011076021566061071,
      "grad_norm": 0.48733270168304443,
      "learning_rate": 1.7995188853644646e-05,
      "loss": 1.6582,
      "step": 1153
    },
    {
      "epoch": 0.011085627829344732,
      "grad_norm": 0.6662628650665283,
      "learning_rate": 1.7919445026907865e-05,
      "loss": 1.8553,
      "step": 1154
    },
    {
      "epoch": 0.011095234092628393,
      "grad_norm": 0.5016735196113586,
      "learning_rate": 1.784382612689477e-05,
      "loss": 2.0963,
      "step": 1155
    },
    {
      "epoch": 0.011104840355912055,
      "grad_norm": 0.48013660311698914,
      "learning_rate": 1.776833244807723e-05,
      "loss": 2.4028,
      "step": 1156
    },
    {
      "epoch": 0.011114446619195716,
      "grad_norm": 0.5376973748207092,
      "learning_rate": 1.7692964284439505e-05,
      "loss": 2.4312,
      "step": 1157
    },
    {
      "epoch": 0.011124052882479377,
      "grad_norm": 0.6370044946670532,
      "learning_rate": 1.7617721929477093e-05,
      "loss": 0.7586,
      "step": 1158
    },
    {
      "epoch": 0.011133659145763038,
      "grad_norm": 0.6221287846565247,
      "learning_rate": 1.7542605676195506e-05,
      "loss": 0.5241,
      "step": 1159
    },
    {
      "epoch": 0.011143265409046699,
      "grad_norm": 1.0271189212799072,
      "learning_rate": 1.746761581710927e-05,
      "loss": 1.4939,
      "step": 1160
    },
    {
      "epoch": 0.01115287167233036,
      "grad_norm": 1.0297560691833496,
      "learning_rate": 1.739275264424067e-05,
      "loss": 1.4919,
      "step": 1161
    },
    {
      "epoch": 0.01116247793561402,
      "grad_norm": 0.779731035232544,
      "learning_rate": 1.7318016449118695e-05,
      "loss": 0.6876,
      "step": 1162
    },
    {
      "epoch": 0.011172084198897682,
      "grad_norm": 0.785140335559845,
      "learning_rate": 1.7243407522777806e-05,
      "loss": 1.26,
      "step": 1163
    },
    {
      "epoch": 0.011181690462181343,
      "grad_norm": 1.7486060857772827,
      "learning_rate": 1.7168926155756893e-05,
      "loss": 0.7225,
      "step": 1164
    },
    {
      "epoch": 0.011191296725465004,
      "grad_norm": 0.7934461236000061,
      "learning_rate": 1.7094572638098123e-05,
      "loss": 1.2661,
      "step": 1165
    },
    {
      "epoch": 0.011200902988748665,
      "grad_norm": 0.7809988856315613,
      "learning_rate": 1.7020347259345763e-05,
      "loss": 0.7069,
      "step": 1166
    },
    {
      "epoch": 0.011210509252032324,
      "grad_norm": 0.9379634857177734,
      "learning_rate": 1.6946250308545125e-05,
      "loss": 0.8051,
      "step": 1167
    },
    {
      "epoch": 0.011220115515315985,
      "grad_norm": 1.0367298126220703,
      "learning_rate": 1.687228207424137e-05,
      "loss": 1.2884,
      "step": 1168
    },
    {
      "epoch": 0.011229721778599646,
      "grad_norm": 0.8595003485679626,
      "learning_rate": 1.6798442844478445e-05,
      "loss": 1.2417,
      "step": 1169
    },
    {
      "epoch": 0.011239328041883307,
      "grad_norm": 0.8204405903816223,
      "learning_rate": 1.672473290679789e-05,
      "loss": 0.5817,
      "step": 1170
    },
    {
      "epoch": 0.011248934305166968,
      "grad_norm": 0.7739542126655579,
      "learning_rate": 1.6651152548237802e-05,
      "loss": 1.3192,
      "step": 1171
    },
    {
      "epoch": 0.01125854056845063,
      "grad_norm": 0.8143975138664246,
      "learning_rate": 1.6577702055331663e-05,
      "loss": 0.9366,
      "step": 1172
    },
    {
      "epoch": 0.01126814683173429,
      "grad_norm": 0.8258153200149536,
      "learning_rate": 1.6504381714107252e-05,
      "loss": 0.7281,
      "step": 1173
    },
    {
      "epoch": 0.011277753095017952,
      "grad_norm": 0.8898383975028992,
      "learning_rate": 1.6431191810085477e-05,
      "loss": 1.2691,
      "step": 1174
    },
    {
      "epoch": 0.011287359358301613,
      "grad_norm": 0.877154529094696,
      "learning_rate": 1.6358132628279322e-05,
      "loss": 1.0645,
      "step": 1175
    },
    {
      "epoch": 0.011296965621585274,
      "grad_norm": 1.1807379722595215,
      "learning_rate": 1.6285204453192777e-05,
      "loss": 1.026,
      "step": 1176
    },
    {
      "epoch": 0.011306571884868935,
      "grad_norm": 1.1935549974441528,
      "learning_rate": 1.6212407568819565e-05,
      "loss": 0.887,
      "step": 1177
    },
    {
      "epoch": 0.011316178148152596,
      "grad_norm": 1.0390746593475342,
      "learning_rate": 1.613974225864222e-05,
      "loss": 1.4029,
      "step": 1178
    },
    {
      "epoch": 0.011325784411436257,
      "grad_norm": 1.556865930557251,
      "learning_rate": 1.6067208805630877e-05,
      "loss": 1.0711,
      "step": 1179
    },
    {
      "epoch": 0.011335390674719918,
      "grad_norm": 1.1142784357070923,
      "learning_rate": 1.599480749224222e-05,
      "loss": 1.3445,
      "step": 1180
    },
    {
      "epoch": 0.011344996938003579,
      "grad_norm": 1.189209222793579,
      "learning_rate": 1.5922538600418318e-05,
      "loss": 1.136,
      "step": 1181
    },
    {
      "epoch": 0.01135460320128724,
      "grad_norm": 1.1040972471237183,
      "learning_rate": 1.5850402411585603e-05,
      "loss": 1.6425,
      "step": 1182
    },
    {
      "epoch": 0.011364209464570901,
      "grad_norm": 1.6171393394470215,
      "learning_rate": 1.5778399206653734e-05,
      "loss": 1.0274,
      "step": 1183
    },
    {
      "epoch": 0.011373815727854562,
      "grad_norm": 1.6282167434692383,
      "learning_rate": 1.5706529266014524e-05,
      "loss": 1.6883,
      "step": 1184
    },
    {
      "epoch": 0.011383421991138221,
      "grad_norm": 1.6567491292953491,
      "learning_rate": 1.563479286954078e-05,
      "loss": 0.9166,
      "step": 1185
    },
    {
      "epoch": 0.011393028254421882,
      "grad_norm": 1.7951308488845825,
      "learning_rate": 1.5563190296585344e-05,
      "loss": 1.2749,
      "step": 1186
    },
    {
      "epoch": 0.011402634517705543,
      "grad_norm": 1.3069489002227783,
      "learning_rate": 1.54917218259799e-05,
      "loss": 2.0707,
      "step": 1187
    },
    {
      "epoch": 0.011412240780989204,
      "grad_norm": 1.0857038497924805,
      "learning_rate": 1.5420387736033864e-05,
      "loss": 2.2489,
      "step": 1188
    },
    {
      "epoch": 0.011421847044272865,
      "grad_norm": 2.5009469985961914,
      "learning_rate": 1.5349188304533413e-05,
      "loss": 1.3615,
      "step": 1189
    },
    {
      "epoch": 0.011431453307556527,
      "grad_norm": 1.8224256038665771,
      "learning_rate": 1.527812380874033e-05,
      "loss": 2.1883,
      "step": 1190
    },
    {
      "epoch": 0.011441059570840188,
      "grad_norm": 1.6372029781341553,
      "learning_rate": 1.5207194525390938e-05,
      "loss": 2.6672,
      "step": 1191
    },
    {
      "epoch": 0.011450665834123849,
      "grad_norm": 2.473048686981201,
      "learning_rate": 1.5136400730695e-05,
      "loss": 0.9348,
      "step": 1192
    },
    {
      "epoch": 0.01146027209740751,
      "grad_norm": 2.3728885650634766,
      "learning_rate": 1.5065742700334678e-05,
      "loss": 1.4092,
      "step": 1193
    },
    {
      "epoch": 0.01146987836069117,
      "grad_norm": 2.189683675765991,
      "learning_rate": 1.4995220709463482e-05,
      "loss": 2.5405,
      "step": 1194
    },
    {
      "epoch": 0.011479484623974832,
      "grad_norm": 2.235046148300171,
      "learning_rate": 1.4924835032705064e-05,
      "loss": 2.9695,
      "step": 1195
    },
    {
      "epoch": 0.011489090887258493,
      "grad_norm": 3.5824530124664307,
      "learning_rate": 1.4854585944152377e-05,
      "loss": 2.9233,
      "step": 1196
    },
    {
      "epoch": 0.011498697150542154,
      "grad_norm": 2.5793144702911377,
      "learning_rate": 1.4784473717366387e-05,
      "loss": 2.1335,
      "step": 1197
    },
    {
      "epoch": 0.011508303413825815,
      "grad_norm": 2.6323935985565186,
      "learning_rate": 1.4714498625375146e-05,
      "loss": 1.8674,
      "step": 1198
    },
    {
      "epoch": 0.011517909677109476,
      "grad_norm": 5.027894973754883,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 2.3273,
      "step": 1199
    },
    {
      "epoch": 0.011527515940393137,
      "grad_norm": 2.37316632270813,
      "learning_rate": 1.4574960935217774e-05,
      "loss": 2.6175,
      "step": 1200
    },
    {
      "epoch": 0.011537122203676798,
      "grad_norm": 0.24304594099521637,
      "learning_rate": 1.4505398880433369e-05,
      "loss": 2.0155,
      "step": 1201
    },
    {
      "epoch": 0.011546728466960459,
      "grad_norm": 0.31746044754981995,
      "learning_rate": 1.4435975047204969e-05,
      "loss": 1.9052,
      "step": 1202
    },
    {
      "epoch": 0.011556334730244118,
      "grad_norm": 0.7279465794563293,
      "learning_rate": 1.4366689705879898e-05,
      "loss": 1.5242,
      "step": 1203
    },
    {
      "epoch": 0.01156594099352778,
      "grad_norm": 0.4245074689388275,
      "learning_rate": 1.4297543126266155e-05,
      "loss": 1.688,
      "step": 1204
    },
    {
      "epoch": 0.01157554725681144,
      "grad_norm": 0.5998055934906006,
      "learning_rate": 1.4228535577631442e-05,
      "loss": 1.7135,
      "step": 1205
    },
    {
      "epoch": 0.011585153520095101,
      "grad_norm": 0.5406530499458313,
      "learning_rate": 1.4159667328701937e-05,
      "loss": 2.2145,
      "step": 1206
    },
    {
      "epoch": 0.011594759783378763,
      "grad_norm": 0.570108950138092,
      "learning_rate": 1.4090938647661461e-05,
      "loss": 1.3736,
      "step": 1207
    },
    {
      "epoch": 0.011604366046662424,
      "grad_norm": 0.6127389669418335,
      "learning_rate": 1.4022349802150303e-05,
      "loss": 1.4309,
