{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 6198,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016134236850596966,
      "grad_norm": 3.2937216758728027,
      "learning_rate": 0.0,
      "loss": 4.697282791137695,
      "step": 1
    },
    {
      "epoch": 0.00032268473701193933,
      "grad_norm": 1.8991987705230713,
      "learning_rate": 6.25e-06,
      "loss": 3.822343349456787,
      "step": 2
    },
    {
      "epoch": 0.000484027105517909,
      "grad_norm": 1.5379608869552612,
      "learning_rate": 1.25e-05,
      "loss": 3.825277328491211,
      "step": 3
    },
    {
      "epoch": 0.0006453694740238787,
      "grad_norm": 1.3724831342697144,
      "learning_rate": 1.875e-05,
      "loss": 3.645108699798584,
      "step": 4
    },
    {
      "epoch": 0.0008067118425298483,
      "grad_norm": 1.3183643817901611,
      "learning_rate": 2.5e-05,
      "loss": 3.801774740219116,
      "step": 5
    },
    {
      "epoch": 0.000968054211035818,
      "grad_norm": 1.5425629615783691,
      "learning_rate": 3.125e-05,
      "loss": 4.086655139923096,
      "step": 6
    },
    {
      "epoch": 0.0011293965795417876,
      "grad_norm": 1.8867943286895752,
      "learning_rate": 3.75e-05,
      "loss": 4.094951629638672,
      "step": 7
    },
    {
      "epoch": 0.0012907389480477573,
      "grad_norm": 1.5533658266067505,
      "learning_rate": 4.375e-05,
      "loss": 3.885037899017334,
      "step": 8
    },
    {
      "epoch": 0.001452081316553727,
      "grad_norm": 1.8594249486923218,
      "learning_rate": 5e-05,
      "loss": 3.843696117401123,
      "step": 9
    },
    {
      "epoch": 0.0016134236850596966,
      "grad_norm": 1.3997585773468018,
      "learning_rate": 5.6250000000000005e-05,
      "loss": 3.8958077430725098,
      "step": 10
    },
    {
      "epoch": 0.0017747660535656663,
      "grad_norm": 1.2593034505844116,
      "learning_rate": 6.25e-05,
      "loss": 3.6179256439208984,
      "step": 11
    },
    {
      "epoch": 0.001936108422071636,
      "grad_norm": 2.0564138889312744,
      "learning_rate": 6.875e-05,
      "loss": 3.7616753578186035,
      "step": 12
    },
    {
      "epoch": 0.002097450790577606,
      "grad_norm": 1.4568206071853638,
      "learning_rate": 7.5e-05,
      "loss": 3.6334080696105957,
      "step": 13
    },
    {
      "epoch": 0.002258793159083575,
      "grad_norm": 1.8639119863510132,
      "learning_rate": 8.125000000000001e-05,
      "loss": 3.7859084606170654,
      "step": 14
    },
    {
      "epoch": 0.002420135527589545,
      "grad_norm": 1.5626275539398193,
      "learning_rate": 8.75e-05,
      "loss": 3.5656545162200928,
      "step": 15
    },
    {
      "epoch": 0.0025814778960955146,
      "grad_norm": 1.4653570652008057,
      "learning_rate": 9.375e-05,
      "loss": 3.9613301753997803,
      "step": 16
    },
    {
      "epoch": 0.0027428202646014844,
      "grad_norm": 2.201716899871826,
      "learning_rate": 0.0001,
      "loss": 4.420013904571533,
      "step": 17
    },
    {
      "epoch": 0.002904162633107454,
      "grad_norm": 1.6758142709732056,
      "learning_rate": 0.00010625,
      "loss": 3.535153865814209,
      "step": 18
    },
    {
      "epoch": 0.003065505001613424,
      "grad_norm": 1.520393967628479,
      "learning_rate": 0.00011250000000000001,
      "loss": 3.582017660140991,
      "step": 19
    },
    {
      "epoch": 0.003226847370119393,
      "grad_norm": 1.5639102458953857,
      "learning_rate": 0.00011875,
      "loss": 3.5169811248779297,
      "step": 20
    },
    {
      "epoch": 0.003388189738625363,
      "grad_norm": 1.3940562009811401,
      "learning_rate": 0.000125,
      "loss": 4.207969665527344,
      "step": 21
    },
    {
      "epoch": 0.0035495321071313327,
      "grad_norm": 1.604730248451233,
      "learning_rate": 0.00013125000000000002,
      "loss": 3.82344913482666,
      "step": 22
    },
    {
      "epoch": 0.0037108744756373024,
      "grad_norm": 2.0010297298431396,
      "learning_rate": 0.0001375,
      "loss": 3.77018666267395,
      "step": 23
    },
    {
      "epoch": 0.003872216844143272,
      "grad_norm": 1.6270376443862915,
      "learning_rate": 0.00014375,
      "loss": 3.2159104347229004,
      "step": 24
    },
    {
      "epoch": 0.004033559212649242,
      "grad_norm": 1.6079480648040771,
      "learning_rate": 0.00015,
      "loss": 3.4189260005950928,
      "step": 25
    },
    {
      "epoch": 0.004194901581155212,
      "grad_norm": 1.5214911699295044,
      "learning_rate": 0.00015625,
      "loss": 3.8557116985321045,
      "step": 26
    },
    {
      "epoch": 0.004356243949661181,
      "grad_norm": 1.785361409187317,
      "learning_rate": 0.00016250000000000002,
      "loss": 3.8567752838134766,
      "step": 27
    },
    {
      "epoch": 0.00451758631816715,
      "grad_norm": 1.5584688186645508,
      "learning_rate": 0.00016875,
      "loss": 3.589975357055664,
      "step": 28
    },
    {
      "epoch": 0.00467892868667312,
      "grad_norm": 1.6266164779663086,
      "learning_rate": 0.000175,
      "loss": 3.910428047180176,
      "step": 29
    },
    {
      "epoch": 0.00484027105517909,
      "grad_norm": 1.4989351034164429,
      "learning_rate": 0.00018125,
      "loss": 3.865623950958252,
      "step": 30
    },
    {
      "epoch": 0.0050016134236850595,
      "grad_norm": 1.4021748304367065,
      "learning_rate": 0.0001875,
      "loss": 4.00720739364624,
      "step": 31
    },
    {
      "epoch": 0.005162955792191029,
      "grad_norm": 2.1958837509155273,
      "learning_rate": 0.00019375000000000002,
      "loss": 3.780177593231201,
      "step": 32
    },
    {
      "epoch": 0.005324298160696999,
      "grad_norm": 1.7857956886291504,
      "learning_rate": 0.0002,
      "loss": 3.631948947906494,
      "step": 33
    },
    {
      "epoch": 0.005485640529202969,
      "grad_norm": 1.6889210939407349,
      "learning_rate": 0.00020625,
      "loss": 3.7388389110565186,
      "step": 34
    },
    {
      "epoch": 0.0056469828977089385,
      "grad_norm": 1.4513267278671265,
      "learning_rate": 0.0002125,
      "loss": 3.9068703651428223,
      "step": 35
    },
    {
      "epoch": 0.005808325266214908,
      "grad_norm": 1.6280912160873413,
      "learning_rate": 0.00021875,
      "loss": 3.692418098449707,
      "step": 36
    },
    {
      "epoch": 0.005969667634720878,
      "grad_norm": 1.7336057424545288,
      "learning_rate": 0.00022500000000000002,
      "loss": 3.513784885406494,
      "step": 37
    },
    {
      "epoch": 0.006131010003226848,
      "grad_norm": 1.5870596170425415,
      "learning_rate": 0.00023125,
      "loss": 3.6550514698028564,
      "step": 38
    },
    {
      "epoch": 0.0062923523717328175,
      "grad_norm": 1.5565085411071777,
      "learning_rate": 0.0002375,
      "loss": 3.8411898612976074,
      "step": 39
    },
    {
      "epoch": 0.006453694740238786,
      "grad_norm": 1.6788463592529297,
      "learning_rate": 0.00024375,
      "loss": 3.79123592376709,
      "step": 40
    },
    {
      "epoch": 0.006615037108744756,
      "grad_norm": 2.1756134033203125,
      "learning_rate": 0.00025,
      "loss": 3.6755518913269043,
      "step": 41
    },
    {
      "epoch": 0.006776379477250726,
      "grad_norm": 1.712059497833252,
      "learning_rate": 0.00025624999999999997,
      "loss": 2.867650032043457,
      "step": 42
    },
    {
      "epoch": 0.006937721845756696,
      "grad_norm": 1.6851009130477905,
      "learning_rate": 0.00026250000000000004,
      "loss": 3.9782352447509766,
      "step": 43
    },
    {
      "epoch": 0.007099064214262665,
      "grad_norm": 2.4328672885894775,
      "learning_rate": 0.00026875,
      "loss": 3.800018787384033,
      "step": 44
    },
    {
      "epoch": 0.007260406582768635,
      "grad_norm": 1.8289200067520142,
      "learning_rate": 0.000275,
      "loss": 3.6191539764404297,
      "step": 45
    },
    {
      "epoch": 0.007421748951274605,
      "grad_norm": 1.5387808084487915,
      "learning_rate": 0.00028125000000000003,
      "loss": 3.781719207763672,
      "step": 46
    },
    {
      "epoch": 0.0075830913197805746,
      "grad_norm": 1.7275835275650024,
      "learning_rate": 0.0002875,
      "loss": 3.5585062503814697,
      "step": 47
    },
    {
      "epoch": 0.007744433688286544,
      "grad_norm": 1.6622196435928345,
      "learning_rate": 0.00029375,
      "loss": 3.812946081161499,
      "step": 48
    },
    {
      "epoch": 0.007905776056792513,
      "grad_norm": 1.5160530805587769,
      "learning_rate": 0.0003,
      "loss": 3.7433762550354004,
      "step": 49
    },
    {
      "epoch": 0.008067118425298484,
      "grad_norm": 1.4692721366882324,
      "learning_rate": 0.00030625000000000004,
      "loss": 3.194249153137207,
      "step": 50
    },
    {
      "epoch": 0.008228460793804453,
      "grad_norm": 2.0471856594085693,
      "learning_rate": 0.0003125,
      "loss": 3.6226327419281006,
      "step": 51
    },
    {
      "epoch": 0.008389803162310423,
      "grad_norm": 1.719726800918579,
      "learning_rate": 0.00031874999999999997,
      "loss": 3.5444226264953613,
      "step": 52
    },
    {
      "epoch": 0.008551145530816392,
      "grad_norm": 1.7122915983200073,
      "learning_rate": 0.00032500000000000004,
      "loss": 3.6201953887939453,
      "step": 53
    },
    {
      "epoch": 0.008712487899322363,
      "grad_norm": 1.3830492496490479,
      "learning_rate": 0.00033125,
      "loss": 3.7984583377838135,
      "step": 54
    },
    {
      "epoch": 0.008873830267828332,
      "grad_norm": 1.9534919261932373,
      "learning_rate": 0.0003375,
      "loss": 3.9989089965820312,
      "step": 55
    },
    {
      "epoch": 0.0090351726363343,
      "grad_norm": 1.644492268562317,
      "learning_rate": 0.00034375,
      "loss": 3.9307756423950195,
      "step": 56
    },
    {
      "epoch": 0.009196515004840271,
      "grad_norm": 1.7288904190063477,
      "learning_rate": 0.00035,
      "loss": 3.5896596908569336,
      "step": 57
    },
    {
      "epoch": 0.00935785737334624,
      "grad_norm": 1.6374150514602661,
      "learning_rate": 0.00035625,
      "loss": 3.712336540222168,
      "step": 58
    },
    {
      "epoch": 0.00951919974185221,
      "grad_norm": 1.9983317852020264,
      "learning_rate": 0.0003625,
      "loss": 3.6833810806274414,
      "step": 59
    },
    {
      "epoch": 0.00968054211035818,
      "grad_norm": 1.7046561241149902,
      "learning_rate": 0.00036875000000000005,
      "loss": 3.437739849090576,
      "step": 60
    },
    {
      "epoch": 0.00984188447886415,
      "grad_norm": 1.9450174570083618,
      "learning_rate": 0.000375,
      "loss": 3.803840160369873,
      "step": 61
    },
    {
      "epoch": 0.010003226847370119,
      "grad_norm": 1.7923344373703003,
      "learning_rate": 0.00038124999999999997,
      "loss": 3.702732563018799,
      "step": 62
    },
    {
      "epoch": 0.01016456921587609,
      "grad_norm": 1.6393837928771973,
      "learning_rate": 0.00038750000000000004,
      "loss": 4.114499092102051,
      "step": 63
    },
    {
      "epoch": 0.010325911584382058,
      "grad_norm": 1.7265268564224243,
      "learning_rate": 0.00039375,
      "loss": 3.6940412521362305,
      "step": 64
    },
    {
      "epoch": 0.010487253952888029,
      "grad_norm": 1.4106998443603516,
      "learning_rate": 0.0004,
      "loss": 3.6383328437805176,
      "step": 65
    },
    {
      "epoch": 0.010648596321393998,
      "grad_norm": 1.5595842599868774,
      "learning_rate": 0.00040625000000000004,
      "loss": 3.7100284099578857,
      "step": 66
    },
    {
      "epoch": 0.010809938689899969,
      "grad_norm": 1.716234564781189,
      "learning_rate": 0.0004125,
      "loss": 3.712735176086426,
      "step": 67
    },
    {
      "epoch": 0.010971281058405937,
      "grad_norm": 1.6054946184158325,
      "learning_rate": 0.00041875,
      "loss": 3.492037296295166,
      "step": 68
    },
    {
      "epoch": 0.011132623426911906,
      "grad_norm": 2.889824867248535,
      "learning_rate": 0.000425,
      "loss": 3.614187240600586,
      "step": 69
    },
    {
      "epoch": 0.011293965795417877,
      "grad_norm": 1.5802068710327148,
      "learning_rate": 0.00043125000000000005,
      "loss": 3.6142196655273438,
      "step": 70
    },
    {
      "epoch": 0.011455308163923846,
      "grad_norm": 1.7368475198745728,
      "learning_rate": 0.0004375,
      "loss": 3.981456995010376,
      "step": 71
    },
    {
      "epoch": 0.011616650532429816,
      "grad_norm": 1.5304979085922241,
      "learning_rate": 0.00044374999999999997,
      "loss": 3.825592517852783,
      "step": 72
    },
    {
      "epoch": 0.011777992900935785,
      "grad_norm": 1.6582465171813965,
      "learning_rate": 0.00045000000000000004,
      "loss": 3.708951234817505,
      "step": 73
    },
    {
      "epoch": 0.011939335269441756,
      "grad_norm": 2.323261260986328,
      "learning_rate": 0.00045625,
      "loss": 3.8038501739501953,
      "step": 74
    },
    {
      "epoch": 0.012100677637947725,
      "grad_norm": 1.9502477645874023,
      "learning_rate": 0.0004625,
      "loss": 4.040788173675537,
      "step": 75
    },
    {
      "epoch": 0.012262020006453695,
      "grad_norm": 2.2693307399749756,
      "learning_rate": 0.00046875,
      "loss": 3.973229169845581,
      "step": 76
    },
    {
      "epoch": 0.012423362374959664,
      "grad_norm": 1.6923143863677979,
      "learning_rate": 0.000475,
      "loss": 3.943993091583252,
      "step": 77
    },
    {
      "epoch": 0.012584704743465635,
      "grad_norm": 1.5953521728515625,
      "learning_rate": 0.00048125,
      "loss": 3.131744861602783,
      "step": 78
    },
    {
      "epoch": 0.012746047111971604,
      "grad_norm": 1.9842677116394043,
      "learning_rate": 0.0004875,
      "loss": 3.8448288440704346,
      "step": 79
    },
    {
      "epoch": 0.012907389480477573,
      "grad_norm": 1.9305503368377686,
      "learning_rate": 0.00049375,
      "loss": 3.978891611099243,
      "step": 80
    },
    {
      "epoch": 0.013068731848983543,
      "grad_norm": 1.5912785530090332,
      "learning_rate": 0.0005,
      "loss": 3.6208302974700928,
      "step": 81
    },
    {
      "epoch": 0.013230074217489512,
      "grad_norm": 1.5902907848358154,
      "learning_rate": 0.0005,
      "loss": 3.7949161529541016,
      "step": 82
    },
    {
      "epoch": 0.013391416585995483,
      "grad_norm": 2.0203490257263184,
      "learning_rate": 0.0005,
      "loss": 3.698604106903076,
      "step": 83
    },
    {
      "epoch": 0.013552758954501452,
      "grad_norm": 1.583745002746582,
      "learning_rate": 0.0005,
      "loss": 3.39426326751709,
      "step": 84
    },
    {
      "epoch": 0.013714101323007422,
      "grad_norm": 1.7308366298675537,
      "learning_rate": 0.0005,
      "loss": 3.9529078006744385,
      "step": 85
    },
    {
      "epoch": 0.013875443691513391,
      "grad_norm": 2.3999505043029785,
      "learning_rate": 0.0005,
      "loss": 4.176935195922852,
      "step": 86
    },
    {
      "epoch": 0.014036786060019362,
      "grad_norm": 2.545725107192993,
      "learning_rate": 0.0005,
      "loss": 3.820488452911377,
      "step": 87
    },
    {
      "epoch": 0.01419812842852533,
      "grad_norm": 1.909967303276062,
      "learning_rate": 0.0005,
      "loss": 4.032298564910889,
      "step": 88
    },
    {
      "epoch": 0.014359470797031301,
      "grad_norm": 2.524451732635498,
      "learning_rate": 0.0005,
      "loss": 3.5708370208740234,
      "step": 89
    },
    {
      "epoch": 0.01452081316553727,
      "grad_norm": 2.225862503051758,
      "learning_rate": 0.0005,
      "loss": 3.9952831268310547,
      "step": 90
    },
    {
      "epoch": 0.014682155534043239,
      "grad_norm": 1.9627829790115356,
      "learning_rate": 0.0005,
      "loss": 4.104012489318848,
      "step": 91
    },
    {
      "epoch": 0.01484349790254921,
      "grad_norm": 2.379063367843628,
      "learning_rate": 0.0005,
      "loss": 3.562069892883301,
      "step": 92
    },
    {
      "epoch": 0.015004840271055178,
      "grad_norm": 1.9125769138336182,
      "learning_rate": 0.0005,
      "loss": 3.7350246906280518,
      "step": 93
    },
    {
      "epoch": 0.015166182639561149,
      "grad_norm": 6.401985168457031,
      "learning_rate": 0.0005,
      "loss": 3.9207539558410645,
      "step": 94
    },
    {
      "epoch": 0.015327525008067118,
      "grad_norm": 2.3297975063323975,
      "learning_rate": 0.0005,
      "loss": 3.7669050693511963,
      "step": 95
    },
    {
      "epoch": 0.015488867376573089,
      "grad_norm": 1.729200005531311,
      "learning_rate": 0.0005,
      "loss": 4.143070220947266,
      "step": 96
    },
    {
      "epoch": 0.01565020974507906,
      "grad_norm": 1.5756016969680786,
      "learning_rate": 0.0005,
      "loss": 3.816279411315918,
      "step": 97
    },
    {
      "epoch": 0.015811552113585026,
      "grad_norm": 1.8842626810073853,
      "learning_rate": 0.0005,
      "loss": 3.692542314529419,
      "step": 98
    },
    {
      "epoch": 0.015972894482090997,
      "grad_norm": 1.937996506690979,
      "learning_rate": 0.0005,
      "loss": 3.5239105224609375,
      "step": 99
    },
    {
      "epoch": 0.016134236850596968,
      "grad_norm": 1.6298748254776,
      "learning_rate": 0.0005,
      "loss": 3.704568386077881,
      "step": 100
    },
    {
      "epoch": 0.016295579219102935,
      "grad_norm": 1.7370280027389526,
      "learning_rate": 0.0005,
      "loss": 3.982927083969116,
      "step": 101
    },
    {
      "epoch": 0.016456921587608905,
      "grad_norm": 2.1509196758270264,
      "learning_rate": 0.0005,
      "loss": 3.7261459827423096,
      "step": 102
    },
    {
      "epoch": 0.016618263956114876,
      "grad_norm": 1.6552603244781494,
      "learning_rate": 0.0005,
      "loss": 3.9928157329559326,
      "step": 103
    },
    {
      "epoch": 0.016779606324620847,
      "grad_norm": 1.9558533430099487,
      "learning_rate": 0.0005,
      "loss": 4.328695297241211,
      "step": 104
    },
    {
      "epoch": 0.016940948693126814,
      "grad_norm": 1.590743899345398,
      "learning_rate": 0.0005,
      "loss": 3.618778705596924,
      "step": 105
    },
    {
      "epoch": 0.017102291061632784,
      "grad_norm": 1.7812769412994385,
      "learning_rate": 0.0005,
      "loss": 3.7600150108337402,
      "step": 106
    },
    {
      "epoch": 0.017263633430138755,
      "grad_norm": 1.8761284351348877,
      "learning_rate": 0.0005,
      "loss": 3.4829063415527344,
      "step": 107
    },
    {
      "epoch": 0.017424975798644726,
      "grad_norm": 1.8785477876663208,
      "learning_rate": 0.0005,
      "loss": 3.4510579109191895,
      "step": 108
    },
    {
      "epoch": 0.017586318167150693,
      "grad_norm": 1.686873197555542,
      "learning_rate": 0.0005,
      "loss": 4.224066734313965,
      "step": 109
    },
    {
      "epoch": 0.017747660535656663,
      "grad_norm": 2.9169933795928955,
      "learning_rate": 0.0005,
      "loss": 4.036013126373291,
      "step": 110
    },
    {
      "epoch": 0.017909002904162634,
      "grad_norm": 1.9844744205474854,
      "learning_rate": 0.0005,
      "loss": 3.956897258758545,
      "step": 111
    },
    {
      "epoch": 0.0180703452726686,
      "grad_norm": 2.5851595401763916,
      "learning_rate": 0.0005,
      "loss": 4.031371116638184,
      "step": 112
    },
    {
      "epoch": 0.01823168764117457,
      "grad_norm": 2.103290319442749,
      "learning_rate": 0.0005,
      "loss": 4.344532012939453,
      "step": 113
    },
    {
      "epoch": 0.018393030009680542,
      "grad_norm": 1.8002307415008545,
      "learning_rate": 0.0005,
      "loss": 3.94378399848938,
      "step": 114
    },
    {
      "epoch": 0.018554372378186513,
      "grad_norm": 2.0613574981689453,
      "learning_rate": 0.0005,
      "loss": 3.476944923400879,
      "step": 115
    },
    {
      "epoch": 0.01871571474669248,
      "grad_norm": 2.2201764583587646,
      "learning_rate": 0.0005,
      "loss": 4.1407928466796875,
      "step": 116
    },
    {
      "epoch": 0.01887705711519845,
      "grad_norm": 2.6888511180877686,
      "learning_rate": 0.0005,
      "loss": 3.9569005966186523,
      "step": 117
    },
    {
      "epoch": 0.01903839948370442,
      "grad_norm": 1.5450550317764282,
      "learning_rate": 0.0005,
      "loss": 3.5972683429718018,
      "step": 118
    },
    {
      "epoch": 0.019199741852210392,
      "grad_norm": 1.3564611673355103,
      "learning_rate": 0.0005,
      "loss": 4.152257919311523,
      "step": 119
    },
    {
      "epoch": 0.01936108422071636,
      "grad_norm": 2.031658172607422,
      "learning_rate": 0.0005,
      "loss": 3.9204928874969482,
      "step": 120
    },
    {
      "epoch": 0.01952242658922233,
      "grad_norm": 2.78377366065979,
      "learning_rate": 0.0005,
      "loss": 3.334777355194092,
      "step": 121
    },
    {
      "epoch": 0.0196837689577283,
      "grad_norm": 1.8208904266357422,
      "learning_rate": 0.0005,
      "loss": 3.3851518630981445,
      "step": 122
    },
    {
      "epoch": 0.01984511132623427,
      "grad_norm": 1.6407755613327026,
      "learning_rate": 0.0005,
      "loss": 3.3786237239837646,
      "step": 123
    },
    {
      "epoch": 0.020006453694740238,
      "grad_norm": 1.7172393798828125,
      "learning_rate": 0.0005,
      "loss": 3.981109142303467,
      "step": 124
    },
    {
      "epoch": 0.02016779606324621,
      "grad_norm": 1.543595552444458,
      "learning_rate": 0.0005,
      "loss": 4.073455333709717,
      "step": 125
    },
    {
      "epoch": 0.02032913843175218,
      "grad_norm": 1.9170037508010864,
      "learning_rate": 0.0005,
      "loss": 3.9693593978881836,
      "step": 126
    },
    {
      "epoch": 0.020490480800258146,
      "grad_norm": 1.7213956117630005,
      "learning_rate": 0.0005,
      "loss": 3.55720591545105,