      "step": 1208
    },
    {
      "epoch": 0.011613972309946085,
      "grad_norm": 0.7525798678398132,
      "learning_rate": 1.3953901059264191e-05,
      "loss": 1.2158,
      "step": 1209
    },
    {
      "epoch": 0.011623578573229746,
      "grad_norm": 0.8749638199806213,
      "learning_rate": 1.388559268555329e-05,
      "loss": 1.8972,
      "step": 1210
    },
    {
      "epoch": 0.011633184836513407,
      "grad_norm": 0.8091251254081726,
      "learning_rate": 1.3817424947021151e-05,
      "loss": 0.6333,
      "step": 1211
    },
    {
      "epoch": 0.011642791099797068,
      "grad_norm": 1.0810160636901855,
      "learning_rate": 1.3749398109123679e-05,
      "loss": 1.5151,
      "step": 1212
    },
    {
      "epoch": 0.011652397363080729,
      "grad_norm": 0.7448549270629883,
      "learning_rate": 1.3681512436768045e-05,
      "loss": 1.9587,
      "step": 1213
    },
    {
      "epoch": 0.01166200362636439,
      "grad_norm": 0.7804606556892395,
      "learning_rate": 1.361376819431172e-05,
      "loss": 2.0766,
      "step": 1214
    },
    {
      "epoch": 0.01167160988964805,
      "grad_norm": 0.7752441167831421,
      "learning_rate": 1.3546165645561487e-05,
      "loss": 0.5087,
      "step": 1215
    },
    {
      "epoch": 0.011681216152931712,
      "grad_norm": 0.9953719973564148,
      "learning_rate": 1.3478705053772244e-05,
      "loss": 0.6756,
      "step": 1216
    },
    {
      "epoch": 0.011690822416215373,
      "grad_norm": 0.9373853206634521,
      "learning_rate": 1.3411386681646164e-05,
      "loss": 0.6682,
      "step": 1217
    },
    {
      "epoch": 0.011700428679499034,
      "grad_norm": 0.8456129431724548,
      "learning_rate": 1.3344210791331557e-05,
      "loss": 0.6289,
      "step": 1218
    },
    {
      "epoch": 0.011710034942782695,
      "grad_norm": 0.8211795687675476,
      "learning_rate": 1.3277177644421924e-05,
      "loss": 1.0296,
      "step": 1219
    },
    {
      "epoch": 0.011719641206066356,
      "grad_norm": 0.6909956932067871,
      "learning_rate": 1.3210287501954844e-05,
      "loss": 1.3598,
      "step": 1220
    },
    {
      "epoch": 0.011729247469350015,
      "grad_norm": 1.2240545749664307,
      "learning_rate": 1.314354062441106e-05,
      "loss": 1.6009,
      "step": 1221
    },
    {
      "epoch": 0.011738853732633676,
      "grad_norm": 0.7934696674346924,
      "learning_rate": 1.3076937271713402e-05,
      "loss": 0.54,
      "step": 1222
    },
    {
      "epoch": 0.011748459995917337,
      "grad_norm": 0.9403862357139587,
      "learning_rate": 1.301047770322581e-05,
      "loss": 1.7715,
      "step": 1223
    },
    {
      "epoch": 0.011758066259200999,
      "grad_norm": 0.9596639275550842,
      "learning_rate": 1.2944162177752244e-05,
      "loss": 1.4277,
      "step": 1224
    },
    {
      "epoch": 0.01176767252248466,
      "grad_norm": 1.2583380937576294,
      "learning_rate": 1.287799095353584e-05,
      "loss": 1.2259,
      "step": 1225
    },
    {
      "epoch": 0.01177727878576832,
      "grad_norm": 1.407221794128418,
      "learning_rate": 1.2811964288257733e-05,
      "loss": 0.9312,
      "step": 1226
    },
    {
      "epoch": 0.011786885049051982,
      "grad_norm": 1.2157663106918335,
      "learning_rate": 1.2746082439036117e-05,
      "loss": 0.8866,
      "step": 1227
    },
    {
      "epoch": 0.011796491312335643,
      "grad_norm": 0.9835970401763916,
      "learning_rate": 1.268034566242527e-05,
      "loss": 1.1338,
      "step": 1228
    },
    {
      "epoch": 0.011806097575619304,
      "grad_norm": 1.4043480157852173,
      "learning_rate": 1.2614754214414548e-05,
      "loss": 1.3643,
      "step": 1229
    },
    {
      "epoch": 0.011815703838902965,
      "grad_norm": 1.0805062055587769,
      "learning_rate": 1.2549308350427363e-05,
      "loss": 0.7288,
      "step": 1230
    },
    {
      "epoch": 0.011825310102186626,
      "grad_norm": 1.1290277242660522,
      "learning_rate": 1.2484008325320174e-05,
      "loss": 1.1034,
      "step": 1231
    },
    {
      "epoch": 0.011834916365470287,
      "grad_norm": 1.1785166263580322,
      "learning_rate": 1.2418854393381552e-05,
      "loss": 0.5611,
      "step": 1232
    },
    {
      "epoch": 0.011844522628753948,
      "grad_norm": 1.630568027496338,
      "learning_rate": 1.2353846808331154e-05,
      "loss": 1.6643,
      "step": 1233
    },
    {
      "epoch": 0.011854128892037609,
      "grad_norm": 1.3291112184524536,
      "learning_rate": 1.2288985823318733e-05,
      "loss": 1.7026,
      "step": 1234
    },
    {
      "epoch": 0.01186373515532127,
      "grad_norm": 1.0357156991958618,
      "learning_rate": 1.2224271690923155e-05,
      "loss": 0.7482,
      "step": 1235
    },
    {
      "epoch": 0.011873341418604931,
      "grad_norm": 1.8691824674606323,
      "learning_rate": 1.2159704663151433e-05,
      "loss": 1.8483,
      "step": 1236
    },
    {
      "epoch": 0.011882947681888592,
      "grad_norm": 1.6788218021392822,
      "learning_rate": 1.2095284991437733e-05,
      "loss": 1.9115,
      "step": 1237
    },
    {
      "epoch": 0.011892553945172253,
      "grad_norm": 1.2689547538757324,
      "learning_rate": 1.2031012926642354e-05,
      "loss": 1.2755,
      "step": 1238
    },
    {
      "epoch": 0.011902160208455912,
      "grad_norm": 1.5324326753616333,
      "learning_rate": 1.1966888719050829e-05,
      "loss": 2.3364,
      "step": 1239
    },
    {
      "epoch": 0.011911766471739573,
      "grad_norm": 1.3587154150009155,
      "learning_rate": 1.1902912618372902e-05,
      "loss": 1.7077,
      "step": 1240
    },
    {
      "epoch": 0.011921372735023235,
      "grad_norm": 1.7131679058074951,
      "learning_rate": 1.1839084873741584e-05,
      "loss": 1.4707,
      "step": 1241
    },
    {
      "epoch": 0.011930978998306896,
      "grad_norm": 2.2139065265655518,
      "learning_rate": 1.1775405733712119e-05,
      "loss": 1.5948,
      "step": 1242
    },
    {
      "epoch": 0.011940585261590557,
      "grad_norm": 1.5664963722229004,
      "learning_rate": 1.1711875446261094e-05,
      "loss": 1.8955,
      "step": 1243
    },
    {
      "epoch": 0.011950191524874218,
      "grad_norm": 2.0715324878692627,
      "learning_rate": 1.1648494258785442e-05,
      "loss": 1.051,
      "step": 1244
    },
    {
      "epoch": 0.011959797788157879,
      "grad_norm": 1.7170823812484741,
      "learning_rate": 1.1585262418101467e-05,
      "loss": 1.731,
      "step": 1245
    },
    {
      "epoch": 0.01196940405144154,
      "grad_norm": 2.7178666591644287,
      "learning_rate": 1.1522180170443891e-05,
      "loss": 1.6746,
      "step": 1246
    },
    {
      "epoch": 0.0119790103147252,
      "grad_norm": 1.7841171026229858,
      "learning_rate": 1.1459247761464909e-05,
      "loss": 1.7435,
      "step": 1247
    },
    {
      "epoch": 0.011988616578008862,
      "grad_norm": 2.7875335216522217,
      "learning_rate": 1.1396465436233228e-05,
      "loss": 1.5466,
      "step": 1248
    },
    {
      "epoch": 0.011998222841292523,
      "grad_norm": 3.184530258178711,
      "learning_rate": 1.1333833439233055e-05,
      "loss": 2.6222,
      "step": 1249
    },
    {
      "epoch": 0.012007829104576184,
      "grad_norm": 3.3153328895568848,
      "learning_rate": 1.1271352014363252e-05,
      "loss": 2.2373,
      "step": 1250
    },
    {
      "epoch": 0.012017435367859845,
      "grad_norm": 0.23886770009994507,
      "learning_rate": 1.1209021404936304e-05,
      "loss": 1.9003,
      "step": 1251
    },
    {
      "epoch": 0.012027041631143506,
      "grad_norm": 0.4039011001586914,
      "learning_rate": 1.1146841853677414e-05,
      "loss": 2.151,
      "step": 1252
    },
    {
      "epoch": 0.012036647894427167,
      "grad_norm": 0.5774843692779541,
      "learning_rate": 1.1084813602723515e-05,
      "loss": 2.0925,
      "step": 1253
    },
    {
      "epoch": 0.012046254157710828,
      "grad_norm": 0.4592426121234894,
      "learning_rate": 1.1022936893622349e-05,
      "loss": 1.7455,
      "step": 1254
    },
    {
      "epoch": 0.012055860420994489,
      "grad_norm": 0.5096227526664734,
      "learning_rate": 1.0961211967331597e-05,
      "loss": 2.169,
      "step": 1255
    },
    {
      "epoch": 0.01206546668427815,
      "grad_norm": 0.46232953667640686,
      "learning_rate": 1.089963906421778e-05,
      "loss": 1.8871,
      "step": 1256
    },
    {
      "epoch": 0.01207507294756181,
      "grad_norm": 0.9018414616584778,
      "learning_rate": 1.083821842405548e-05,
      "loss": 1.5008,
      "step": 1257
    },
    {
      "epoch": 0.01208467921084547,
      "grad_norm": 0.7207086682319641,
      "learning_rate": 1.0776950286026321e-05,
      "loss": 0.9103,
      "step": 1258
    },
    {
      "epoch": 0.012094285474129132,
      "grad_norm": 0.5234125256538391,
      "learning_rate": 1.0715834888718074e-05,
      "loss": 2.1002,
      "step": 1259
    },
    {
      "epoch": 0.012103891737412793,
      "grad_norm": 0.8336639404296875,
      "learning_rate": 1.0654872470123678e-05,
      "loss": 1.7906,
      "step": 1260
    },
    {
      "epoch": 0.012113498000696454,
      "grad_norm": 1.0882858037948608,
      "learning_rate": 1.0594063267640386e-05,
      "loss": 0.8547,
      "step": 1261
    },
    {
      "epoch": 0.012123104263980115,
      "grad_norm": 0.8288137912750244,
      "learning_rate": 1.0533407518068794e-05,
      "loss": 0.3988,
      "step": 1262
    },
    {
      "epoch": 0.012132710527263776,
      "grad_norm": 0.7192359566688538,
      "learning_rate": 1.0472905457611936e-05,
      "loss": 1.0409,
      "step": 1263
    },
    {
      "epoch": 0.012142316790547437,
      "grad_norm": 1.1397526264190674,