      "step": 127
    },
    {
      "epoch": 0.020651823168764117,
      "grad_norm": 1.756934404373169,
      "learning_rate": 0.0005,
      "loss": 3.7489023208618164,
      "step": 128
    },
    {
      "epoch": 0.020813165537270088,
      "grad_norm": 1.9603620767593384,
      "learning_rate": 0.0005,
      "loss": 3.782850742340088,
      "step": 129
    },
    {
      "epoch": 0.020974507905776058,
      "grad_norm": 1.739915132522583,
      "learning_rate": 0.0005,
      "loss": 3.560126781463623,
      "step": 130
    },
    {
      "epoch": 0.021135850274282025,
      "grad_norm": 2.63690447807312,
      "learning_rate": 0.0005,
      "loss": 3.476025104522705,
      "step": 131
    },
    {
      "epoch": 0.021297192642787996,
      "grad_norm": 1.897168755531311,
      "learning_rate": 0.0005,
      "loss": 3.901270627975464,
      "step": 132
    },
    {
      "epoch": 0.021458535011293967,
      "grad_norm": 2.1199967861175537,
      "learning_rate": 0.0005,
      "loss": 3.913398504257202,
      "step": 133
    },
    {
      "epoch": 0.021619877379799937,
      "grad_norm": 2.865140438079834,
      "learning_rate": 0.0005,
      "loss": 3.350311040878296,
      "step": 134
    },
    {
      "epoch": 0.021781219748305904,
      "grad_norm": 1.6121375560760498,
      "learning_rate": 0.0005,
      "loss": 4.117663860321045,
      "step": 135
    },
    {
      "epoch": 0.021942562116811875,
      "grad_norm": 1.6274309158325195,
      "learning_rate": 0.0005,
      "loss": 4.1975555419921875,
      "step": 136
    },
    {
      "epoch": 0.022103904485317846,
      "grad_norm": 1.746893048286438,
      "learning_rate": 0.0005,
      "loss": 3.650538682937622,
      "step": 137
    },
    {
      "epoch": 0.022265246853823813,
      "grad_norm": 1.7156040668487549,
      "learning_rate": 0.0005,
      "loss": 3.3957581520080566,
      "step": 138
    },
    {
      "epoch": 0.022426589222329783,
      "grad_norm": 2.5068325996398926,
      "learning_rate": 0.0005,
      "loss": 3.657660484313965,
      "step": 139
    },
    {
      "epoch": 0.022587931590835754,
      "grad_norm": 2.096181631088257,
      "learning_rate": 0.0005,
      "loss": 4.241362571716309,
      "step": 140
    },
    {
      "epoch": 0.022749273959341725,
      "grad_norm": 1.8788502216339111,
      "learning_rate": 0.0005,
      "loss": 3.233828544616699,
      "step": 141
    },
    {
      "epoch": 0.02291061632784769,
      "grad_norm": 1.6733638048171997,
      "learning_rate": 0.0005,
      "loss": 3.7879374027252197,
      "step": 142
    },
    {
      "epoch": 0.023071958696353662,
      "grad_norm": 1.9382944107055664,
      "learning_rate": 0.0005,
      "loss": 3.638514518737793,
      "step": 143
    },
    {
      "epoch": 0.023233301064859633,
      "grad_norm": 1.7569117546081543,
      "learning_rate": 0.0005,
      "loss": 3.1015994548797607,
      "step": 144
    },
    {
      "epoch": 0.023394643433365604,
      "grad_norm": 1.8348877429962158,
      "learning_rate": 0.0005,
      "loss": 4.108867645263672,
      "step": 145
    },
    {
      "epoch": 0.02355598580187157,
      "grad_norm": 1.5654364824295044,
      "learning_rate": 0.0005,
      "loss": 3.8511407375335693,
      "step": 146
    },
    {
      "epoch": 0.02371732817037754,
      "grad_norm": 1.7872904539108276,
      "learning_rate": 0.0005,
      "loss": 3.6766374111175537,
      "step": 147
    },
    {
      "epoch": 0.023878670538883512,
      "grad_norm": 1.6129177808761597,
      "learning_rate": 0.0005,
      "loss": 3.957709312438965,
      "step": 148
    },
    {
      "epoch": 0.02404001290738948,
      "grad_norm": 1.9307143688201904,
      "learning_rate": 0.0005,
      "loss": 4.173839569091797,
      "step": 149
    },
    {
      "epoch": 0.02420135527589545,
      "grad_norm": 2.295534133911133,
      "learning_rate": 0.0005,
      "loss": 4.022490501403809,
      "step": 150
    },
    {
      "epoch": 0.02436269764440142,
      "grad_norm": 1.890688419342041,
      "learning_rate": 0.0005,
      "loss": 3.701305627822876,
      "step": 151
    },
    {
      "epoch": 0.02452404001290739,
      "grad_norm": 1.6258654594421387,
      "learning_rate": 0.0005,
      "loss": 3.4901442527770996,
      "step": 152
    },
    {
      "epoch": 0.024685382381413358,
      "grad_norm": 1.6912999153137207,
      "learning_rate": 0.0005,
      "loss": 3.7606985569000244,
      "step": 153
    },
    {
      "epoch": 0.02484672474991933,
      "grad_norm": 1.8071328401565552,
      "learning_rate": 0.0005,
      "loss": 4.116213798522949,
      "step": 154
    },
    {
      "epoch": 0.0250080671184253,
      "grad_norm": 1.6707719564437866,
      "learning_rate": 0.0005,
      "loss": 4.257476806640625,
      "step": 155
    },
    {
      "epoch": 0.02516940948693127,
      "grad_norm": 1.3586889505386353,
      "learning_rate": 0.0005,
      "loss": 4.2368059158325195,
      "step": 156
    },
    {
      "epoch": 0.025330751855437237,
      "grad_norm": 1.5064977407455444,
      "learning_rate": 0.0005,
      "loss": 3.6601979732513428,
      "step": 157
    },
    {
      "epoch": 0.025492094223943208,
      "grad_norm": 1.86359441280365,
      "learning_rate": 0.0005,
      "loss": 3.311826705932617,
      "step": 158
    },
    {
      "epoch": 0.025653436592449178,
      "grad_norm": 1.6516469717025757,
      "learning_rate": 0.0005,
      "loss": 3.836104154586792,
      "step": 159
    },
    {
      "epoch": 0.025814778960955145,
      "grad_norm": 2.1211631298065186,
      "learning_rate": 0.0005,
      "loss": 3.809910774230957,
      "step": 160
    },
    {
      "epoch": 0.025976121329461116,
      "grad_norm": 1.8421896696090698,
      "learning_rate": 0.0005,
      "loss": 3.6158621311187744,
      "step": 161
    },
    {
      "epoch": 0.026137463697967087,
      "grad_norm": 1.4786028861999512,
      "learning_rate": 0.0005,
      "loss": 4.280796527862549,
      "step": 162
    },
    {
      "epoch": 0.026298806066473057,
      "grad_norm": 1.503812313079834,
      "learning_rate": 0.0005,
      "loss": 4.098982334136963,
      "step": 163
    },
    {
      "epoch": 0.026460148434979024,
      "grad_norm": 1.658280849456787,
      "learning_rate": 0.0005,
      "loss": 3.786858081817627,
      "step": 164
    },
    {
      "epoch": 0.026621490803484995,
      "grad_norm": 1.9899982213974,
      "learning_rate": 0.0005,
      "loss": 3.790334701538086,
      "step": 165
    },
    {
      "epoch": 0.026782833171990966,
      "grad_norm": 2.0640430450439453,
      "learning_rate": 0.0005,
      "loss": 4.271302223205566,
      "step": 166
    },
    {
      "epoch": 0.026944175540496936,
      "grad_norm": 2.4997222423553467,
      "learning_rate": 0.0005,
      "loss": 4.183464527130127,
      "step": 167
    },
    {
      "epoch": 0.027105517909002903,
      "grad_norm": 1.6380971670150757,
      "learning_rate": 0.0005,
      "loss": 3.9134130477905273,
      "step": 168
    },
    {
      "epoch": 0.027266860277508874,
      "grad_norm": 1.7980656623840332,
      "learning_rate": 0.0005,
      "loss": 3.63045072555542,
      "step": 169
    },
    {
      "epoch": 0.027428202646014845,
      "grad_norm": 2.1523263454437256,
      "learning_rate": 0.0005,
      "loss": 3.808274030685425,
      "step": 170
    },
    {
      "epoch": 0.02758954501452081,
      "grad_norm": 2.307243824005127,
      "learning_rate": 0.0005,
      "loss": 4.095818519592285,
      "step": 171
    },
    {
      "epoch": 0.027750887383026782,
      "grad_norm": 1.6254268884658813,
      "learning_rate": 0.0005,
      "loss": 3.5805773735046387,
      "step": 172
    },
    {
      "epoch": 0.027912229751532753,
      "grad_norm": 1.9915759563446045,
      "learning_rate": 0.0005,
      "loss": 3.4115262031555176,
      "step": 173
    },
    {
      "epoch": 0.028073572120038724,
      "grad_norm": 2.3210620880126953,
      "learning_rate": 0.0005,
      "loss": 3.908936023712158,
      "step": 174
    },
    {
      "epoch": 0.02823491448854469,
      "grad_norm": 1.9246020317077637,
      "learning_rate": 0.0005,
      "loss": 3.4424633979797363,
      "step": 175
    },
    {
      "epoch": 0.02839625685705066,
      "grad_norm": 1.7443593740463257,
      "learning_rate": 0.0005,
      "loss": 4.024843215942383,
      "step": 176
    },
    {
      "epoch": 0.028557599225556632,
      "grad_norm": 1.6009154319763184,
      "learning_rate": 0.0005,
      "loss": 4.226607799530029,
      "step": 177
    },
    {
      "epoch": 0.028718941594062602,
      "grad_norm": 1.6767107248306274,
      "learning_rate": 0.0005,
      "loss": 3.792473077774048,
      "step": 178
    },
    {
      "epoch": 0.02888028396256857,
      "grad_norm": 1.784453272819519,
      "learning_rate": 0.0005,
      "loss": 4.0723772048950195,
      "step": 179
    },
    {
      "epoch": 0.02904162633107454,
      "grad_norm": 1.6870675086975098,
      "learning_rate": 0.0005,
      "loss": 4.080337047576904,
      "step": 180
    },
    {
      "epoch": 0.02920296869958051,
      "grad_norm": 2.0079729557037354,
      "learning_rate": 0.0005,
      "loss": 3.4368934631347656,
      "step": 181
    },
    {
      "epoch": 0.029364311068086478,
      "grad_norm": 1.8040213584899902,
      "learning_rate": 0.0005,
      "loss": 3.8649847507476807,
      "step": 182
    },
    {
      "epoch": 0.02952565343659245,
      "grad_norm": 2.1018006801605225,
      "learning_rate": 0.0005,
      "loss": 3.961228370666504,
      "step": 183
    },
    {
      "epoch": 0.02968699580509842,
      "grad_norm": 1.9402824640274048,
      "learning_rate": 0.0005,
      "loss": 3.694855213165283,
      "step": 184
    },
    {
      "epoch": 0.02984833817360439,
      "grad_norm": 1.6735762357711792,
      "learning_rate": 0.0005,
      "loss": 4.186314582824707,
      "step": 185
    },
    {
      "epoch": 0.030009680542110357,
      "grad_norm": 1.5015850067138672,
      "learning_rate": 0.0005,
      "loss": 4.187116622924805,
      "step": 186
    },
    {
      "epoch": 0.030171022910616328,
      "grad_norm": 1.6218384504318237,
      "learning_rate": 0.0005,
      "loss": 3.890174388885498,
      "step": 187
    },
    {
      "epoch": 0.030332365279122298,
      "grad_norm": 1.8117235898971558,
      "learning_rate": 0.0005,
      "loss": 3.389458417892456,
      "step": 188
    },
    {
      "epoch": 0.03049370764762827,
      "grad_norm": 1.6708970069885254,
      "learning_rate": 0.0005,
      "loss": 3.9820871353149414,
      "step": 189
    },
    {
      "epoch": 0.030655050016134236,
      "grad_norm": 1.833825945854187,
      "learning_rate": 0.0005,
      "loss": 3.7204203605651855,
      "step": 190
    },
    {
      "epoch": 0.030816392384640207,
      "grad_norm": 1.8497483730316162,
      "learning_rate": 0.0005,
      "loss": 3.9803617000579834,
      "step": 191
    },
    {
      "epoch": 0.030977734753146177,
      "grad_norm": 1.7894777059555054,
      "learning_rate": 0.0005,
      "loss": 3.57387113571167,
      "step": 192
    },
    {
      "epoch": 0.031139077121652144,
      "grad_norm": 1.5983452796936035,
      "learning_rate": 0.0005,
      "loss": 4.049222469329834,
      "step": 193
    },
    {
      "epoch": 0.03130041949015812,
      "grad_norm": 1.9456113576889038,
      "learning_rate": 0.0005,
      "loss": 4.034397602081299,
      "step": 194
    },
    {
      "epoch": 0.03146176185866408,
      "grad_norm": 1.7071622610092163,
      "learning_rate": 0.0005,
      "loss": 3.409407615661621,
      "step": 195
    },
    {
      "epoch": 0.03162310422717005,
      "grad_norm": 2.122523784637451,
      "learning_rate": 0.0005,
      "loss": 3.7674593925476074,
      "step": 196
    },
    {
      "epoch": 0.03178444659567602,
      "grad_norm": 2.3791277408599854,
      "learning_rate": 0.0005,
      "loss": 3.4305973052978516,
      "step": 197
    },
    {
      "epoch": 0.031945788964181994,
      "grad_norm": 2.867266893386841,
      "learning_rate": 0.0005,
      "loss": 3.8476009368896484,
      "step": 198
    },
    {
      "epoch": 0.032107131332687965,
      "grad_norm": 1.8219835758209229,
      "learning_rate": 0.0005,
      "loss": 4.117099761962891,
      "step": 199
    },
    {
      "epoch": 0.032268473701193935,
      "grad_norm": 1.7833558320999146,
      "learning_rate": 0.0005,
      "loss": 3.9820005893707275,
      "step": 200
    },
    {
      "epoch": 0.032429816069699906,
      "grad_norm": 1.4711006879806519,
      "learning_rate": 0.0005,
      "loss": 3.9984893798828125,
      "step": 201
    },
    {
      "epoch": 0.03259115843820587,
      "grad_norm": 1.5931620597839355,
      "learning_rate": 0.0005,
      "loss": 4.178135871887207,
      "step": 202
    },
    {
      "epoch": 0.03275250080671184,
      "grad_norm": 1.5871498584747314,
      "learning_rate": 0.0005,
      "loss": 3.6205148696899414,
      "step": 203
    },
    {
      "epoch": 0.03291384317521781,
      "grad_norm": 1.5283159017562866,
      "learning_rate": 0.0005,
      "loss": 4.214707374572754,
      "step": 204
    },
    {
      "epoch": 0.03307518554372378,
      "grad_norm": 1.5045685768127441,
      "learning_rate": 0.0005,
      "loss": 3.568512439727783,
      "step": 205
    },
    {
      "epoch": 0.03323652791222975,
      "grad_norm": 1.6255754232406616,
      "learning_rate": 0.0005,
      "loss": 3.7084548473358154,
      "step": 206
    },
    {
      "epoch": 0.03339787028073572,
      "grad_norm": 1.4374195337295532,
      "learning_rate": 0.0005,
      "loss": 3.4599742889404297,
      "step": 207
    },
    {
      "epoch": 0.03355921264924169,
      "grad_norm": 1.6335958242416382,
      "learning_rate": 0.0005,
      "loss": 3.7004497051239014,
      "step": 208
    },
    {
      "epoch": 0.033720555017747664,
      "grad_norm": 1.6259132623672485,
      "learning_rate": 0.0005,
      "loss": 4.146002769470215,
      "step": 209
    },
    {
      "epoch": 0.03388189738625363,
      "grad_norm": 1.7281410694122314,
      "learning_rate": 0.0005,
      "loss": 3.456839084625244,
      "step": 210
    },
    {
      "epoch": 0.0340432397547596,
      "grad_norm": 2.4081923961639404,
      "learning_rate": 0.0005,
      "loss": 3.6289494037628174,
      "step": 211
    },
    {
      "epoch": 0.03420458212326557,
      "grad_norm": 1.8745523691177368,
      "learning_rate": 0.0005,
      "loss": 3.6371381282806396,
      "step": 212
    },
    {
      "epoch": 0.03436592449177154,
      "grad_norm": 1.59743332862854,
      "learning_rate": 0.0005,
      "loss": 4.119265556335449,
      "step": 213
    },
    {
      "epoch": 0.03452726686027751,
      "grad_norm": 2.2122111320495605,
      "learning_rate": 0.0005,
      "loss": 3.9407331943511963,
      "step": 214
    },
    {
      "epoch": 0.03468860922878348,
      "grad_norm": 1.463703989982605,
      "learning_rate": 0.0005,
      "loss": 3.6649012565612793,
      "step": 215
    },
    {
      "epoch": 0.03484995159728945,
      "grad_norm": 1.6332275867462158,
      "learning_rate": 0.0005,
      "loss": 3.6755218505859375,
      "step": 216
    },
    {
      "epoch": 0.035011293965795415,
      "grad_norm": 1.9339878559112549,
      "learning_rate": 0.0005,
      "loss": 3.6306004524230957,
      "step": 217
    },
    {
      "epoch": 0.035172636334301385,
      "grad_norm": 1.567100167274475,
      "learning_rate": 0.0005,
      "loss": 4.0915703773498535,
      "step": 218
    },
    {
      "epoch": 0.035333978702807356,
      "grad_norm": 2.10530424118042,
      "learning_rate": 0.0005,
      "loss": 4.287796974182129,
      "step": 219
    },
    {
      "epoch": 0.03549532107131333,
      "grad_norm": 1.8070707321166992,
      "learning_rate": 0.0005,
      "loss": 3.449542284011841,
      "step": 220
    },
    {
      "epoch": 0.0356566634398193,
      "grad_norm": 3.3603150844573975,
      "learning_rate": 0.0005,
      "loss": 3.924975872039795,
      "step": 221
    },
    {
      "epoch": 0.03581800580832527,
      "grad_norm": 1.5986071825027466,
      "learning_rate": 0.0005,
      "loss": 3.6453282833099365,
      "step": 222
    },
    {
      "epoch": 0.03597934817683124,
      "grad_norm": 1.4432514905929565,
      "learning_rate": 0.0005,
      "loss": 3.8869683742523193,
      "step": 223
    },
    {
      "epoch": 0.0361406905453372,
      "grad_norm": 2.061307907104492,
      "learning_rate": 0.0005,
      "loss": 4.101002216339111,
      "step": 224
    },
    {
      "epoch": 0.03630203291384317,
      "grad_norm": 2.0730297565460205,
      "learning_rate": 0.0005,
      "loss": 3.6507039070129395,
      "step": 225
    },
    {
      "epoch": 0.03646337528234914,
      "grad_norm": 1.9362785816192627,
      "learning_rate": 0.0005,
      "loss": 3.661008358001709,
      "step": 226
    },
    {
      "epoch": 0.036624717650855114,
      "grad_norm": 2.1007440090179443,
      "learning_rate": 0.0005,
      "loss": 3.3823423385620117,
      "step": 227
    },
    {
      "epoch": 0.036786060019361085,
      "grad_norm": 1.435632586479187,
      "learning_rate": 0.0005,
      "loss": 3.611876964569092,
      "step": 228
    },
    {
      "epoch": 0.036947402387867055,
      "grad_norm": 2.2104461193084717,
      "learning_rate": 0.0005,
      "loss": 3.9610462188720703,
      "step": 229
    },
    {
      "epoch": 0.037108744756373026,
      "grad_norm": 1.5415838956832886,
      "learning_rate": 0.0005,
      "loss": 3.982112169265747,
      "step": 230
    },
    {
      "epoch": 0.037270087124878996,
      "grad_norm": 2.0163464546203613,
      "learning_rate": 0.0005,
      "loss": 3.4960389137268066,
      "step": 231
    },
    {
      "epoch": 0.03743142949338496,
      "grad_norm": 1.9255237579345703,
      "learning_rate": 0.0005,
      "loss": 3.8962574005126953,
      "step": 232
    },
    {
      "epoch": 0.03759277186189093,
      "grad_norm": 1.8190158605575562,
      "learning_rate": 0.0005,
      "loss": 3.9270708560943604,
      "step": 233
    },
    {
      "epoch": 0.0377541142303969,
      "grad_norm": 1.774232268333435,
      "learning_rate": 0.0005,
      "loss": 3.815530300140381,
      "step": 234
    },
    {
      "epoch": 0.03791545659890287,
      "grad_norm": 2.0597827434539795,
      "learning_rate": 0.0005,
      "loss": 4.301422119140625,
      "step": 235
    },
    {
      "epoch": 0.03807679896740884,
      "grad_norm": 2.1982808113098145,
      "learning_rate": 0.0005,
      "loss": 3.9013657569885254,
      "step": 236
    },
    {
      "epoch": 0.03823814133591481,
      "grad_norm": 1.826454997062683,
      "learning_rate": 0.0005,
      "loss": 3.7357003688812256,
      "step": 237
    },
    {
      "epoch": 0.038399483704420784,
      "grad_norm": 2.9696407318115234,
      "learning_rate": 0.0005,
      "loss": 3.209362268447876,
      "step": 238
    },
    {
      "epoch": 0.03856082607292675,
      "grad_norm": 1.7195324897766113,
      "learning_rate": 0.0005,
      "loss": 3.847287178039551,
      "step": 239
    },
    {
      "epoch": 0.03872216844143272,
      "grad_norm": 1.6519323587417603,
      "learning_rate": 0.0005,
      "loss": 3.8222196102142334,
      "step": 240
    },
    {
      "epoch": 0.03888351080993869,
      "grad_norm": 2.505284309387207,
      "learning_rate": 0.0005,
      "loss": 3.816159248352051,
      "step": 241
    },
    {
      "epoch": 0.03904485317844466,
      "grad_norm": 1.6136398315429688,
      "learning_rate": 0.0005,
      "loss": 3.7256970405578613,
      "step": 242
    },
    {
      "epoch": 0.03920619554695063,
      "grad_norm": 1.707737922668457,
      "learning_rate": 0.0005,
      "loss": 3.803673505783081,
      "step": 243
    },
    {
      "epoch": 0.0393675379154566,
      "grad_norm": 2.253389596939087,
      "learning_rate": 0.0005,
      "loss": 3.9261391162872314,
      "step": 244
    },
    {
      "epoch": 0.03952888028396257,
      "grad_norm": 2.635683059692383,
      "learning_rate": 0.0005,
      "loss": 4.3557233810424805,
      "step": 245
    },
    {
      "epoch": 0.03969022265246854,
      "grad_norm": 1.7862495183944702,
      "learning_rate": 0.0005,
      "loss": 3.6516001224517822,
      "step": 246
    },
    {
      "epoch": 0.039851565020974505,
      "grad_norm": 2.2561585903167725,
      "learning_rate": 0.0005,
      "loss": 3.853353500366211,
      "step": 247
    },
    {
      "epoch": 0.040012907389480476,
      "grad_norm": 2.068510055541992,
      "learning_rate": 0.0005,
      "loss": 3.479156017303467,
      "step": 248
    },
    {
      "epoch": 0.04017424975798645,
      "grad_norm": 1.7385543584823608,
      "learning_rate": 0.0005,
      "loss": 3.753164291381836,
      "step": 249
    },
    {
      "epoch": 0.04033559212649242,
      "grad_norm": 1.9857534170150757,
      "learning_rate": 0.0005,
      "loss": 3.802302837371826,
      "step": 250
    },
    {
      "epoch": 0.04049693449499839,
      "grad_norm": 3.0880508422851562,
      "learning_rate": 0.0005,
      "loss": 3.1842408180236816,
      "step": 251
    },
    {
      "epoch": 0.04065827686350436,
      "grad_norm": 1.5290542840957642,
      "learning_rate": 0.0005,
      "loss": 3.6561427116394043,
      "step": 252
    },
    {
      "epoch": 0.04081961923201033,
      "grad_norm": 1.8543565273284912,
      "learning_rate": 0.0005,
      "loss": 3.5847177505493164,
      "step": 253
    },
    {
      "epoch": 0.04098096160051629,
      "grad_norm": 1.4389320611953735,
      "learning_rate": 0.0005,
      "loss": 4.18020486831665,
      "step": 254
    },
    {
      "epoch": 0.04114230396902226,
      "grad_norm": 1.3794807195663452,
      "learning_rate": 0.0005,
      "loss": 3.644705295562744,
      "step": 255
    },
    {
      "epoch": 0.041303646337528234,
      "grad_norm": 1.8779358863830566,
      "learning_rate": 0.0005,
      "loss": 3.7088165283203125,
      "step": 256
    },
    {
      "epoch": 0.041464988706034205,
      "grad_norm": 2.76206374168396,
      "learning_rate": 0.0005,