      "learning_rate": 1.0412557321874305e-05,
      "loss": 1.1109,
      "step": 1264
    },
    {
      "epoch": 0.012151923053831098,
      "grad_norm": 0.986792266368866,
      "learning_rate": 1.0352363345861065e-05,
      "loss": 0.6528,
      "step": 1265
    },
    {
      "epoch": 0.012161529317114759,
      "grad_norm": 0.8717612028121948,
      "learning_rate": 1.0292323763977024e-05,
      "loss": 0.8045,
      "step": 1266
    },
    {
      "epoch": 0.01217113558039842,
      "grad_norm": 0.6383687853813171,
      "learning_rate": 1.023243881002573e-05,
      "loss": 0.8346,
      "step": 1267
    },
    {
      "epoch": 0.012180741843682081,
      "grad_norm": 0.8243313431739807,
      "learning_rate": 1.0172708717208607e-05,
      "loss": 1.6805,
      "step": 1268
    },
    {
      "epoch": 0.012190348106965742,
      "grad_norm": 1.1205098628997803,
      "learning_rate": 1.0113133718124035e-05,
      "loss": 1.5009,
      "step": 1269
    },
    {
      "epoch": 0.012199954370249403,
      "grad_norm": 0.9703836441040039,
      "learning_rate": 1.0053714044766433e-05,
      "loss": 1.9895,
      "step": 1270
    },
    {
      "epoch": 0.012209560633533064,
      "grad_norm": 1.0298919677734375,
      "learning_rate": 9.994449928525324e-06,
      "loss": 0.6098,
      "step": 1271
    },
    {
      "epoch": 0.012219166896816725,
      "grad_norm": 0.8976793885231018,
      "learning_rate": 9.93534160018451e-06,
      "loss": 1.1091,
      "step": 1272
    },
    {
      "epoch": 0.012228773160100386,
      "grad_norm": 1.0641522407531738,
      "learning_rate": 9.876389289921106e-06,
      "loss": 0.9322,
      "step": 1273
    },
    {
      "epoch": 0.012238379423384047,
      "grad_norm": 1.1619539260864258,
      "learning_rate": 9.817593227304678e-06,
      "loss": 1.214,
      "step": 1274
    },
    {
      "epoch": 0.012247985686667706,
      "grad_norm": 0.9666778445243835,
      "learning_rate": 9.758953641296331e-06,
      "loss": 0.8025,
      "step": 1275
    },
    {
      "epoch": 0.012257591949951368,
      "grad_norm": 1.0337514877319336,
      "learning_rate": 9.700470760247838e-06,
      "loss": 2.2822,
      "step": 1276
    },
    {
      "epoch": 0.012267198213235029,
      "grad_norm": 0.9917118549346924,
      "learning_rate": 9.642144811900739e-06,
      "loss": 1.1873,
      "step": 1277
    },
    {
      "epoch": 0.01227680447651869,
      "grad_norm": 1.3022019863128662,
      "learning_rate": 9.583976023385422e-06,
      "loss": 0.8498,
      "step": 1278
    },
    {
      "epoch": 0.01228641073980235,
      "grad_norm": 1.0588277578353882,
      "learning_rate": 9.5259646212203e-06,
      "loss": 1.1031,
      "step": 1279
    },
    {
      "epoch": 0.012296017003086012,
      "grad_norm": 0.894415020942688,
      "learning_rate": 9.468110831310895e-06,
      "loss": 0.6799,
      "step": 1280
    },
    {
      "epoch": 0.012305623266369673,
      "grad_norm": 1.7153661251068115,
      "learning_rate": 9.410414878948975e-06,
      "loss": 1.905,
      "step": 1281
    },
    {
      "epoch": 0.012315229529653334,
      "grad_norm": 1.240562081336975,
      "learning_rate": 9.352876988811615e-06,
      "loss": 2.0129,
      "step": 1282
    },
    {
      "epoch": 0.012324835792936995,
      "grad_norm": 1.2109215259552002,
      "learning_rate": 9.295497384960416e-06,
      "loss": 1.5702,
      "step": 1283
    },
    {
      "epoch": 0.012334442056220656,
      "grad_norm": 1.1345182657241821,
      "learning_rate": 9.238276290840586e-06,
      "loss": 0.7934,
      "step": 1284
    },
    {
      "epoch": 0.012344048319504317,
      "grad_norm": 1.5537837743759155,
      "learning_rate": 9.181213929280046e-06,
      "loss": 1.2742,
      "step": 1285
    },
    {
      "epoch": 0.012353654582787978,
      "grad_norm": 1.3512520790100098,
      "learning_rate": 9.12431052248861e-06,
      "loss": 0.9148,
      "step": 1286
    },
    {
      "epoch": 0.012363260846071639,
      "grad_norm": 2.137115716934204,
      "learning_rate": 9.067566292057084e-06,
      "loss": 2.1244,
      "step": 1287
    },
    {
      "epoch": 0.0123728671093553,
      "grad_norm": 1.3759469985961914,
      "learning_rate": 9.010981458956441e-06,
      "loss": 1.5954,
      "step": 1288
    },
    {
      "epoch": 0.012382473372638961,
      "grad_norm": 1.2014045715332031,
      "learning_rate": 8.954556243536877e-06,
      "loss": 0.351,
      "step": 1289
    },
    {
      "epoch": 0.012392079635922622,
      "grad_norm": 1.3174604177474976,
      "learning_rate": 8.898290865527077e-06,
      "loss": 2.3941,
      "step": 1290
    },
    {
      "epoch": 0.012401685899206283,
      "grad_norm": 1.6204346418380737,
      "learning_rate": 8.842185544033255e-06,
      "loss": 1.9556,
      "step": 1291
    },
    {
      "epoch": 0.012411292162489944,
      "grad_norm": 2.3952038288116455,
      "learning_rate": 8.786240497538368e-06,
      "loss": 0.8159,
      "step": 1292
    },
    {
      "epoch": 0.012420898425773604,
      "grad_norm": 1.7509487867355347,
      "learning_rate": 8.7304559439012e-06,
      "loss": 1.2616,
      "step": 1293
    },
    {
      "epoch": 0.012430504689057265,
      "grad_norm": 1.5139427185058594,
      "learning_rate": 8.674832100355567e-06,
      "loss": 1.8556,
      "step": 1294
    },
    {
      "epoch": 0.012440110952340926,
      "grad_norm": 2.4402108192443848,
      "learning_rate": 8.619369183509501e-06,
      "loss": 2.1442,
      "step": 1295
    },
    {
      "epoch": 0.012449717215624587,
      "grad_norm": 1.9534533023834229,
      "learning_rate": 8.564067409344279e-06,
      "loss": 1.8495,
      "step": 1296
    },
    {
      "epoch": 0.012459323478908248,
      "grad_norm": 2.0038352012634277,
      "learning_rate": 8.508926993213712e-06,
      "loss": 1.0759,
      "step": 1297
    },
    {
      "epoch": 0.012468929742191909,
      "grad_norm": 1.7788993120193481,
      "learning_rate": 8.453948149843244e-06,
      "loss": 1.8516,
      "step": 1298
    },
    {
      "epoch": 0.01247853600547557,
      "grad_norm": 2.8507580757141113,
      "learning_rate": 8.39913109332916e-06,
      "loss": 1.9621,
      "step": 1299
    },
    {
      "epoch": 0.01248814226875923,
      "grad_norm": 3.111874580383301,
      "learning_rate": 8.344476037137655e-06,
      "loss": 2.3904,
      "step": 1300
    },
    {
      "epoch": 0.012497748532042892,
      "grad_norm": 0.22093941271305084,
      "learning_rate": 8.28998319410413e-06,
      "loss": 1.9976,
      "step": 1301
    },
    {
      "epoch": 0.012507354795326553,
      "grad_norm": 0.4744524657726288,
      "learning_rate": 8.235652776432274e-06,
      "loss": 1.7633,
      "step": 1302
    },
    {
      "epoch": 0.012516961058610214,
      "grad_norm": 0.38851699233055115,
      "learning_rate": 8.181484995693295e-06,
      "loss": 2.2067,
      "step": 1303
    },
    {
      "epoch": 0.012526567321893875,
      "grad_norm": 0.52333664894104,
      "learning_rate": 8.127480062825e-06,
      "loss": 1.736,
      "step": 1304
    },
    {
      "epoch": 0.012536173585177536,
      "grad_norm": 0.47692447900772095,
      "learning_rate": 8.073638188131128e-06,
      "loss": 1.9328,
      "step": 1305
    },
    {
      "epoch": 0.012545779848461197,
      "grad_norm": 0.6180505156517029,
      "learning_rate": 8.019959581280401e-06,
      "loss": 1.7122,
      "step": 1306
    },
    {
      "epoch": 0.012555386111744858,
      "grad_norm": 0.9801742434501648,
      "learning_rate": 7.966444451305726e-06,
      "loss": 1.6527,
      "step": 1307
    },
    {
      "epoch": 0.01256499237502852,
      "grad_norm": 0.8951825499534607,
      "learning_rate": 7.913093006603445e-06,
      "loss": 0.2918,
      "step": 1308
    },
    {
      "epoch": 0.01257459863831218,
      "grad_norm": 0.5541632771492004,
      "learning_rate": 7.859905454932471e-06,
      "loss": 0.6242,
      "step": 1309
    },
    {
      "epoch": 0.012584204901595841,
      "grad_norm": 0.8574060797691345,
      "learning_rate": 7.806882003413496e-06,
      "loss": 1.9855,
      "step": 1310
    },
    {
      "epoch": 0.0125938111648795,
      "grad_norm": 0.7803676724433899,
      "learning_rate": 7.75402285852816e-06,
      "loss": 1.5606,
      "step": 1311
    },
    {
      "epoch": 0.012603417428163162,
      "grad_norm": 0.6776403188705444,
      "learning_rate": 7.701328226118288e-06,
      "loss": 1.2598,
      "step": 1312
    },
    {
      "epoch": 0.012613023691446823,
      "grad_norm": 0.7974050641059875,
      "learning_rate": 7.648798311385058e-06,
      "loss": 1.5879,
      "step": 1313
    },
    {
      "epoch": 0.012622629954730484,
      "grad_norm": 0.7562013864517212,
      "learning_rate": 7.596433318888224e-06,
      "loss": 2.1482,
      "step": 1314
    },
    {
      "epoch": 0.012632236218014145,
      "grad_norm": 0.8396043181419373,
      "learning_rate": 7.5442334525452964e-06,
      "loss": 0.9644,
      "step": 1315
    },
    {
      "epoch": 0.012641842481297806,
      "grad_norm": 1.0126715898513794,
      "learning_rate": 7.4921989156307545e-06,
      "loss": 0.7277,
      "step": 1316
    },
    {
      "epoch": 0.012651448744581467,
      "grad_norm": 1.0184913873672485,
      "learning_rate": 7.440329910775273e-06,
      "loss": 1.574,
      "step": 1317
    },
    {
      "epoch": 0.012661055007865128,
      "grad_norm": 0.928348183631897,
      "learning_rate": 7.3886266399649014e-06,
      "loss": 1.5747,
      "step": 1318
    },
    {
      "epoch": 0.012670661271148789,
      "grad_norm": 0.7095616459846497,
      "learning_rate": 7.337089304540301e-06,
      "loss": 1.2142,
      "step": 1319
    },
    {
      "epoch": 0.01268026753443245,