      "loss": 3.860353946685791,
      "step": 257
    },
    {
      "epoch": 0.041626331074540175,
      "grad_norm": 1.6648705005645752,
      "learning_rate": 0.0005,
      "loss": 3.985262870788574,
      "step": 258
    },
    {
      "epoch": 0.041787673443046146,
      "grad_norm": 3.2317733764648438,
      "learning_rate": 0.0005,
      "loss": 3.6793856620788574,
      "step": 259
    },
    {
      "epoch": 0.041949015811552116,
      "grad_norm": 1.9054064750671387,
      "learning_rate": 0.0005,
      "loss": 3.751638650894165,
      "step": 260
    },
    {
      "epoch": 0.04211035818005808,
      "grad_norm": 1.4695699214935303,
      "learning_rate": 0.0005,
      "loss": 3.840585708618164,
      "step": 261
    },
    {
      "epoch": 0.04227170054856405,
      "grad_norm": 1.509694218635559,
      "learning_rate": 0.0005,
      "loss": 4.1636061668396,
      "step": 262
    },
    {
      "epoch": 0.04243304291707002,
      "grad_norm": 2.0882227420806885,
      "learning_rate": 0.0005,
      "loss": 3.904690742492676,
      "step": 263
    },
    {
      "epoch": 0.04259438528557599,
      "grad_norm": 2.263718366622925,
      "learning_rate": 0.0005,
      "loss": 3.7639834880828857,
      "step": 264
    },
    {
      "epoch": 0.04275572765408196,
      "grad_norm": 1.6825648546218872,
      "learning_rate": 0.0005,
      "loss": 3.463143825531006,
      "step": 265
    },
    {
      "epoch": 0.04291707002258793,
      "grad_norm": 1.8113653659820557,
      "learning_rate": 0.0005,
      "loss": 3.182161331176758,
      "step": 266
    },
    {
      "epoch": 0.043078412391093904,
      "grad_norm": 1.8139724731445312,
      "learning_rate": 0.0005,
      "loss": 4.190359115600586,
      "step": 267
    },
    {
      "epoch": 0.043239754759599874,
      "grad_norm": 1.6005282402038574,
      "learning_rate": 0.0005,
      "loss": 3.9513180255889893,
      "step": 268
    },
    {
      "epoch": 0.04340109712810584,
      "grad_norm": 1.551796317100525,
      "learning_rate": 0.0005,
      "loss": 3.855140447616577,
      "step": 269
    },
    {
      "epoch": 0.04356243949661181,
      "grad_norm": 1.686972737312317,
      "learning_rate": 0.0005,
      "loss": 4.173229217529297,
      "step": 270
    },
    {
      "epoch": 0.04372378186511778,
      "grad_norm": 1.6105238199234009,
      "learning_rate": 0.0005,
      "loss": 3.5884220600128174,
      "step": 271
    },
    {
      "epoch": 0.04388512423362375,
      "grad_norm": 2.5123045444488525,
      "learning_rate": 0.0005,
      "loss": 3.978445529937744,
      "step": 272
    },
    {
      "epoch": 0.04404646660212972,
      "grad_norm": 1.7388246059417725,
      "learning_rate": 0.0005,
      "loss": 3.7986879348754883,
      "step": 273
    },
    {
      "epoch": 0.04420780897063569,
      "grad_norm": 1.7773470878601074,
      "learning_rate": 0.0005,
      "loss": 3.68320894241333,
      "step": 274
    },
    {
      "epoch": 0.04436915133914166,
      "grad_norm": 2.386969566345215,
      "learning_rate": 0.0005,
      "loss": 4.087874412536621,
      "step": 275
    },
    {
      "epoch": 0.044530493707647625,
      "grad_norm": 1.9948956966400146,
      "learning_rate": 0.0005,
      "loss": 3.4420294761657715,
      "step": 276
    },
    {
      "epoch": 0.044691836076153596,
      "grad_norm": 2.1509783267974854,
      "learning_rate": 0.0005,
      "loss": 3.6991071701049805,
      "step": 277
    },
    {
      "epoch": 0.04485317844465957,
      "grad_norm": 1.9978688955307007,
      "learning_rate": 0.0005,
      "loss": 3.5018293857574463,
      "step": 278
    },
    {
      "epoch": 0.04501452081316554,
      "grad_norm": 1.4840238094329834,
      "learning_rate": 0.0005,
      "loss": 3.6428933143615723,
      "step": 279
    },
    {
      "epoch": 0.04517586318167151,
      "grad_norm": 2.621103286743164,
      "learning_rate": 0.0005,
      "loss": 3.486868381500244,
      "step": 280
    },
    {
      "epoch": 0.04533720555017748,
      "grad_norm": 1.7493524551391602,
      "learning_rate": 0.0005,
      "loss": 3.587141990661621,
      "step": 281
    },
    {
      "epoch": 0.04549854791868345,
      "grad_norm": 2.057356357574463,
      "learning_rate": 0.0005,
      "loss": 4.156789779663086,
      "step": 282
    },
    {
      "epoch": 0.04565989028718941,
      "grad_norm": 2.1833267211914062,
      "learning_rate": 0.0005,
      "loss": 3.498016119003296,
      "step": 283
    },
    {
      "epoch": 0.04582123265569538,
      "grad_norm": 3.3869223594665527,
      "learning_rate": 0.0005,
      "loss": 3.673426628112793,
      "step": 284
    },
    {
      "epoch": 0.045982575024201354,
      "grad_norm": 1.6718190908432007,
      "learning_rate": 0.0005,
      "loss": 3.559427499771118,
      "step": 285
    },
    {
      "epoch": 0.046143917392707325,
      "grad_norm": 1.4983912706375122,
      "learning_rate": 0.0005,
      "loss": 3.715085983276367,
      "step": 286
    },
    {
      "epoch": 0.046305259761213295,
      "grad_norm": 1.9504854679107666,
      "learning_rate": 0.0005,
      "loss": 4.118968963623047,
      "step": 287
    },
    {
      "epoch": 0.046466602129719266,
      "grad_norm": 3.8117477893829346,
      "learning_rate": 0.0005,
      "loss": 3.553724765777588,
      "step": 288
    },
    {
      "epoch": 0.046627944498225236,
      "grad_norm": 2.0613934993743896,
      "learning_rate": 0.0005,
      "loss": 3.9402527809143066,
      "step": 289
    },
    {
      "epoch": 0.04678928686673121,
      "grad_norm": 2.3512818813323975,
      "learning_rate": 0.0005,
      "loss": 3.7405006885528564,
      "step": 290
    },
    {
      "epoch": 0.04695062923523717,
      "grad_norm": 1.8999593257904053,
      "learning_rate": 0.0005,
      "loss": 3.499368906021118,
      "step": 291
    },
    {
      "epoch": 0.04711197160374314,
      "grad_norm": 1.9647417068481445,
      "learning_rate": 0.0005,
      "loss": 3.719339370727539,
      "step": 292
    },
    {
      "epoch": 0.04727331397224911,
      "grad_norm": 2.1547911167144775,
      "learning_rate": 0.0005,
      "loss": 3.8355154991149902,
      "step": 293
    },
    {
      "epoch": 0.04743465634075508,
      "grad_norm": 1.791892409324646,
      "learning_rate": 0.0005,
      "loss": 3.7113237380981445,
      "step": 294
    },
    {
      "epoch": 0.04759599870926105,
      "grad_norm": 1.5961365699768066,
      "learning_rate": 0.0005,
      "loss": 3.6573426723480225,
      "step": 295
    },
    {
      "epoch": 0.047757341077767024,
      "grad_norm": 1.688428521156311,
      "learning_rate": 0.0005,
      "loss": 3.594708204269409,
      "step": 296
    },
    {
      "epoch": 0.047918683446272994,
      "grad_norm": 1.6568124294281006,
      "learning_rate": 0.0005,
      "loss": 4.1178879737854,
      "step": 297
    },
    {
      "epoch": 0.04808002581477896,
      "grad_norm": 1.7708876132965088,
      "learning_rate": 0.0005,
      "loss": 4.180855751037598,
      "step": 298
    },
    {
      "epoch": 0.04824136818328493,
      "grad_norm": 1.9019243717193604,
      "learning_rate": 0.0005,
      "loss": 3.767639398574829,
      "step": 299
    },
    {
      "epoch": 0.0484027105517909,
      "grad_norm": 1.7042697668075562,
      "learning_rate": 0.0005,
      "loss": 3.5656840801239014,
      "step": 300
    },
    {
      "epoch": 0.04856405292029687,
      "grad_norm": 1.7852646112442017,
      "learning_rate": 0.0005,
      "loss": 3.637722969055176,
      "step": 301
    },
    {
      "epoch": 0.04872539528880284,
      "grad_norm": 1.711585283279419,
      "learning_rate": 0.0005,
      "loss": 3.9480371475219727,
      "step": 302
    },
    {
      "epoch": 0.04888673765730881,
      "grad_norm": 1.7569230794906616,
      "learning_rate": 0.0005,
      "loss": 3.7103915214538574,
      "step": 303
    },
    {
      "epoch": 0.04904808002581478,
      "grad_norm": 1.6070717573165894,
      "learning_rate": 0.0005,
      "loss": 3.9377267360687256,
      "step": 304
    },
    {
      "epoch": 0.049209422394320745,
      "grad_norm": 1.9685367345809937,
      "learning_rate": 0.0005,
      "loss": 3.5023016929626465,
      "step": 305
    },
    {
      "epoch": 0.049370764762826716,
      "grad_norm": 1.7400538921356201,
      "learning_rate": 0.0005,
      "loss": 3.521878719329834,
      "step": 306
    },
    {
      "epoch": 0.04953210713133269,
      "grad_norm": 1.473921298980713,
      "learning_rate": 0.0005,
      "loss": 3.710885763168335,
      "step": 307
    },
    {
      "epoch": 0.04969344949983866,
      "grad_norm": 2.0788097381591797,
      "learning_rate": 0.0005,
      "loss": 3.7947475910186768,
      "step": 308
    },
    {
      "epoch": 0.04985479186834463,
      "grad_norm": 1.5637671947479248,
      "learning_rate": 0.0005,
      "loss": 3.7966156005859375,
      "step": 309
    },
    {
      "epoch": 0.0500161342368506,
      "grad_norm": 1.4650269746780396,
      "learning_rate": 0.0005,
      "loss": 3.972667932510376,
      "step": 310
    },
    {
      "epoch": 0.05017747660535657,
      "grad_norm": 1.8886820077896118,
      "learning_rate": 0.0005,
      "loss": 3.774378776550293,
      "step": 311
    },
    {
      "epoch": 0.05033881897386254,
      "grad_norm": 1.5701028108596802,
      "learning_rate": 0.0005,
      "loss": 3.897712230682373,
      "step": 312
    },
    {
      "epoch": 0.0505001613423685,
      "grad_norm": 1.8818471431732178,
      "learning_rate": 0.0005,
      "loss": 3.405470609664917,
      "step": 313
    },
    {
      "epoch": 0.050661503710874474,
      "grad_norm": 2.9136219024658203,
      "learning_rate": 0.0005,
      "loss": 3.638857126235962,
      "step": 314
    },
    {
      "epoch": 0.050822846079380445,
      "grad_norm": 1.6812776327133179,
      "learning_rate": 0.0005,
      "loss": 3.7910683155059814,
      "step": 315
    },
    {
      "epoch": 0.050984188447886415,
      "grad_norm": 1.8165885210037231,
      "learning_rate": 0.0005,
      "loss": 3.5073957443237305,
      "step": 316
    },
    {
      "epoch": 0.051145530816392386,
      "grad_norm": 1.9092568159103394,
      "learning_rate": 0.0005,
      "loss": 3.8197669982910156,
      "step": 317
    },
    {
      "epoch": 0.051306873184898356,
      "grad_norm": 1.432987093925476,
      "learning_rate": 0.0005,
      "loss": 3.636195659637451,
      "step": 318
    },
    {
      "epoch": 0.05146821555340433,
      "grad_norm": 1.7322067022323608,
      "learning_rate": 0.0005,
      "loss": 3.257153034210205,
      "step": 319
    },
    {
      "epoch": 0.05162955792191029,
      "grad_norm": 2.047982692718506,
      "learning_rate": 0.0005,
      "loss": 4.03449010848999,
      "step": 320
    },
    {
      "epoch": 0.05179090029041626,
      "grad_norm": 1.9375005960464478,
      "learning_rate": 0.0005,
      "loss": 3.6620078086853027,
      "step": 321
    },
    {
      "epoch": 0.05195224265892223,
      "grad_norm": 1.7410407066345215,
      "learning_rate": 0.0005,
      "loss": 3.612645149230957,
      "step": 322
    },
    {
      "epoch": 0.0521135850274282,
      "grad_norm": 1.5330657958984375,
      "learning_rate": 0.0005,
      "loss": 3.9547297954559326,
      "step": 323
    },
    {
      "epoch": 0.05227492739593417,
      "grad_norm": 1.44402277469635,
      "learning_rate": 0.0005,
      "loss": 4.216649055480957,
      "step": 324
    },
    {
      "epoch": 0.052436269764440144,
      "grad_norm": 1.5100005865097046,
      "learning_rate": 0.0005,
      "loss": 3.493849277496338,
      "step": 325
    },
    {
      "epoch": 0.052597612132946114,
      "grad_norm": 3.4212656021118164,
      "learning_rate": 0.0005,
      "loss": 3.4669933319091797,
      "step": 326
    },
    {
      "epoch": 0.05275895450145208,
      "grad_norm": 2.001523971557617,
      "learning_rate": 0.0005,
      "loss": 3.722991466522217,
      "step": 327
    },
    {
      "epoch": 0.05292029686995805,
      "grad_norm": 2.1505284309387207,
      "learning_rate": 0.0005,
      "loss": 3.7787065505981445,
      "step": 328
    },
    {
      "epoch": 0.05308163923846402,
      "grad_norm": 1.7350083589553833,
      "learning_rate": 0.0005,
      "loss": 3.7946553230285645,
      "step": 329
    },
    {
      "epoch": 0.05324298160696999,
      "grad_norm": 1.6456412076950073,
      "learning_rate": 0.0005,
      "loss": 3.6094534397125244,
      "step": 330
    },
    {
      "epoch": 0.05340432397547596,
      "grad_norm": 2.1436634063720703,
      "learning_rate": 0.0005,
      "loss": 3.9451346397399902,
      "step": 331
    },
    {
      "epoch": 0.05356566634398193,
      "grad_norm": 1.9364928007125854,
      "learning_rate": 0.0005,
      "loss": 3.633387565612793,
      "step": 332
    },
    {
      "epoch": 0.0537270087124879,
      "grad_norm": 1.6823105812072754,
      "learning_rate": 0.0005,
      "loss": 3.9615442752838135,
      "step": 333
    },
    {
      "epoch": 0.05388835108099387,
      "grad_norm": 1.769943118095398,
      "learning_rate": 0.0005,
      "loss": 4.236525535583496,
      "step": 334
    },
    {
      "epoch": 0.054049693449499836,
      "grad_norm": 1.6915085315704346,
      "learning_rate": 0.0005,
      "loss": 3.3854401111602783,
      "step": 335
    },
    {
      "epoch": 0.05421103581800581,
      "grad_norm": 2.040456533432007,
      "learning_rate": 0.0005,
      "loss": 3.814945697784424,
      "step": 336
    },
    {
      "epoch": 0.05437237818651178,
      "grad_norm": 1.666247010231018,
      "learning_rate": 0.0005,
      "loss": 3.7808804512023926,
      "step": 337
    },
    {
      "epoch": 0.05453372055501775,
      "grad_norm": 1.894027590751648,
      "learning_rate": 0.0005,
      "loss": 3.5893466472625732,
      "step": 338
    },
    {
      "epoch": 0.05469506292352372,
      "grad_norm": 1.5716091394424438,
      "learning_rate": 0.0005,
      "loss": 4.07357120513916,
      "step": 339
    },
    {
      "epoch": 0.05485640529202969,
      "grad_norm": 1.8439245223999023,
      "learning_rate": 0.0005,
      "loss": 3.662179946899414,
      "step": 340
    },
    {
      "epoch": 0.05501774766053566,
      "grad_norm": 1.6969674825668335,
      "learning_rate": 0.0005,
      "loss": 3.6851675510406494,
      "step": 341
    },
    {
      "epoch": 0.05517909002904162,
      "grad_norm": 1.4925774335861206,
      "learning_rate": 0.0005,
      "loss": 3.7759077548980713,
      "step": 342
    },
    {
      "epoch": 0.055340432397547594,
      "grad_norm": 1.8306593894958496,
      "learning_rate": 0.0005,
      "loss": 3.7302236557006836,
      "step": 343
    },
    {
      "epoch": 0.055501774766053565,
      "grad_norm": 1.7572100162506104,
      "learning_rate": 0.0005,
      "loss": 3.4793455600738525,
      "step": 344
    },
    {
      "epoch": 0.055663117134559535,
      "grad_norm": 1.6993207931518555,
      "learning_rate": 0.0005,
      "loss": 3.533000946044922,
      "step": 345
    },
    {
      "epoch": 0.055824459503065506,
      "grad_norm": 1.6014456748962402,
      "learning_rate": 0.0005,
      "loss": 4.166924476623535,
      "step": 346
    },
    {
      "epoch": 0.055985801871571476,
      "grad_norm": 1.7405414581298828,
      "learning_rate": 0.0005,
      "loss": 3.78293776512146,
      "step": 347
    },
    {
      "epoch": 0.05614714424007745,
      "grad_norm": 1.6690726280212402,
      "learning_rate": 0.0005,
      "loss": 3.5109612941741943,
      "step": 348
    },
    {
      "epoch": 0.05630848660858341,
      "grad_norm": 1.6238971948623657,
      "learning_rate": 0.0005,
      "loss": 3.7581167221069336,
      "step": 349
    },
    {
      "epoch": 0.05646982897708938,
      "grad_norm": 1.4104397296905518,
      "learning_rate": 0.0005,
      "loss": 3.755502223968506,
      "step": 350
    },
    {
      "epoch": 0.05663117134559535,
      "grad_norm": 1.8763809204101562,
      "learning_rate": 0.0005,
      "loss": 3.505098342895508,
      "step": 351
    },
    {
      "epoch": 0.05679251371410132,
      "grad_norm": 1.9442284107208252,
      "learning_rate": 0.0005,
      "loss": 3.608780860900879,
      "step": 352
    },
    {
      "epoch": 0.05695385608260729,
      "grad_norm": 1.9033323526382446,
      "learning_rate": 0.0005,
      "loss": 4.045020580291748,
      "step": 353
    },
    {
      "epoch": 0.057115198451113264,
      "grad_norm": 1.850993275642395,
      "learning_rate": 0.0005,
      "loss": 3.923877000808716,
      "step": 354
    },
    {
      "epoch": 0.057276540819619234,
      "grad_norm": 2.227093458175659,
      "learning_rate": 0.0005,
      "loss": 3.650355815887451,
      "step": 355
    },
    {
      "epoch": 0.057437883188125205,
      "grad_norm": 2.402878999710083,
      "learning_rate": 0.0005,
      "loss": 4.015517234802246,
      "step": 356
    },
    {
      "epoch": 0.05759922555663117,
      "grad_norm": 2.862846851348877,
      "learning_rate": 0.0005,
      "loss": 4.258655071258545,
      "step": 357
    },
    {
      "epoch": 0.05776056792513714,
      "grad_norm": 1.3751212358474731,
      "learning_rate": 0.0005,
      "loss": 4.248862266540527,
      "step": 358
    },
    {
      "epoch": 0.05792191029364311,
      "grad_norm": 1.634464979171753,
      "learning_rate": 0.0005,
      "loss": 3.54984188079834,
      "step": 359
    },
    {
      "epoch": 0.05808325266214908,
      "grad_norm": 1.9893933534622192,
      "learning_rate": 0.0005,
      "loss": 3.6372809410095215,
      "step": 360
    },
    {
      "epoch": 0.05824459503065505,
      "grad_norm": 1.670742392539978,
      "learning_rate": 0.0005,
      "loss": 3.5903754234313965,
      "step": 361
    },
    {
      "epoch": 0.05840593739916102,
      "grad_norm": 1.7198092937469482,
      "learning_rate": 0.0005,
      "loss": 3.5649795532226562,
      "step": 362
    },
    {
      "epoch": 0.05856727976766699,
      "grad_norm": 1.9534143209457397,
      "learning_rate": 0.0005,
      "loss": 3.4893975257873535,
      "step": 363
    },
    {
      "epoch": 0.058728622136172956,
      "grad_norm": 1.5516530275344849,
      "learning_rate": 0.0005,
      "loss": 3.843825340270996,
      "step": 364
    },
    {
      "epoch": 0.05888996450467893,
      "grad_norm": 1.5775654315948486,
      "learning_rate": 0.0005,
      "loss": 3.843435764312744,
      "step": 365
    },
    {
      "epoch": 0.0590513068731849,
      "grad_norm": 2.2160377502441406,
      "learning_rate": 0.0005,
      "loss": 3.590217113494873,
      "step": 366
    },
    {
      "epoch": 0.05921264924169087,
      "grad_norm": 1.6312971115112305,
      "learning_rate": 0.0005,
      "loss": 3.430457353591919,
      "step": 367
    },
    {
      "epoch": 0.05937399161019684,
      "grad_norm": 1.505373239517212,
      "learning_rate": 0.0005,
      "loss": 4.135183334350586,
      "step": 368
    },
    {
      "epoch": 0.05953533397870281,
      "grad_norm": 2.0283946990966797,
      "learning_rate": 0.0005,
      "loss": 3.62835693359375,
      "step": 369
    },
    {
      "epoch": 0.05969667634720878,
      "grad_norm": 2.094789743423462,
      "learning_rate": 0.0005,
      "loss": 3.1189494132995605,
      "step": 370
    },
    {
      "epoch": 0.05985801871571474,
      "grad_norm": 1.5052175521850586,
      "learning_rate": 0.0005,
      "loss": 3.9221012592315674,
      "step": 371
    },
    {
      "epoch": 0.060019361084220714,
      "grad_norm": 1.5928525924682617,
      "learning_rate": 0.0005,
      "loss": 3.6716413497924805,
      "step": 372
    },
    {
      "epoch": 0.060180703452726685,
      "grad_norm": 1.2986866235733032,
      "learning_rate": 0.0005,
      "loss": 3.9881255626678467,
      "step": 373
    },
    {
      "epoch": 0.060342045821232655,
      "grad_norm": 1.5519121885299683,
      "learning_rate": 0.0005,
      "loss": 3.420689105987549,
      "step": 374
    },
    {
      "epoch": 0.060503388189738626,
      "grad_norm": 1.4720865488052368,
      "learning_rate": 0.0005,
      "loss": 4.004240036010742,
      "step": 375
    },
    {
      "epoch": 0.060664730558244596,
      "grad_norm": 1.5103728771209717,
      "learning_rate": 0.0005,
      "loss": 3.685181140899658,
      "step": 376
    },
    {
      "epoch": 0.06082607292675057,
      "grad_norm": 1.7732645273208618,
      "learning_rate": 0.0005,
      "loss": 3.468315839767456,
      "step": 377
    },
    {
      "epoch": 0.06098741529525654,
      "grad_norm": 1.6722220182418823,
      "learning_rate": 0.0005,
      "loss": 3.7526607513427734,
      "step": 378
    },
    {
      "epoch": 0.0611487576637625,
      "grad_norm": 1.689052700996399,
      "learning_rate": 0.0005,
      "loss": 3.7573251724243164,
      "step": 379
    },
    {
      "epoch": 0.06131010003226847,
      "grad_norm": 2.003558397293091,
      "learning_rate": 0.0005,
      "loss": 3.7703962326049805,
      "step": 380
    },
    {
      "epoch": 0.06147144240077444,
      "grad_norm": 1.6686861515045166,
      "learning_rate": 0.0005,
      "loss": 4.00231409072876,
      "step": 381
    },
    {
      "epoch": 0.06163278476928041,
      "grad_norm": 1.6637336015701294,
      "learning_rate": 0.0005,
      "loss": 3.6000723838806152,
      "step": 382
    },
    {
      "epoch": 0.061794127137786384,
      "grad_norm": 3.3742241859436035,
      "learning_rate": 0.0005,
      "loss": 3.6870667934417725,
      "step": 383
    },
    {
      "epoch": 0.061955469506292354,
      "grad_norm": 2.263453960418701,
      "learning_rate": 0.0005,
      "loss": 4.426929950714111,
      "step": 384
    },
    {
      "epoch": 0.062116811874798325,
      "grad_norm": 1.399986982345581,
      "learning_rate": 0.0005,
      "loss": 3.914943218231201,
      "step": 385
    },
    {
      "epoch": 0.06227815424330429,
      "grad_norm": 2.899787425994873,
      "learning_rate": 0.0005,
      "loss": 4.17704963684082,
      "step": 386
    },
    {
      "epoch": 0.06243949661181026,
      "grad_norm": 2.055629014968872,
      "learning_rate": 0.0005,
      "loss": 3.5643839836120605,