      "grad_norm": 0.8285765051841736,
      "learning_rate": 7.28571810519596e-06,
      "loss": 0.6966,
      "step": 1320
    },
    {
      "epoch": 0.012689873797716111,
      "grad_norm": 0.944232165813446,
      "learning_rate": 7.234513241979418e-06,
      "loss": 1.0616,
      "step": 1321
    },
    {
      "epoch": 0.012699480060999772,
      "grad_norm": 0.8697264194488525,
      "learning_rate": 7.183474914290433e-06,
      "loss": 1.8868,
      "step": 1322
    },
    {
      "epoch": 0.012709086324283433,
      "grad_norm": 0.9995365738868713,
      "learning_rate": 7.132603320880294e-06,
      "loss": 0.9878,
      "step": 1323
    },
    {
      "epoch": 0.012718692587567094,
      "grad_norm": 0.7990480065345764,
      "learning_rate": 7.081898659850977e-06,
      "loss": 0.9188,
      "step": 1324
    },
    {
      "epoch": 0.012728298850850755,
      "grad_norm": 1.0068573951721191,
      "learning_rate": 7.031361128654401e-06,
      "loss": 0.612,
      "step": 1325
    },
    {
      "epoch": 0.012737905114134416,
      "grad_norm": 0.8989691734313965,
      "learning_rate": 6.980990924091663e-06,
      "loss": 1.3281,
      "step": 1326
    },
    {
      "epoch": 0.012747511377418077,
      "grad_norm": 1.0953826904296875,
      "learning_rate": 6.930788242312253e-06,
      "loss": 0.718,
      "step": 1327
    },
    {
      "epoch": 0.012757117640701738,
      "grad_norm": 1.1208527088165283,
      "learning_rate": 6.880753278813312e-06,
      "loss": 1.3806,
      "step": 1328
    },
    {
      "epoch": 0.0127667239039854,
      "grad_norm": 1.1740251779556274,
      "learning_rate": 6.830886228438837e-06,
      "loss": 0.7611,
      "step": 1329
    },
    {
      "epoch": 0.012776330167269059,
      "grad_norm": 0.9152091145515442,
      "learning_rate": 6.781187285378965e-06,
      "loss": 0.6383,
      "step": 1330
    },
    {
      "epoch": 0.01278593643055272,
      "grad_norm": 1.1075633764266968,
      "learning_rate": 6.731656643169204e-06,
      "loss": 1.2403,
      "step": 1331
    },
    {
      "epoch": 0.01279554269383638,
      "grad_norm": 1.1017249822616577,
      "learning_rate": 6.682294494689656e-06,
      "loss": 0.7527,
      "step": 1332
    },
    {
      "epoch": 0.012805148957120042,
      "grad_norm": 0.9468207359313965,
      "learning_rate": 6.633101032164274e-06,
      "loss": 0.5834,
      "step": 1333
    },
    {
      "epoch": 0.012814755220403703,
      "grad_norm": 2.0187582969665527,
      "learning_rate": 6.584076447160137e-06,
      "loss": 1.1991,
      "step": 1334
    },
    {
      "epoch": 0.012824361483687364,
      "grad_norm": 1.5514315366744995,
      "learning_rate": 6.535220930586705e-06,
      "loss": 1.4413,
      "step": 1335
    },
    {
      "epoch": 0.012833967746971025,
      "grad_norm": 1.136216402053833,
      "learning_rate": 6.486534672695016e-06,
      "loss": 1.0502,
      "step": 1336
    },
    {
      "epoch": 0.012843574010254686,
      "grad_norm": 1.4092392921447754,
      "learning_rate": 6.4380178630770225e-06,
      "loss": 1.5213,
      "step": 1337
    },
    {
      "epoch": 0.012853180273538347,
      "grad_norm": 2.0249874591827393,
      "learning_rate": 6.389670690664801e-06,
      "loss": 1.8229,
      "step": 1338
    },
    {
      "epoch": 0.012862786536822008,
      "grad_norm": 1.2564117908477783,
      "learning_rate": 6.341493343729854e-06,
      "loss": 2.1412,
      "step": 1339
    },
    {
      "epoch": 0.012872392800105669,
      "grad_norm": 1.7565979957580566,
      "learning_rate": 6.2934860098823215e-06,
      "loss": 0.5932,
      "step": 1340
    },
    {
      "epoch": 0.01288199906338933,
      "grad_norm": 1.4513801336288452,
      "learning_rate": 6.2456488760703205e-06,
      "loss": 1.4253,
      "step": 1341
    },
    {
      "epoch": 0.012891605326672991,
      "grad_norm": 1.5311293601989746,
      "learning_rate": 6.1979821285791666e-06,
      "loss": 0.815,
      "step": 1342
    },
    {
      "epoch": 0.012901211589956652,
      "grad_norm": 2.166646957397461,
      "learning_rate": 6.150485953030677e-06,
      "loss": 0.9291,
      "step": 1343
    },
    {
      "epoch": 0.012910817853240313,
      "grad_norm": 1.7635732889175415,
      "learning_rate": 6.103160534382402e-06,
      "loss": 1.4243,
      "step": 1344
    },
    {
      "epoch": 0.012920424116523974,
      "grad_norm": 1.7761321067810059,
      "learning_rate": 6.056006056926977e-06,
      "loss": 1.5863,
      "step": 1345
    },
    {
      "epoch": 0.012930030379807635,
      "grad_norm": 2.130981922149658,
      "learning_rate": 6.009022704291361e-06,
      "loss": 2.7922,
      "step": 1346
    },
    {
      "epoch": 0.012939636643091296,
      "grad_norm": 1.645337700843811,
      "learning_rate": 5.962210659436091e-06,
      "loss": 0.5797,
      "step": 1347
    },
    {
      "epoch": 0.012949242906374956,
      "grad_norm": 2.2915232181549072,
      "learning_rate": 5.9155701046546384e-06,
      "loss": 2.1254,
      "step": 1348
    },
    {
      "epoch": 0.012958849169658617,
      "grad_norm": 3.340772867202759,
      "learning_rate": 5.869101221572654e-06,
      "loss": 1.6675,
      "step": 1349
    },
    {
      "epoch": 0.012968455432942278,
      "grad_norm": 2.850680112838745,
      "learning_rate": 5.82280419114728e-06,
      "loss": 2.7344,
      "step": 1350
    },
    {
      "epoch": 0.012978061696225939,
      "grad_norm": 0.2448108196258545,
      "learning_rate": 5.776679193666412e-06,
      "loss": 1.8639,
      "step": 1351
    },
    {
      "epoch": 0.0129876679595096,
      "grad_norm": 0.48849502205848694,
      "learning_rate": 5.730726408748055e-06,
      "loss": 2.4691,
      "step": 1352
    },
    {
      "epoch": 0.012997274222793261,
      "grad_norm": 1.0094072818756104,
      "learning_rate": 5.6849460153395706e-06,
      "loss": 0.7901,
      "step": 1353
    },
    {
      "epoch": 0.013006880486076922,
      "grad_norm": 1.0659600496292114,
      "learning_rate": 5.639338191717009e-06,
      "loss": 0.6873,
      "step": 1354
    },
    {
      "epoch": 0.013016486749360583,
      "grad_norm": 0.9116507768630981,
      "learning_rate": 5.5939031154844e-06,
      "loss": 1.9804,
      "step": 1355
    },
    {
      "epoch": 0.013026093012644244,
      "grad_norm": 0.6070476770401001,
      "learning_rate": 5.5486409635730815e-06,
      "loss": 1.8359,
      "step": 1356
    },
    {
      "epoch": 0.013035699275927905,
      "grad_norm": 0.6397705674171448,
      "learning_rate": 5.5035519122409895e-06,
      "loss": 1.7745,
      "step": 1357
    },
    {
      "epoch": 0.013045305539211566,
      "grad_norm": 0.5579898357391357,
      "learning_rate": 5.4586361370719654e-06,
      "loss": 0.9213,
      "step": 1358
    },
    {
      "epoch": 0.013054911802495227,
      "grad_norm": 0.8104727268218994,
      "learning_rate": 5.413893812975096e-06,
      "loss": 1.5619,
      "step": 1359
    },
    {
      "epoch": 0.013064518065778888,
      "grad_norm": 1.7391012907028198,
      "learning_rate": 5.369325114184032e-06,
      "loss": 0.6587,
      "step": 1360
    },
    {
      "epoch": 0.01307412432906255,
      "grad_norm": 0.7869606614112854,
      "learning_rate": 5.324930214256302e-06,
      "loss": 2.1848,
      "step": 1361
    },
    {
      "epoch": 0.01308373059234621,
      "grad_norm": 0.9439055919647217,
      "learning_rate": 5.280709286072605e-06,
      "loss": 1.5044,
      "step": 1362
    },
    {
      "epoch": 0.013093336855629871,
      "grad_norm": 0.6741794347763062,
      "learning_rate": 5.236662501836192e-06,
      "loss": 1.2642,
      "step": 1363
    },
    {
      "epoch": 0.013102943118913532,
      "grad_norm": 0.722041130065918,
      "learning_rate": 5.19279003307217e-06,
      "loss": 0.66,
      "step": 1364
    },
    {
      "epoch": 0.013112549382197193,
      "grad_norm": 0.7658513784408569,
      "learning_rate": 5.149092050626825e-06,
      "loss": 1.2821,
      "step": 1365
    },
    {
      "epoch": 0.013122155645480853,
      "grad_norm": 0.9324590563774109,
      "learning_rate": 5.10556872466697e-06,
      "loss": 1.4646,
      "step": 1366
    },
    {
      "epoch": 0.013131761908764514,
      "grad_norm": 0.8357721567153931,
      "learning_rate": 5.062220224679276e-06,
      "loss": 0.5584,
      "step": 1367
    },
    {
      "epoch": 0.013141368172048175,
      "grad_norm": 1.147518277168274,
      "learning_rate": 5.0190467194696276e-06,
      "loss": 1.2917,
      "step": 1368
    },
    {
      "epoch": 0.013150974435331836,
      "grad_norm": 0.7350714206695557,
      "learning_rate": 4.9760483771624236e-06,
      "loss": 1.0685,
      "step": 1369
    },
    {
      "epoch": 0.013160580698615497,
      "grad_norm": 0.9620083570480347,
      "learning_rate": 4.933225365199967e-06,
      "loss": 0.5959,
      "step": 1370
    },
    {
      "epoch": 0.013170186961899158,
      "grad_norm": 0.9014163613319397,
      "learning_rate": 4.89057785034181e-06,
      "loss": 1.3587,
      "step": 1371
    },
    {
      "epoch": 0.013179793225182819,
      "grad_norm": 0.9376206994056702,
      "learning_rate": 4.848105998664082e-06,
      "loss": 0.9965,
      "step": 1372
    },
    {
      "epoch": 0.01318939948846648,
      "grad_norm": 1.0249470472335815,
      "learning_rate": 4.805809975558828e-06,
      "loss": 1.0462,
      "step": 1373
    },
    {
      "epoch": 0.013199005751750141,
      "grad_norm": 1.0543888807296753,
      "learning_rate": 4.763689945733418e-06,
      "loss": 1.2151,
      "step": 1374
    },
    {
      "epoch": 0.013208612015033802,
      "grad_norm": 0.9313603043556213,
      "learning_rate": 4.721746073209893e-06,
      "loss": 1.3105,
      "step": 1375
    },
    {
      "epoch": 0.013218218278317463,
      "grad_norm": 0.8977903723716736,
      "learning_rate": 4.679978521324257e-06,
      "loss": 1.5354,
      "step": 1376
    },
    {
      "epoch": 0.013227824541601124,
      "grad_norm": 1.0173827409744263,
      "learning_rate": 4.6383874527259345e-06,
      "loss": 1.6489,
      "step": 1377
    },
    {
      "epoch": 0.013237430804884785,
      "grad_norm": 1.3930602073669434,
      "learning_rate": 4.5969730293770776e-06,
      "loss": 1.0227,
      "step": 1378
    },
    {
      "epoch": 0.013247037068168446,
      "grad_norm": 1.0031945705413818,
      "learning_rate": 4.555735412551975e-06,
      "loss": 0.5266,
      "step": 1379
    },
    {
      "epoch": 0.013256643331452107,
      "grad_norm": 1.2070573568344116,
      "learning_rate": 4.514674762836369e-06,
      "loss": 1.4161,
      "step": 1380
    },
    {
      "epoch": 0.013266249594735768,
      "grad_norm": 1.2639909982681274,
      "learning_rate": 4.47379124012689e-06,
      "loss": 0.7972,
      "step": 1381
    },
    {
      "epoch": 0.01327585585801943,
      "grad_norm": 0.8697017431259155,
      "learning_rate": 4.433085003630411e-06,
      "loss": 0.4543,
      "step": 1382
    },
    {
      "epoch": 0.01328546212130309,
      "grad_norm": 1.457741618156433,
      "learning_rate": 4.3925562118634135e-06,
      "loss": 0.936,
      "step": 1383
    },
    {
      "epoch": 0.01329506838458675,
      "grad_norm": 1.92344331741333,
      "learning_rate": 4.352205022651368e-06,
      "loss": 1.0984,
      "step": 1384
    },
    {
      "epoch": 0.01330467464787041,
      "grad_norm": 1.0835517644882202,
      "learning_rate": 4.312031593128163e-06,
      "loss": 0.6498,
      "step": 1385
    },
    {
      "epoch": 0.013314280911154072,
      "grad_norm": 1.0525588989257812,
      "learning_rate": 4.272036079735447e-06,
      "loss": 0.4696,
      "step": 1386
    },
    {
      "epoch": 0.013323887174437733,
      "grad_norm": 1.5031206607818604,
      "learning_rate": 4.232218638222029e-06,
      "loss": 1.3497,
      "step": 1387
    },
    {
      "epoch": 0.013333493437721394,
      "grad_norm": 1.6809800863265991,
      "learning_rate": 4.192579423643289e-06,
      "loss": 1.8485,
      "step": 1388
    },
    {
      "epoch": 0.013343099701005055,
      "grad_norm": 1.566742181777954,
      "learning_rate": 4.153118590360561e-06,
      "loss": 1.9606,
      "step": 1389
    },
    {
      "epoch": 0.013352705964288716,
      "grad_norm": 2.624690532684326,
      "learning_rate": 4.113836292040535e-06,
      "loss": 1.4085,
      "step": 1390
    },
    {
      "epoch": 0.013362312227572377,
      "grad_norm": 1.7665361166000366,
      "learning_rate": 4.074732681654647e-06,
      "loss": 1.3442,
      "step": 1391
    },
    {
      "epoch": 0.013371918490856038,
      "grad_norm": 1.2678183317184448,
      "learning_rate": 4.035807911478501e-06,
      "loss": 1.623,
      "step": 1392
    },
    {
      "epoch": 0.0133815247541397,
      "grad_norm": 1.4695771932601929,
      "learning_rate": 3.997062133091284e-06,
      "loss": 1.2938,
      "step": 1393
    },
    {
      "epoch": 0.01339113101742336,
      "grad_norm": 1.8408315181732178,
      "learning_rate": 3.958495497375125e-06,
      "loss": 1.7552,
      "step": 1394
    },
    {
      "epoch": 0.013400737280707021,
      "grad_norm": 1.8904510736465454,
      "learning_rate": 3.920108154514585e-06,
      "loss": 1.9988,
      "step": 1395
    },
    {
      "epoch": 0.013410343543990682,
      "grad_norm": 2.1557648181915283,
      "learning_rate": 3.881900253996007e-06,
      "loss": 1.3713,
      "step": 1396
    },
    {
      "epoch": 0.013419949807274343,
      "grad_norm": 1.6359537839889526,
      "learning_rate": 3.843871944606969e-06,
      "loss": 1.0889,
      "step": 1397
    },
    {
      "epoch": 0.013429556070558004,
      "grad_norm": 2.364774465560913,
      "learning_rate": 3.8060233744356633e-06,
      "loss": 2.1195,
      "step": 1398
    },
    {
      "epoch": 0.013439162333841665,
      "grad_norm": 3.0365054607391357,
      "learning_rate": 3.7683546908703903e-06,
      "loss": 1.579,
      "step": 1399
    },
    {
      "epoch": 0.013448768597125326,
      "grad_norm": 4.097827911376953,
      "learning_rate": 3.7308660405989236e-06,
      "loss": 2.1318,
      "step": 1400
    },
    {
      "epoch": 0.013458374860408987,
      "grad_norm": 0.2823294699192047,
      "learning_rate": 3.693557569607947e-06,
      "loss": 2.2207,
      "step": 1401
    },
    {
      "epoch": 0.013467981123692647,
      "grad_norm": 0.29963523149490356,
      "learning_rate": 3.656429423182517e-06,
      "loss": 2.1717,
      "step": 1402
    },
    {
      "epoch": 0.013477587386976308,
      "grad_norm": 0.4321596324443817,
      "learning_rate": 3.6194817459054676e-06,
      "loss": 1.9181,
      "step": 1403
    },
    {
      "epoch": 0.013487193650259969,
      "grad_norm": 0.493495911359787,
      "learning_rate": 3.5827146816568844e-06,
      "loss": 1.314,
      "step": 1404
    },
    {
      "epoch": 0.01349679991354363,
      "grad_norm": 0.42508795857429504,
      "learning_rate": 3.5461283736134722e-06,
      "loss": 1.7113,
      "step": 1405
    },
    {
      "epoch": 0.013506406176827291,
      "grad_norm": 0.5699524879455566,
      "learning_rate": 3.5097229642480743e-06,
      "loss": 2.1593,
      "step": 1406
    },
    {
      "epoch": 0.013516012440110952,
      "grad_norm": 0.9306424856185913,
      "learning_rate": 3.4734985953290778e-06,
      "loss": 0.7923,
      "step": 1407
    },
    {
      "epoch": 0.013525618703394613,
      "grad_norm": 0.6619681119918823,
      "learning_rate": 3.4374554079198594e-06,
      "loss": 1.9823,
      "step": 1408
    },
    {
      "epoch": 0.013535224966678274,
      "grad_norm": 0.8809049129486084,
      "learning_rate": 3.401593542378262e-06,
      "loss": 0.225,
      "step": 1409
    },
    {
      "epoch": 0.013544831229961935,
      "grad_norm": 0.6989437937736511,
      "learning_rate": 3.365913138356025e-06,
      "loss": 1.975,
      "step": 1410
    },
    {
      "epoch": 0.013554437493245596,
      "grad_norm": 0.6701787114143372,
      "learning_rate": 3.330414334798265e-06,
      "loss": 0.6392,
      "step": 1411
    },
    {
      "epoch": 0.013564043756529257,
      "grad_norm": 0.8192588686943054,
      "learning_rate": 3.2950972699428817e-06,
      "loss": 2.832,
      "step": 1412
    },
    {
      "epoch": 0.013573650019812918,
      "grad_norm": 0.9694103002548218,
      "learning_rate": 3.2599620813200837e-06,
      "loss": 1.5731,
      "step": 1413
    },
    {
      "epoch": 0.01358325628309658,
      "grad_norm": 1.0692158937454224,
      "learning_rate": 3.2250089057518307e-06,
      "loss": 0.712,
      "step": 1414
    },
    {
      "epoch": 0.01359286254638024,
      "grad_norm": 0.7780011296272278,
      "learning_rate": 3.1902378793512657e-06,
      "loss": 1.7028,
      "step": 1415
    },
    {
      "epoch": 0.013602468809663901,
      "grad_norm": 1.214434027671814,
      "learning_rate": 3.1556491375222375e-06,
      "loss": 1.1284,
      "step": 1416
    },
    {
      "epoch": 0.013612075072947562,
      "grad_norm": 0.9437004327774048,
      "learning_rate": 3.121242814958747e-06,
      "loss": 1.269,
      "step": 1417
    },
    {
      "epoch": 0.013621681336231223,
      "grad_norm": 0.9650669693946838,
      "learning_rate": 3.0870190456444237e-06,
      "loss": 1.3202,
      "step": 1418
    },
    {
      "epoch": 0.013631287599514885,
      "grad_norm": 0.8260666728019714,
      "learning_rate": 3.0529779628519992e-06,
      "loss": 1.6349,
      "step": 1419
    },
    {
      "epoch": 0.013640893862798544,
      "grad_norm": 0.7508506178855896,
      "learning_rate": 3.019119699142797e-06,
      "loss": 1.5069,
      "step": 1420
    },
    {
      "epoch": 0.013650500126082205,
      "grad_norm": 1.0261093378067017,
      "learning_rate": 2.9854443863662262e-06,
      "loss": 1.1122,
      "step": 1421
    },
    {
      "epoch": 0.013660106389365866,
      "grad_norm": 0.9321567416191101,
      "learning_rate": 2.9519521556592488e-06,
      "loss": 0.6697,
      "step": 1422
    },
    {
      "epoch": 0.013669712652649527,
      "grad_norm": 0.6702725291252136,
      "learning_rate": 2.918643137445859e-06,
      "loss": 0.5922,
      "step": 1423
    },
    {
      "epoch": 0.013679318915933188,
      "grad_norm": 0.8566491007804871,
      "learning_rate": 2.8855174614366322e-06,
      "loss": 0.7623,
      "step": 1424
    },
    {
      "epoch": 0.013688925179216849,
      "grad_norm": 1.276656150817871,
      "learning_rate": 2.8525752566281482e-06,
      "loss": 0.6412,
      "step": 1425
    },
    {
      "epoch": 0.01369853144250051,
      "grad_norm": 0.8470809459686279,
      "learning_rate": 2.8198166513025314e-06,
      "loss": 0.8668,
      "step": 1426
    },
    {
      "epoch": 0.013708137705784171,
      "grad_norm": 1.352955937385559,
      "learning_rate": 2.787241773026933e-06,
      "loss": 1.9754,
      "step": 1427
    },
    {
      "epoch": 0.013717743969067832,
      "grad_norm": 1.2425868511199951,
      "learning_rate": 2.7548507486530596e-06,
      "loss": 1.1799,
      "step": 1428
    },
    {
      "epoch": 0.013727350232351493,
      "grad_norm": 1.1900222301483154,
      "learning_rate": 2.722643704316652e-06,
      "loss": 2.1851,
      "step": 1429
    },
    {
      "epoch": 0.013736956495635154,
      "grad_norm": 1.0071966648101807,
      "learning_rate": 2.690620765436991e-06,
      "loss": 1.6281,
      "step": 1430
    },
    {
      "epoch": 0.013746562758918815,
      "grad_norm": 1.1236071586608887,
      "learning_rate": 2.658782056716441e-06,
      "loss": 1.2622,
      "step": 1431