      "step": 387
    },
    {
      "epoch": 0.06260083898031624,
      "grad_norm": 1.4206058979034424,
      "learning_rate": 0.0005,
      "loss": 3.521569013595581,
      "step": 388
    },
    {
      "epoch": 0.06276218134882221,
      "grad_norm": 1.79546320438385,
      "learning_rate": 0.0005,
      "loss": 4.058502197265625,
      "step": 389
    },
    {
      "epoch": 0.06292352371732816,
      "grad_norm": 2.7066709995269775,
      "learning_rate": 0.0005,
      "loss": 3.5593128204345703,
      "step": 390
    },
    {
      "epoch": 0.06308486608583413,
      "grad_norm": 1.5680298805236816,
      "learning_rate": 0.0005,
      "loss": 3.6083478927612305,
      "step": 391
    },
    {
      "epoch": 0.0632462084543401,
      "grad_norm": 1.5575255155563354,
      "learning_rate": 0.0005,
      "loss": 3.433809757232666,
      "step": 392
    },
    {
      "epoch": 0.06340755082284608,
      "grad_norm": 1.8930593729019165,
      "learning_rate": 0.0005,
      "loss": 3.825043201446533,
      "step": 393
    },
    {
      "epoch": 0.06356889319135205,
      "grad_norm": 1.58977210521698,
      "learning_rate": 0.0005,
      "loss": 3.681983470916748,
      "step": 394
    },
    {
      "epoch": 0.06373023555985802,
      "grad_norm": 1.7598522901535034,
      "learning_rate": 0.0005,
      "loss": 3.780383825302124,
      "step": 395
    },
    {
      "epoch": 0.06389157792836399,
      "grad_norm": 1.675121784210205,
      "learning_rate": 0.0005,
      "loss": 3.6066410541534424,
      "step": 396
    },
    {
      "epoch": 0.06405292029686996,
      "grad_norm": 1.3700097799301147,
      "learning_rate": 0.0005,
      "loss": 4.271283149719238,
      "step": 397
    },
    {
      "epoch": 0.06421426266537593,
      "grad_norm": 1.2710374593734741,
      "learning_rate": 0.0005,
      "loss": 4.145166397094727,
      "step": 398
    },
    {
      "epoch": 0.0643756050338819,
      "grad_norm": 1.5598262548446655,
      "learning_rate": 0.0005,
      "loss": 3.5572447776794434,
      "step": 399
    },
    {
      "epoch": 0.06453694740238787,
      "grad_norm": 2.7071433067321777,
      "learning_rate": 0.0005,
      "loss": 4.055405616760254,
      "step": 400
    },
    {
      "epoch": 0.06469828977089384,
      "grad_norm": 1.5891152620315552,
      "learning_rate": 0.0005,
      "loss": 3.6041975021362305,
      "step": 401
    },
    {
      "epoch": 0.06485963213939981,
      "grad_norm": 1.80879807472229,
      "learning_rate": 0.0005,
      "loss": 4.092473030090332,
      "step": 402
    },
    {
      "epoch": 0.06502097450790578,
      "grad_norm": 1.8499172925949097,
      "learning_rate": 0.0005,
      "loss": 3.9844202995300293,
      "step": 403
    },
    {
      "epoch": 0.06518231687641174,
      "grad_norm": 1.9042143821716309,
      "learning_rate": 0.0005,
      "loss": 3.3810033798217773,
      "step": 404
    },
    {
      "epoch": 0.06534365924491771,
      "grad_norm": 1.5597805976867676,
      "learning_rate": 0.0005,
      "loss": 3.8114612102508545,
      "step": 405
    },
    {
      "epoch": 0.06550500161342368,
      "grad_norm": 1.7153222560882568,
      "learning_rate": 0.0005,
      "loss": 3.7065324783325195,
      "step": 406
    },
    {
      "epoch": 0.06566634398192965,
      "grad_norm": 1.7167415618896484,
      "learning_rate": 0.0005,
      "loss": 3.262462615966797,
      "step": 407
    },
    {
      "epoch": 0.06582768635043562,
      "grad_norm": 1.6031310558319092,
      "learning_rate": 0.0005,
      "loss": 3.4330365657806396,
      "step": 408
    },
    {
      "epoch": 0.06598902871894159,
      "grad_norm": 1.643920660018921,
      "learning_rate": 0.0005,
      "loss": 3.6223344802856445,
      "step": 409
    },
    {
      "epoch": 0.06615037108744756,
      "grad_norm": 1.6313714981079102,
      "learning_rate": 0.0005,
      "loss": 3.840764045715332,
      "step": 410
    },
    {
      "epoch": 0.06631171345595353,
      "grad_norm": 1.7402000427246094,
      "learning_rate": 0.0005,
      "loss": 3.8675248622894287,
      "step": 411
    },
    {
      "epoch": 0.0664730558244595,
      "grad_norm": 1.6692306995391846,
      "learning_rate": 0.0005,
      "loss": 3.5809013843536377,
      "step": 412
    },
    {
      "epoch": 0.06663439819296547,
      "grad_norm": 1.7654870748519897,
      "learning_rate": 0.0005,
      "loss": 3.299990177154541,
      "step": 413
    },
    {
      "epoch": 0.06679574056147145,
      "grad_norm": 1.6056486368179321,
      "learning_rate": 0.0005,
      "loss": 3.7178447246551514,
      "step": 414
    },
    {
      "epoch": 0.06695708292997742,
      "grad_norm": 1.4362404346466064,
      "learning_rate": 0.0005,
      "loss": 3.561370372772217,
      "step": 415
    },
    {
      "epoch": 0.06711842529848339,
      "grad_norm": 1.6423568725585938,
      "learning_rate": 0.0005,
      "loss": 3.9479546546936035,
      "step": 416
    },
    {
      "epoch": 0.06727976766698936,
      "grad_norm": 1.343170404434204,
      "learning_rate": 0.0005,
      "loss": 3.609325885772705,
      "step": 417
    },
    {
      "epoch": 0.06744111003549533,
      "grad_norm": 1.4620505571365356,
      "learning_rate": 0.0005,
      "loss": 3.8973264694213867,
      "step": 418
    },
    {
      "epoch": 0.06760245240400128,
      "grad_norm": 1.927358627319336,
      "learning_rate": 0.0005,
      "loss": 3.814401865005493,
      "step": 419
    },
    {
      "epoch": 0.06776379477250725,
      "grad_norm": 2.2731637954711914,
      "learning_rate": 0.0005,
      "loss": 3.701080560684204,
      "step": 420
    },
    {
      "epoch": 0.06792513714101323,
      "grad_norm": 2.6489272117614746,
      "learning_rate": 0.0005,
      "loss": 3.427114963531494,
      "step": 421
    },
    {
      "epoch": 0.0680864795095192,
      "grad_norm": 1.7613461017608643,
      "learning_rate": 0.0005,
      "loss": 3.3893966674804688,
      "step": 422
    },
    {
      "epoch": 0.06824782187802517,
      "grad_norm": 1.6284370422363281,
      "learning_rate": 0.0005,
      "loss": 3.8069896697998047,
      "step": 423
    },
    {
      "epoch": 0.06840916424653114,
      "grad_norm": 1.8111398220062256,
      "learning_rate": 0.0005,
      "loss": 3.880843162536621,
      "step": 424
    },
    {
      "epoch": 0.06857050661503711,
      "grad_norm": 1.5879168510437012,
      "learning_rate": 0.0005,
      "loss": 3.685073137283325,
      "step": 425
    },
    {
      "epoch": 0.06873184898354308,
      "grad_norm": 1.525168776512146,
      "learning_rate": 0.0005,
      "loss": 3.8632473945617676,
      "step": 426
    },
    {
      "epoch": 0.06889319135204905,
      "grad_norm": 1.3050518035888672,
      "learning_rate": 0.0005,
      "loss": 3.2876644134521484,
      "step": 427
    },
    {
      "epoch": 0.06905453372055502,
      "grad_norm": 1.641542911529541,
      "learning_rate": 0.0005,
      "loss": 4.182554244995117,
      "step": 428
    },
    {
      "epoch": 0.06921587608906099,
      "grad_norm": 1.737646460533142,
      "learning_rate": 0.0005,
      "loss": 4.041416168212891,
      "step": 429
    },
    {
      "epoch": 0.06937721845756696,
      "grad_norm": 1.7702863216400146,
      "learning_rate": 0.0005,
      "loss": 3.326080799102783,
      "step": 430
    },
    {
      "epoch": 0.06953856082607293,
      "grad_norm": 1.4167492389678955,
      "learning_rate": 0.0005,
      "loss": 4.011853218078613,
      "step": 431
    },
    {
      "epoch": 0.0696999031945789,
      "grad_norm": 1.7214311361312866,
      "learning_rate": 0.0005,
      "loss": 3.462425470352173,
      "step": 432
    },
    {
      "epoch": 0.06986124556308487,
      "grad_norm": 1.9922640323638916,
      "learning_rate": 0.0005,
      "loss": 4.128365516662598,
      "step": 433
    },
    {
      "epoch": 0.07002258793159083,
      "grad_norm": 1.9983222484588623,
      "learning_rate": 0.0005,
      "loss": 3.960446357727051,
      "step": 434
    },
    {
      "epoch": 0.0701839303000968,
      "grad_norm": 1.5248733758926392,
      "learning_rate": 0.0005,
      "loss": 3.4846301078796387,
      "step": 435
    },
    {
      "epoch": 0.07034527266860277,
      "grad_norm": 1.5860148668289185,
      "learning_rate": 0.0005,
      "loss": 3.9181907176971436,
      "step": 436
    },
    {
      "epoch": 0.07050661503710874,
      "grad_norm": 1.3339401483535767,
      "learning_rate": 0.0005,
      "loss": 3.5158956050872803,
      "step": 437
    },
    {
      "epoch": 0.07066795740561471,
      "grad_norm": 1.6796517372131348,
      "learning_rate": 0.0005,
      "loss": 3.3529751300811768,
      "step": 438
    },
    {
      "epoch": 0.07082929977412068,
      "grad_norm": 1.5366053581237793,
      "learning_rate": 0.0005,
      "loss": 3.3329668045043945,
      "step": 439
    },
    {
      "epoch": 0.07099064214262665,
      "grad_norm": 1.9562941789627075,
      "learning_rate": 0.0005,
      "loss": 3.587702751159668,
      "step": 440
    },
    {
      "epoch": 0.07115198451113262,
      "grad_norm": 1.4829702377319336,
      "learning_rate": 0.0005,
      "loss": 3.654507637023926,
      "step": 441
    },
    {
      "epoch": 0.0713133268796386,
      "grad_norm": 1.8807986974716187,
      "learning_rate": 0.0005,
      "loss": 3.7835147380828857,
      "step": 442
    },
    {
      "epoch": 0.07147466924814457,
      "grad_norm": 2.292381763458252,
      "learning_rate": 0.0005,
      "loss": 3.5230467319488525,
      "step": 443
    },
    {
      "epoch": 0.07163601161665054,
      "grad_norm": 1.9452873468399048,
      "learning_rate": 0.0005,
      "loss": 3.9886367321014404,
      "step": 444
    },
    {
      "epoch": 0.0717973539851565,
      "grad_norm": 1.9832496643066406,
      "learning_rate": 0.0005,
      "loss": 3.455256938934326,
      "step": 445
    },
    {
      "epoch": 0.07195869635366248,
      "grad_norm": 2.074950695037842,
      "learning_rate": 0.0005,
      "loss": 3.3767199516296387,
      "step": 446
    },
    {
      "epoch": 0.07212003872216845,
      "grad_norm": 1.4143823385238647,
      "learning_rate": 0.0005,
      "loss": 3.9101192951202393,
      "step": 447
    },
    {
      "epoch": 0.0722813810906744,
      "grad_norm": 1.2333894968032837,
      "learning_rate": 0.0005,
      "loss": 3.6661131381988525,
      "step": 448
    },
    {
      "epoch": 0.07244272345918037,
      "grad_norm": 1.7344447374343872,
      "learning_rate": 0.0005,
      "loss": 4.163272857666016,
      "step": 449
    },
    {
      "epoch": 0.07260406582768635,
      "grad_norm": 1.5370875597000122,
      "learning_rate": 0.0005,
      "loss": 3.9125852584838867,
      "step": 450
    },
    {
      "epoch": 0.07276540819619232,
      "grad_norm": 1.7388519048690796,
      "learning_rate": 0.0005,
      "loss": 3.639862537384033,
      "step": 451
    },
    {
      "epoch": 0.07292675056469829,
      "grad_norm": 1.6159418821334839,
      "learning_rate": 0.0005,
      "loss": 3.5164523124694824,