    },
    {
      "epoch": 0.013756169022202476,
      "grad_norm": 1.721375823020935,
      "learning_rate": 2.6271277021399423e-06,
      "loss": 0.9908,
      "step": 1432
    },
    {
      "epoch": 0.013765775285486137,
      "grad_norm": 1.129111886024475,
      "learning_rate": 2.5956578249745236e-06,
      "loss": 1.2414,
      "step": 1433
    },
    {
      "epoch": 0.013775381548769798,
      "grad_norm": 1.0341899394989014,
      "learning_rate": 2.564372547768834e-06,
      "loss": 1.0645,
      "step": 1434
    },
    {
      "epoch": 0.01378498781205346,
      "grad_norm": 2.1746985912323,
      "learning_rate": 2.533271992352659e-06,
      "loss": 1.7303,
      "step": 1435
    },
    {
      "epoch": 0.01379459407533712,
      "grad_norm": 1.0520601272583008,
      "learning_rate": 2.5023562798364542e-06,
      "loss": 1.1918,
      "step": 1436
    },
    {
      "epoch": 0.013804200338620782,
      "grad_norm": 1.3040817975997925,
      "learning_rate": 2.4716255306108605e-06,
      "loss": 1.7243,
      "step": 1437
    },
    {
      "epoch": 0.013813806601904441,
      "grad_norm": 1.2131621837615967,
      "learning_rate": 2.441079864346241e-06,
      "loss": 1.6467,
      "step": 1438
    },
    {
      "epoch": 0.013823412865188102,
      "grad_norm": 1.4932360649108887,
      "learning_rate": 2.4107193999922286e-06,
      "loss": 0.7456,
      "step": 1439
    },
    {
      "epoch": 0.013833019128471763,
      "grad_norm": 1.399996280670166,
      "learning_rate": 2.3805442557772506e-06,
      "loss": 1.2783,
      "step": 1440
    },
    {
      "epoch": 0.013842625391755424,
      "grad_norm": 1.5067113637924194,
      "learning_rate": 2.3505545492080395e-06,
      "loss": 1.3007,
      "step": 1441
    },
    {
      "epoch": 0.013852231655039085,
      "grad_norm": 1.757777452468872,
      "learning_rate": 2.3207503970692435e-06,
      "loss": 1.4619,
      "step": 1442
    },
    {
      "epoch": 0.013861837918322746,
      "grad_norm": 1.7575135231018066,
      "learning_rate": 2.291131915422917e-06,
      "loss": 1.3914,
      "step": 1443
    },
    {
      "epoch": 0.013871444181606407,
      "grad_norm": 1.7992342710494995,
      "learning_rate": 2.261699219608071e-06,
      "loss": 1.7884,
      "step": 1444
    },
    {
      "epoch": 0.013881050444890068,
      "grad_norm": 2.1739306449890137,
      "learning_rate": 2.2324524242402613e-06,
      "loss": 2.0404,
      "step": 1445
    },
    {
      "epoch": 0.01389065670817373,
      "grad_norm": 1.924720287322998,
      "learning_rate": 2.2033916432111003e-06,
      "loss": 1.8472,
      "step": 1446
    },
    {
      "epoch": 0.01390026297145739,
      "grad_norm": 1.7615585327148438,
      "learning_rate": 2.1745169896878414e-06,
      "loss": 1.6542,
      "step": 1447
    },
    {
      "epoch": 0.013909869234741051,
      "grad_norm": 2.5631535053253174,
      "learning_rate": 2.145828576112918e-06,
      "loss": 2.1631,
      "step": 1448
    },
    {
      "epoch": 0.013919475498024712,
      "grad_norm": 2.641089677810669,
      "learning_rate": 2.117326514203527e-06,
      "loss": 2.6554,
      "step": 1449
    },
    {
      "epoch": 0.013929081761308373,
      "grad_norm": 4.36536979675293,
      "learning_rate": 2.089010914951173e-06,
      "loss": 2.9866,
      "step": 1450
    },
    {
      "epoch": 0.013938688024592034,
      "grad_norm": 0.23590919375419617,
      "learning_rate": 2.0608818886212576e-06,
      "loss": 1.8436,
      "step": 1451
    },
    {
      "epoch": 0.013948294287875695,
      "grad_norm": 0.27973809838294983,
      "learning_rate": 2.032939544752627e-06,
      "loss": 1.8202,
      "step": 1452
    },
    {
      "epoch": 0.013957900551159357,
      "grad_norm": 0.6220752596855164,
      "learning_rate": 2.0051839921571448e-06,
      "loss": 0.9031,
      "step": 1453
    },
    {
      "epoch": 0.013967506814443018,
      "grad_norm": 0.3801034092903137,
      "learning_rate": 1.977615338919331e-06,
      "loss": 2.0523,
      "step": 1454
    },
    {
      "epoch": 0.013977113077726679,
      "grad_norm": 0.5059555172920227,
      "learning_rate": 1.9502336923958255e-06,
      "loss": 1.5032,
      "step": 1455
    },
    {
      "epoch": 0.013986719341010338,
      "grad_norm": 0.8687939643859863,
      "learning_rate": 1.923039159215073e-06,
      "loss": 1.4558,
      "step": 1456
    },
    {
      "epoch": 0.013996325604293999,
      "grad_norm": 0.8506740927696228,
      "learning_rate": 1.8960318452768577e-06,
      "loss": 1.5112,
      "step": 1457
    },
    {
      "epoch": 0.01400593186757766,
      "grad_norm": 0.6744850873947144,
      "learning_rate": 1.8692118557519124e-06,
      "loss": 1.4923,
      "step": 1458
    },
    {
      "epoch": 0.014015538130861321,
      "grad_norm": 1.4139418601989746,
      "learning_rate": 1.8425792950814868e-06,
      "loss": 1.1357,
      "step": 1459
    },
    {
      "epoch": 0.014025144394144982,
      "grad_norm": 1.0256134271621704,
      "learning_rate": 1.8161342669769543e-06,
      "loss": 0.2689,
      "step": 1460
    },
    {
      "epoch": 0.014034750657428643,
      "grad_norm": 1.235180377960205,
      "learning_rate": 1.7898768744194162e-06,
      "loss": 1.9205,
      "step": 1461
    },
    {
      "epoch": 0.014044356920712304,
      "grad_norm": 1.0876485109329224,
      "learning_rate": 1.7638072196592915e-06,
      "loss": 1.6574,
      "step": 1462
    },
    {
      "epoch": 0.014053963183995965,
      "grad_norm": 0.7919371724128723,
      "learning_rate": 1.7379254042158955e-06,
      "loss": 1.8323,
      "step": 1463
    },
    {
      "epoch": 0.014063569447279626,
      "grad_norm": 1.52744460105896,
      "learning_rate": 1.7122315288771063e-06,
      "loss": 1.1787,
      "step": 1464
    },
    {
      "epoch": 0.014073175710563287,
      "grad_norm": 1.0921329259872437,
      "learning_rate": 1.6867256936989096e-06,
      "loss": 1.0773,
      "step": 1465
    },
    {
      "epoch": 0.014082781973846948,
      "grad_norm": 0.866415798664093,
      "learning_rate": 1.6614079980050268e-06,
      "loss": 0.5916,
      "step": 1466
    },
    {
      "epoch": 0.01409238823713061,
      "grad_norm": 0.9815545082092285,
      "learning_rate": 1.6362785403865488e-06,
      "loss": 1.5516,
      "step": 1467
    },
    {
      "epoch": 0.01410199450041427,
      "grad_norm": 0.7811928987503052,
      "learning_rate": 1.6113374187015361e-06,
      "loss": 0.628,
      "step": 1468
    },
    {
      "epoch": 0.014111600763697931,
      "grad_norm": 0.790367841720581,
      "learning_rate": 1.5865847300746417e-06,
      "loss": 0.8627,
      "step": 1469
    },
    {
      "epoch": 0.014121207026981593,
      "grad_norm": 0.8883091807365417,
      "learning_rate": 1.5620205708967162e-06,
      "loss": 1.3902,
      "step": 1470
    },
    {
      "epoch": 0.014130813290265254,
      "grad_norm": 0.7730988264083862,
      "learning_rate": 1.5376450368244589e-06,
      "loss": 0.6142,
      "step": 1471
    },
    {
      "epoch": 0.014140419553548915,
      "grad_norm": 0.9608266949653625,
      "learning_rate": 1.5134582227800287e-06,
      "loss": 1.3799,
      "step": 1472
    },
    {
      "epoch": 0.014150025816832576,
      "grad_norm": 0.9299761652946472,
      "learning_rate": 1.4894602229506892e-06,
      "loss": 1.8379,
      "step": 1473
    },
    {
      "epoch": 0.014159632080116235,
      "grad_norm": 1.4585217237472534,
      "learning_rate": 1.4656511307884146e-06,
      "loss": 0.426,
      "step": 1474
    },
    {
      "epoch": 0.014169238343399896,
      "grad_norm": 1.0566785335540771,
      "learning_rate": 1.4420310390095615e-06,
      "loss": 1.0758,
      "step": 1475
    },
    {
      "epoch": 0.014178844606683557,
      "grad_norm": 1.250511646270752,
      "learning_rate": 1.4186000395944809e-06,
      "loss": 1.6016,
      "step": 1476
    },
    {
      "epoch": 0.014188450869967218,
      "grad_norm": 1.2056167125701904,
      "learning_rate": 1.3953582237871521e-06,
      "loss": 1.2261,
      "step": 1477
    },
    {
      "epoch": 0.014198057133250879,
      "grad_norm": 1.8116402626037598,
      "learning_rate": 1.372305682094871e-06,
      "loss": 0.7249,
      "step": 1478
    },
    {
      "epoch": 0.01420766339653454,
      "grad_norm": 0.9976388216018677,
      "learning_rate": 1.3494425042878622e-06,
      "loss": 0.5366,
      "step": 1479
    },
    {
      "epoch": 0.014217269659818201,
      "grad_norm": 1.1733542680740356,
      "learning_rate": 1.326768779398929e-06,
      "loss": 1.1214,
      "step": 1480
    },
    {
      "epoch": 0.014226875923101862,
      "grad_norm": 1.1367336511611938,
      "learning_rate": 1.3042845957231153e-06,
      "loss": 0.4766,
      "step": 1481
    },
    {
      "epoch": 0.014236482186385523,
      "grad_norm": 1.329309344291687,
      "learning_rate": 1.2819900408173714e-06,
      "loss": 0.9044,
      "step": 1482
    },
    {
      "epoch": 0.014246088449669184,
      "grad_norm": 1.1273263692855835,
      "learning_rate": 1.2598852015001994e-06,
      "loss": 0.6729,
      "step": 1483
    },
    {
      "epoch": 0.014255694712952845,
      "grad_norm": 0.6687442064285278,
      "learning_rate": 1.2379701638513153e-06,
      "loss": 0.6622,
      "step": 1484
    },
    {
      "epoch": 0.014265300976236506,
      "grad_norm": 1.4767571687698364,
      "learning_rate": 1.2162450132113201e-06,
      "loss": 1.3024,
      "step": 1485
    },
    {
      "epoch": 0.014274907239520167,
      "grad_norm": 1.4602934122085571,
      "learning_rate": 1.1947098341813678e-06,
      "loss": 0.9379,
      "step": 1486
    },
    {
      "epoch": 0.014284513502803829,
      "grad_norm": 2.381859540939331,
      "learning_rate": 1.1733647106228375e-06,
      "loss": 0.4574,
      "step": 1487
    },
    {
      "epoch": 0.01429411976608749,
      "grad_norm": 1.9809890985488892,
      "learning_rate": 1.1522097256569776e-06,
      "loss": 1.307,
      "step": 1488
    },
    {
      "epoch": 0.01430372602937115,
      "grad_norm": 2.0697414875030518,
      "learning_rate": 1.1312449616646403e-06,
      "loss": 1.2738,
      "step": 1489
    },
    {
      "epoch": 0.014313332292654812,
      "grad_norm": 2.2143564224243164,
      "learning_rate": 1.110470500285904e-06,
      "loss": 1.9771,
      "step": 1490
    },
    {
      "epoch": 0.014322938555938473,
      "grad_norm": 1.7997151613235474,
      "learning_rate": 1.0898864224197946e-06,
      "loss": 1.0769,
      "step": 1491
    },
    {
      "epoch": 0.014332544819222132,
      "grad_norm": 1.6824840307235718,
      "learning_rate": 1.0694928082239376e-06,
      "loss": 2.1892,
      "step": 1492
    },
    {
      "epoch": 0.014342151082505793,
      "grad_norm": 1.4572582244873047,
      "learning_rate": 1.049289737114273e-06,
      "loss": 1.2788,
      "step": 1493
    },
    {
      "epoch": 0.014351757345789454,
      "grad_norm": 2.2297871112823486,
      "learning_rate": 1.0292772877647462e-06,
      "loss": 1.3602,
      "step": 1494
    },
    {
      "epoch": 0.014361363609073115,
      "grad_norm": 1.9319744110107422,
      "learning_rate": 1.009455538106968e-06,
      "loss": 1.102,
      "step": 1495
    },
    {
      "epoch": 0.014370969872356776,
      "grad_norm": 1.9980595111846924,
      "learning_rate": 9.898245653299554e-07,
      "loss": 2.0976,
      "step": 1496
    },
    {
      "epoch": 0.014380576135640437,
      "grad_norm": 2.023252010345459,
      "learning_rate": 9.703844458797962e-07,
      "loss": 2.0978,
      "step": 1497
    },
    {
      "epoch": 0.014390182398924098,
      "grad_norm": 2.8768935203552246,
      "learning_rate": 9.511352554593789e-07,
      "loss": 1.4897,
      "step": 1498
    },
    {
      "epoch": 0.01439978866220776,
      "grad_norm": 3.7225050926208496,
      "learning_rate": 9.320770690280645e-07,
      "loss": 1.7528,
      "step": 1499
    },
    {
      "epoch": 0.01440939492549142,
      "grad_norm": 3.3216378688812256,
      "learning_rate": 9.132099608014422e-07,
      "loss": 2.7666,
      "step": 1500
    },
    {
      "epoch": 0.014419001188775081,
      "grad_norm": 0.24966958165168762,
      "learning_rate": 8.945340042509797e-07,
      "loss": 1.8411,
      "step": 1501
    },
    {
      "epoch": 0.014428607452058742,
      "grad_norm": 0.3495580852031708,
      "learning_rate": 8.760492721038071e-07,
      "loss": 1.9743,
      "step": 1502
    },
    {
      "epoch": 0.014438213715342403,
      "grad_norm": 0.3945554494857788,
      "learning_rate": 8.577558363423554e-07,
      "loss": 1.8308,
      "step": 1503
    },
    {
      "epoch": 0.014447819978626064,
      "grad_norm": 0.5365820527076721,
      "learning_rate": 8.396537682041572e-07,
      "loss": 1.3296,
      "step": 1504
    },
    {
      "epoch": 0.014457426241909726,
      "grad_norm": 0.9025552868843079,
      "learning_rate": 8.217431381815077e-07,
      "loss": 0.6257,
      "step": 1505
    },
    {
      "epoch": 0.014467032505193387,
      "grad_norm": 0.7715771794319153,
      "learning_rate": 8.040240160212042e-07,
      "loss": 1.6995,
      "step": 1506
    },
    {
      "epoch": 0.014476638768477048,
      "grad_norm": 0.5638648271560669,
      "learning_rate": 7.864964707243072e-07,
      "loss": 1.9291,
      "step": 1507
    },
    {
      "epoch": 0.014486245031760709,
      "grad_norm": 0.9612266421318054,
      "learning_rate": 7.691605705458238e-07,
      "loss": 1.5632,
      "step": 1508
    },
    {
      "epoch": 0.01449585129504437,
      "grad_norm": 0.6024612784385681,
      "learning_rate": 7.520163829944804e-07,
      "loss": 0.8159,
      "step": 1509
    },
    {
      "epoch": 0.014505457558328029,
      "grad_norm": 0.5374557971954346,
      "learning_rate": 7.350639748324339e-07,
      "loss": 1.1212,
      "step": 1510
    },
    {
      "epoch": 0.01451506382161169,
      "grad_norm": 0.758751392364502,
      "learning_rate": 7.183034120750221e-07,
      "loss": 0.8058,
      "step": 1511
    },
    {
      "epoch": 0.014524670084895351,
      "grad_norm": 1.3426655530929565,
      "learning_rate": 7.017347599905244e-07,
      "loss": 0.3788,
      "step": 1512
    },
    {
      "epoch": 0.014534276348179012,
      "grad_norm": 1.2120193243026733,
      "learning_rate": 6.85358083099863e-07,
      "loss": 2.3863,
      "step": 1513
    },
    {
      "epoch": 0.014543882611462673,
      "grad_norm": 0.7500411868095398,
      "learning_rate": 6.69173445176402e-07,
      "loss": 1.7041,
      "step": 1514
    },
    {
      "epoch": 0.014553488874746334,
      "grad_norm": 0.6524091958999634,
      "learning_rate": 6.531809092456598e-07,
      "loss": 0.963,
      "step": 1515
    },
    {
      "epoch": 0.014563095138029995,
      "grad_norm": 1.0304814577102661,
      "learning_rate": 6.373805375850917e-07,
      "loss": 1.1896,
      "step": 1516
    },
    {
      "epoch": 0.014572701401313656,
      "grad_norm": 1.5014792680740356,
      "learning_rate": 6.217723917238128e-07,
      "loss": 0.993,
      "step": 1517
    },
    {
      "epoch": 0.014582307664597317,
      "grad_norm": 0.862172544002533,
      "learning_rate": 6.063565324424147e-07,
      "loss": 1.0368,
      "step": 1518
    },
    {
      "epoch": 0.014591913927880978,
      "grad_norm": 0.8092368245124817,
      "learning_rate": 5.911330197726661e-07,
      "loss": 1.0823,
      "step": 1519
    },
    {
      "epoch": 0.01460152019116464,
      "grad_norm": 0.9634461402893066,
      "learning_rate": 5.761019129973233e-07,
      "loss": 1.141,
      "step": 1520
    },
    {
      "epoch": 0.0146111264544483,
      "grad_norm": 0.6817344427108765,
      "learning_rate": 5.612632706498755e-07,
      "loss": 0.5265,
      "step": 1521
    },
    {
      "epoch": 0.014620732717731962,
      "grad_norm": 0.7728158235549927,
      "learning_rate": 5.466171505143225e-07,
      "loss": 1.2575,
      "step": 1522
    },
    {
      "epoch": 0.014630338981015623,
      "grad_norm": 0.8764125108718872,
      "learning_rate": 5.321636096249749e-07,
      "loss": 1.0003,
      "step": 1523
    },
    {
      "epoch": 0.014639945244299284,
      "grad_norm": 0.8837773203849792,
      "learning_rate": 5.179027042661766e-07,
      "loss": 1.3595,
      "step": 1524
    },
    {
      "epoch": 0.014649551507582945,
      "grad_norm": 1.1360968351364136,
      "learning_rate": 5.038344899721436e-07,
      "loss": 0.6597,
      "step": 1525
    },
    {
      "epoch": 0.014659157770866606,
      "grad_norm": 1.3072655200958252,
      "learning_rate": 4.899590215267091e-07,
      "loss": 1.0458,
      "step": 1526
    },
    {
      "epoch": 0.014668764034150267,
      "grad_norm": 1.0967586040496826,
      "learning_rate": 4.762763529631342e-07,
      "loss": 0.8311,
      "step": 1527
    },
    {
      "epoch": 0.014678370297433926,
      "grad_norm": 1.428826093673706,
      "learning_rate": 4.6278653756387514e-07,
      "loss": 1.2745,
      "step": 1528
    },
    {
      "epoch": 0.014687976560717587,
      "grad_norm": 1.4646788835525513,
      "learning_rate": 4.4948962786039437e-07,
      "loss": 1.8977,
      "step": 1529
    },
    {
      "epoch": 0.014697582824001248,
      "grad_norm": 1.023618459701538,
      "learning_rate": 4.3638567563294986e-07,
      "loss": 0.5181,
      "step": 1530
    },
    {
      "epoch": 0.01470718908728491,
      "grad_norm": 1.4419240951538086,
      "learning_rate": 4.234747319103949e-07,
      "loss": 0.6687,
      "step": 1531
    },
    {
      "epoch": 0.01471679535056857,
      "grad_norm": 1.5811681747436523,
      "learning_rate": 4.107568469699674e-07,
      "loss": 1.565,
      "step": 1532
    },
    {
      "epoch": 0.014726401613852231,
      "grad_norm": 0.9161353707313538,
      "learning_rate": 3.9823207033710676e-07,
      "loss": 0.9838,
      "step": 1533
    },
    {
      "epoch": 0.014736007877135892,
      "grad_norm": 1.7939404249191284,
      "learning_rate": 3.859004507852759e-07,
      "loss": 1.7109,
      "step": 1534
    },
    {
      "epoch": 0.014745614140419553,
      "grad_norm": 1.7447971105575562,
      "learning_rate": 3.737620363357286e-07,
      "loss": 1.5584,
      "step": 1535
    },
    {
      "epoch": 0.014755220403703214,
      "grad_norm": 1.9135932922363281,
      "learning_rate": 3.6181687425736464e-07,
      "loss": 1.4576,
      "step": 1536
    },
    {
      "epoch": 0.014764826666986875,
      "grad_norm": 2.033905267715454,
      "learning_rate": 3.5006501106651937e-07,
      "loss": 1.7224,
      "step": 1537
    },
    {
      "epoch": 0.014774432930270536,
      "grad_norm": 0.867377519607544,
      "learning_rate": 3.385064925267967e-07,
      "loss": 0.5116,
      "step": 1538
    },
    {
      "epoch": 0.014784039193554198,
      "grad_norm": 1.5313493013381958,
      "learning_rate": 3.2714136364888073e-07,
      "loss": 1.5841,
      "step": 1539
    },
    {
      "epoch": 0.014793645456837859,
      "grad_norm": 1.7377293109893799,
      "learning_rate": 3.159696686903746e-07,
      "loss": 1.6841,
      "step": 1540
    },
    {
      "epoch": 0.01480325172012152,
      "grad_norm": 1.6860902309417725,
      "learning_rate": 3.0499145115561176e-07,
      "loss": 2.3476,
      "step": 1541
    },
    {
      "epoch": 0.01481285798340518,
      "grad_norm": 2.2776246070861816,
      "learning_rate": 2.942067537955007e-07,
      "loss": 1.7652,
      "step": 1542
    },
    {
      "epoch": 0.014822464246688842,
      "grad_norm": 2.107292652130127,
      "learning_rate": 2.836156186073413e-07,
      "loss": 0.8998,
      "step": 1543
    },
    {
      "epoch": 0.014832070509972503,
      "grad_norm": 1.7746764421463013,
      "learning_rate": 2.732180868346812e-07,
      "loss": 1.356,
      "step": 1544
    },
    {
      "epoch": 0.014841676773256164,
      "grad_norm": 2.2240185737609863,
      "learning_rate": 2.630141989671542e-07,
      "loss": 1.4766,
      "step": 1545
    },
    {
      "epoch": 0.014851283036539825,
      "grad_norm": 2.506375551223755,
      "learning_rate": 2.5300399474028624e-07,
      "loss": 1.5676,
      "step": 1546
    },
    {
      "epoch": 0.014860889299823484,
      "grad_norm": 2.127063035964966,
      "learning_rate": 2.431875131354011e-07,
      "loss": 1.2854,
      "step": 1547
    },
    {
      "epoch": 0.014870495563107145,
      "grad_norm": 2.4192678928375244,
      "learning_rate": 2.3356479237940932e-07,
      "loss": 3.158,
      "step": 1548
    },
    {
      "epoch": 0.014880101826390806,
      "grad_norm": 4.550783634185791,
      "learning_rate": 2.2413586994470825e-07,
      "loss": 2.2723,
      "step": 1549
    },
    {
      "epoch": 0.014889708089674467,
      "grad_norm": 3.2424111366271973,
      "learning_rate": 2.1490078254899348e-07,
      "loss": 3.6502,
      "step": 1550
    },
    {
      "epoch": 0.014899314352958128,
      "grad_norm": 0.22420750558376312,
      "learning_rate": 2.0585956615515323e-07,
      "loss": 1.8073,
      "step": 1551
    },
    {
      "epoch": 0.01490892061624179,
      "grad_norm": 0.5209183692932129,
      "learning_rate": 1.970122559711074e-07,
      "loss": 1.9495,
      "step": 1552
    },
    {
      "epoch": 0.01491852687952545,
      "grad_norm": 0.6448322534561157,
      "learning_rate": 1.8835888644966325e-07,
      "loss": 0.8256,
      "step": 1553
    },
    {
      "epoch": 0.014928133142809111,
      "grad_norm": 0.44594621658325195,
      "learning_rate": 1.7989949128841554e-07,
      "loss": 1.9334,
      "step": 1554
    },
    {
      "epoch": 0.014937739406092772,
      "grad_norm": 0.5354653000831604,
      "learning_rate": 1.7163410342956875e-07,
      "loss": 1.7156,
      "step": 1555
    },
    {
      "epoch": 0.014947345669376434,
      "grad_norm": 0.5310769081115723,
      "learning_rate": 1.6356275505984286e-07,
      "loss": 1.8007,
      "step": 1556
    },
    {
      "epoch": 0.014956951932660095,
      "grad_norm": 0.6432744264602661,
      "learning_rate": 1.5568547761034004e-07,
      "loss": 1.8852,
      "step": 1557
    },
    {
      "epoch": 0.014966558195943756,
      "grad_norm": 0.448311448097229,
      "learning_rate": 1.4800230175641695e-07,
      "loss": 0.2679,
      "step": 1558
    },
    {
      "epoch": 0.014976164459227417,
      "grad_norm": 0.5687196850776672,
      "learning_rate": 1.4051325741756828e-07,
      "loss": 2.2304,
      "step": 1559
    },
    {
      "epoch": 0.014985770722511078,
      "grad_norm": 0.6607801914215088,
      "learning_rate": 1.3321837375731005e-07,
      "loss": 2.6455,
      "step": 1560
    },
    {
      "epoch": 0.014995376985794739,
      "grad_norm": 0.7040688991546631,
      "learning_rate": 1.2611767918306316e-07,
      "loss": 1.6596,
      "step": 1561
    },
    {
      "epoch": 0.0150049832490784,
      "grad_norm": 1.1383543014526367,
      "learning_rate": 1.192112013460478e-07,
      "loss": 0.1267,
      "step": 1562
    },
    {
      "epoch": 0.01501458951236206,
      "grad_norm": 1.1857229471206665,
      "learning_rate": 1.1249896714117802e-07,
      "loss": 2.4044,
      "step": 1563
    },
    {
      "epoch": 0.015024195775645722,
      "grad_norm": 0.851668119430542,
      "learning_rate": 1.0598100270694522e-07,
      "loss": 1.0669,
      "step": 1564
    },
    {
      "epoch": 0.015033802038929381,
      "grad_norm": 0.8586041331291199,
      "learning_rate": 9.965733342532924e-08,
      "loss": 0.8862,
      "step": 1565
    },
    {
      "epoch": 0.015043408302213042,
      "grad_norm": 1.378212571144104,
      "learning_rate": 9.352798392169848e-08,
      "loss": 0.5062,
      "step": 1566
    },
    {
      "epoch": 0.015053014565496703,
      "grad_norm": 0.8861387372016907,
      "learning_rate": 8.759297806469335e-08,
      "loss": 1.812,
      "step": 1567
    },
    {
      "epoch": 0.015062620828780364,
      "grad_norm": 0.8511702418327332,
      "learning_rate": 8.185233896616517e-08,
      "loss": 1.4519,
      "step": 1568
    },
    {
      "epoch": 0.015072227092064025,
      "grad_norm": 0.8892300128936768,
      "learning_rate": 7.630608898105962e-08,
      "loss": 0.4583,
      "step": 1569
    },
    {
      "epoch": 0.015081833355347686,
      "grad_norm": 1.119099736213684,
      "learning_rate": 7.09542497073501e-08,
      "loss": 1.1426,
      "step": 1570
    },
    {
      "epoch": 0.015091439618631347,
      "grad_norm": 0.932698130607605,
      "learning_rate": 6.579684198594338e-08,
      "loss": 0.8147,
      "step": 1571
    },
    {
      "epoch": 0.015101045881915008,
      "grad_norm": 0.8807966113090515,
      "learning_rate": 6.083388590059081e-08,
      "loss": 0.9775,
      "step": 1572
    },
    {
      "epoch": 0.01511065214519867,
      "grad_norm": 0.842082679271698,
      "learning_rate": 5.606540077782163e-08,
      "loss": 1.5158,
      "step": 1573
    },
    {
      "epoch": 0.01512025840848233,
      "grad_norm": 0.9434205889701843,
      "learning_rate": 5.14914051868709e-08,
      "loss": 1.0191,
      "step": 1574
    },
    {
      "epoch": 0.015129864671765992,
      "grad_norm": 0.8906828165054321,
      "learning_rate": 4.711191693959616e-08,
      "loss": 1.0639,
      "step": 1575
    },
    {
      "epoch": 0.015139470935049653,
      "grad_norm": 1.015246033668518,
      "learning_rate": 4.292695309040529e-08,
      "loss": 1.1553,
      "step": 1576
    },
    {
      "epoch": 0.015149077198333314,
      "grad_norm": 1.3302627801895142,
      "learning_rate": 3.893652993621766e-08,
      "loss": 1.9563,
      "step": 1577
    },
    {
      "epoch": 0.015158683461616975,
      "grad_norm": 1.197374701499939,
      "learning_rate": 3.5140663016364205e-08,
      "loss": 1.8744,
      "step": 1578
    },
    {
      "epoch": 0.015168289724900636,
      "grad_norm": 1.2963511943817139,
      "learning_rate": 3.1539367112543014e-08,
      "loss": 1.3826,
      "step": 1579
    },
    {
      "epoch": 0.015177895988184297,
      "grad_norm": 1.4743306636810303,
      "learning_rate": 2.8132656248769372e-08,
      "loss": 1.2313,
      "step": 1580
    },
    {
      "epoch": 0.015187502251467958,
      "grad_norm": 1.2326154708862305,
      "learning_rate": 2.4920543691309138e-08,
      "loss": 1.0279,
      "step": 1581
    },
    {
      "epoch": 0.015197108514751619,
      "grad_norm": 0.7234031558036804,
      "learning_rate": 2.19030419486399e-08,
      "loss": 0.2631,
      "step": 1582
    },
    {
      "epoch": 0.015206714778035278,
      "grad_norm": 1.2122236490249634,
      "learning_rate": 1.9080162771378808e-08,
      "loss": 1.0213,
      "step": 1583
    },
    {
      "epoch": 0.01521632104131894,
      "grad_norm": 1.6157090663909912,
      "learning_rate": 1.6451917152265906e-08,
      "loss": 0.8347,
      "step": 1584
    },
    {
      "epoch": 0.0152259273046026,
      "grad_norm": 1.1818397045135498,
      "learning_rate": 1.4018315326103094e-08,
      "loss": 1.2913,
      "step": 1585
    },
    {
      "epoch": 0.015235533567886261,
      "grad_norm": 1.3002811670303345,
      "learning_rate": 1.177936676971525e-08,
      "loss": 0.8638,
      "step": 1586
    },
    {
      "epoch": 0.015245139831169922,
      "grad_norm": 1.2042262554168701,
      "learning_rate": 9.735080201922487e-09,
      "loss": 1.9613,
      "step": 1587
    },
    {
      "epoch": 0.015254746094453583,
      "grad_norm": 1.5267974138259888,
      "learning_rate": 7.885463583495733e-09,
      "loss": 1.1249,
      "step": 1588
    },
    {
      "epoch": 0.015264352357737244,
      "grad_norm": 1.7169855833053589,
      "learning_rate": 6.2305241171345395e-09,
      "loss": 2.1914,
      "step": 1589
    },
    {
      "epoch": 0.015273958621020906,
      "grad_norm": 1.3588310480117798,
      "learning_rate": 4.7702682474337625e-09,
      "loss": 1.141,
      "step": 1590
    },
    {
      "epoch": 0.015283564884304567,
      "grad_norm": 1.7462457418441772,
      "learning_rate": 3.5047016608613647e-09,
      "loss": 1.0643,
      "step": 1591
    },
    {
      "epoch": 0.015293171147588228,
      "grad_norm": 2.717428684234619,
      "learning_rate": 2.43382928573066e-09,
      "loss": 2.198,
      "step": 1592
    },
    {
      "epoch": 0.015302777410871889,
      "grad_norm": 2.002877950668335,
      "learning_rate": 1.5576552921836574e-09,
      "loss": 1.2204,
      "step": 1593
    },
    {
      "epoch": 0.01531238367415555,
      "grad_norm": 1.8236584663391113,
      "learning_rate": 8.761830921855119e-10,
      "loss": 1.0132,
      "step": 1594
    },
    {
      "epoch": 0.01532198993743921,
      "grad_norm": 2.3542301654815674,
      "learning_rate": 3.89415339491217e-10,
      "loss": 1.5282,
      "step": 1595
    },
    {
      "epoch": 0.015331596200722872,
      "grad_norm": 2.0443341732025146,
      "learning_rate": 9.735392965115608e-11,
      "loss": 2.1046,
      "step": 1596
    },
    {
      "epoch": 0.015341202464006533,
      "grad_norm": 2.864762783050537,
      "learning_rate": 0.0,
      "loss": 2.3461,
      "step": 1597
    }
  ],
  "logging_steps": 1,
  "max_steps": 1597,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 400,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.5799448097325056e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}