      "step": 452
    },
    {
      "epoch": 0.07308809293320426,
      "grad_norm": 2.323798179626465,
      "learning_rate": 0.0005,
      "loss": 4.009775161743164,
      "step": 453
    },
    {
      "epoch": 0.07324943530171023,
      "grad_norm": 2.2956383228302,
      "learning_rate": 0.0005,
      "loss": 4.029460906982422,
      "step": 454
    },
    {
      "epoch": 0.0734107776702162,
      "grad_norm": 2.253005027770996,
      "learning_rate": 0.0005,
      "loss": 3.98555850982666,
      "step": 455
    },
    {
      "epoch": 0.07357212003872217,
      "grad_norm": 2.3645951747894287,
      "learning_rate": 0.0005,
      "loss": 3.4272727966308594,
      "step": 456
    },
    {
      "epoch": 0.07373346240722814,
      "grad_norm": 1.9060914516448975,
      "learning_rate": 0.0005,
      "loss": 4.272453308105469,
      "step": 457
    },
    {
      "epoch": 0.07389480477573411,
      "grad_norm": 1.3985402584075928,
      "learning_rate": 0.0005,
      "loss": 3.573452949523926,
      "step": 458
    },
    {
      "epoch": 0.07405614714424008,
      "grad_norm": 1.4165980815887451,
      "learning_rate": 0.0005,
      "loss": 3.938532829284668,
      "step": 459
    },
    {
      "epoch": 0.07421748951274605,
      "grad_norm": 1.573847770690918,
      "learning_rate": 0.0005,
      "loss": 3.963306427001953,
      "step": 460
    },
    {
      "epoch": 0.07437883188125202,
      "grad_norm": 1.4623618125915527,
      "learning_rate": 0.0005,
      "loss": 3.7023329734802246,
      "step": 461
    },
    {
      "epoch": 0.07454017424975799,
      "grad_norm": 1.4510688781738281,
      "learning_rate": 0.0005,
      "loss": 3.8594532012939453,
      "step": 462
    },
    {
      "epoch": 0.07470151661826395,
      "grad_norm": 1.5591747760772705,
      "learning_rate": 0.0005,
      "loss": 4.2009687423706055,
      "step": 463
    },
    {
      "epoch": 0.07486285898676992,
      "grad_norm": 1.215246319770813,
      "learning_rate": 0.0005,
      "loss": 3.6864099502563477,
      "step": 464
    },
    {
      "epoch": 0.07502420135527589,
      "grad_norm": 1.9833911657333374,
      "learning_rate": 0.0005,
      "loss": 3.9039382934570312,
      "step": 465
    },
    {
      "epoch": 0.07518554372378186,
      "grad_norm": 1.3234028816223145,
      "learning_rate": 0.0005,
      "loss": 3.448118209838867,
      "step": 466
    },
    {
      "epoch": 0.07534688609228783,
      "grad_norm": 1.7335271835327148,
      "learning_rate": 0.0005,
      "loss": 3.491738796234131,
      "step": 467
    },
    {
      "epoch": 0.0755082284607938,
      "grad_norm": 1.5532065629959106,
      "learning_rate": 0.0005,
      "loss": 3.7802910804748535,
      "step": 468
    },
    {
      "epoch": 0.07566957082929977,
      "grad_norm": 1.561737060546875,
      "learning_rate": 0.0005,
      "loss": 4.234887599945068,
      "step": 469
    },
    {
      "epoch": 0.07583091319780574,
      "grad_norm": 1.3442671298980713,
      "learning_rate": 0.0005,
      "loss": 3.303595781326294,
      "step": 470
    },
    {
      "epoch": 0.07599225556631171,
      "grad_norm": 1.6980648040771484,
      "learning_rate": 0.0005,
      "loss": 4.033544540405273,
      "step": 471
    },
    {
      "epoch": 0.07615359793481769,
      "grad_norm": 1.7123768329620361,
      "learning_rate": 0.0005,
      "loss": 4.041604995727539,
      "step": 472
    },
    {
      "epoch": 0.07631494030332366,
      "grad_norm": 2.141578197479248,
      "learning_rate": 0.0005,
      "loss": 3.429955005645752,
      "step": 473
    },
    {
      "epoch": 0.07647628267182963,
      "grad_norm": 1.5381149053573608,
      "learning_rate": 0.0005,
      "loss": 4.193968772888184,
      "step": 474
    },
    {
      "epoch": 0.0766376250403356,
      "grad_norm": 1.9282156229019165,
      "learning_rate": 0.0005,
      "loss": 3.775520086288452,
      "step": 475
    },
    {
      "epoch": 0.07679896740884157,
      "grad_norm": 1.6447601318359375,
      "learning_rate": 0.0005,
      "loss": 3.8075015544891357,
      "step": 476
    },
    {
      "epoch": 0.07696030977734754,
      "grad_norm": 1.9508285522460938,
      "learning_rate": 0.0005,
      "loss": 4.0658674240112305,
      "step": 477
    },
    {
      "epoch": 0.0771216521458535,
      "grad_norm": 1.6421656608581543,
      "learning_rate": 0.0005,
      "loss": 4.16646671295166,
      "step": 478
    },
    {
      "epoch": 0.07728299451435947,
      "grad_norm": 1.8462263345718384,
      "learning_rate": 0.0005,
      "loss": 3.953028440475464,
      "step": 479
    },
    {
      "epoch": 0.07744433688286544,
      "grad_norm": 1.3713942766189575,
      "learning_rate": 0.0005,
      "loss": 3.31265926361084,
      "step": 480
    },
    {
      "epoch": 0.0776056792513714,
      "grad_norm": 2.821371555328369,
      "learning_rate": 0.0005,
      "loss": 3.887441635131836,
      "step": 481
    },
    {
      "epoch": 0.07776702161987738,
      "grad_norm": 1.6638892889022827,
      "learning_rate": 0.0005,
      "loss": 3.507939338684082,
      "step": 482
    },
    {
      "epoch": 0.07792836398838335,
      "grad_norm": 3.728039026260376,
      "learning_rate": 0.0005,
      "loss": 3.15208101272583,
      "step": 483
    },
    {
      "epoch": 0.07808970635688932,
      "grad_norm": 1.8635610342025757,
      "learning_rate": 0.0005,
      "loss": 3.4561972618103027,
      "step": 484
    },
    {
      "epoch": 0.07825104872539529,
      "grad_norm": 1.7522943019866943,
      "learning_rate": 0.0005,
      "loss": 3.5972180366516113,
      "step": 485
    },
    {
      "epoch": 0.07841239109390126,
      "grad_norm": 1.7135623693466187,
      "learning_rate": 0.0005,
      "loss": 3.1714975833892822,
      "step": 486
    },
    {
      "epoch": 0.07857373346240723,
      "grad_norm": 2.491623640060425,
      "learning_rate": 0.0005,
      "loss": 3.6568450927734375,
      "step": 487
    },
    {
      "epoch": 0.0787350758309132,
      "grad_norm": 1.6507158279418945,
      "learning_rate": 0.0005,
      "loss": 3.589154005050659,
      "step": 488
    },
    {
      "epoch": 0.07889641819941917,
      "grad_norm": 1.5871272087097168,
      "learning_rate": 0.0005,
      "loss": 3.2973015308380127,
      "step": 489
    },
    {
      "epoch": 0.07905776056792514,
      "grad_norm": 1.6571515798568726,
      "learning_rate": 0.0005,
      "loss": 3.7241873741149902,
      "step": 490
    },
    {
      "epoch": 0.07921910293643111,
      "grad_norm": 1.8087058067321777,
      "learning_rate": 0.0005,
      "loss": 3.868893623352051,
      "step": 491
    },
    {
      "epoch": 0.07938044530493708,
      "grad_norm": 1.6994775533676147,
      "learning_rate": 0.0005,
      "loss": 3.474741220474243,
      "step": 492
    },
    {
      "epoch": 0.07954178767344304,
      "grad_norm": 1.3102315664291382,
      "learning_rate": 0.0005,
      "loss": 3.7500274181365967,
      "step": 493
    },
    {
      "epoch": 0.07970313004194901,
      "grad_norm": 1.2714859247207642,
      "learning_rate": 0.0005,
      "loss": 4.013525009155273,
      "step": 494
    },
    {
      "epoch": 0.07986447241045498,
      "grad_norm": 1.3922029733657837,
      "learning_rate": 0.0005,
      "loss": 3.98238468170166,
      "step": 495
    },
    {
      "epoch": 0.08002581477896095,
      "grad_norm": 1.9914151430130005,
      "learning_rate": 0.0005,
      "loss": 3.7003583908081055,
      "step": 496
    },
    {
      "epoch": 0.08018715714746692,
      "grad_norm": 1.6969112157821655,
      "learning_rate": 0.0005,
      "loss": 3.4773001670837402,
      "step": 497
    },
    {
      "epoch": 0.0803484995159729,
      "grad_norm": 1.702730417251587,
      "learning_rate": 0.0005,
      "loss": 3.758166551589966,
      "step": 498
    },
    {
      "epoch": 0.08050984188447886,
      "grad_norm": 1.4129596948623657,
      "learning_rate": 0.0005,
      "loss": 3.649190902709961,
      "step": 499
    },
    {
      "epoch": 0.08067118425298483,
      "grad_norm": 1.622870683670044,
      "learning_rate": 0.0005,
      "loss": 3.8380136489868164,
      "step": 500
    },
    {
      "epoch": 0.0808325266214908,
      "grad_norm": 1.818995475769043,
      "learning_rate": 0.0005,
      "loss": 3.533362865447998,
      "step": 501
    },
    {
      "epoch": 0.08099386898999678,
      "grad_norm": 1.8077824115753174,
      "learning_rate": 0.0005,
      "loss": 3.9180667400360107,
      "step": 502
    },
    {
      "epoch": 0.08115521135850275,
      "grad_norm": 1.708306074142456,
      "learning_rate": 0.0005,
      "loss": 3.5227956771850586,
      "step": 503
    },
    {
      "epoch": 0.08131655372700872,
      "grad_norm": 1.5226664543151855,
      "learning_rate": 0.0005,
      "loss": 3.8080501556396484,
      "step": 504
    },
    {
      "epoch": 0.08147789609551469,
      "grad_norm": 1.4165152311325073,
      "learning_rate": 0.0005,
      "loss": 4.090529918670654,
      "step": 505
    },
    {
      "epoch": 0.08163923846402066,
      "grad_norm": 1.5321029424667358,
      "learning_rate": 0.0005,
      "loss": 3.645972728729248,
      "step": 506
    },
    {
      "epoch": 0.08180058083252661,
      "grad_norm": 1.8012831211090088,
      "learning_rate": 0.0005,
      "loss": 3.377457857131958,
      "step": 507
    },
    {
      "epoch": 0.08196192320103259,
      "grad_norm": 1.5411020517349243,
      "learning_rate": 0.0005,
      "loss": 3.68916392326355,
      "step": 508
    },
    {
      "epoch": 0.08212326556953856,
      "grad_norm": 2.806722402572632,
      "learning_rate": 0.0005,
      "loss": 4.191666603088379,
      "step": 509
    },
    {
      "epoch": 0.08228460793804453,
      "grad_norm": 2.4544520378112793,
      "learning_rate": 0.0005,
      "loss": 3.407309055328369,
      "step": 510
    },
    {
      "epoch": 0.0824459503065505,
      "grad_norm": 1.5842195749282837,
      "learning_rate": 0.0005,
      "loss": 4.232051372528076,
      "step": 511
    },
    {
      "epoch": 0.08260729267505647,
      "grad_norm": 1.9904907941818237,
      "learning_rate": 0.0005,
      "loss": 3.7082149982452393,
      "step": 512
    },
    {
      "epoch": 0.08276863504356244,
      "grad_norm": 1.5886383056640625,
      "learning_rate": 0.0005,
      "loss": 3.126329183578491,
      "step": 513
    },
    {
      "epoch": 0.08292997741206841,
      "grad_norm": 1.6192491054534912,
      "learning_rate": 0.0005,
      "loss": 4.032886028289795,
      "step": 514
    },
    {
      "epoch": 0.08309131978057438,
      "grad_norm": 1.9397634267807007,
      "learning_rate": 0.0005,
      "loss": 3.848464250564575,
      "step": 515
    },
    {
      "epoch": 0.08325266214908035,
      "grad_norm": 1.6072195768356323,
      "learning_rate": 0.0005,
      "loss": 3.647702932357788,
      "step": 516
    },
    {
      "epoch": 0.08341400451758632,
      "grad_norm": 1.7285356521606445,
      "learning_rate": 0.0005,
      "loss": 3.2233285903930664,
      "step": 517
    },
    {
      "epoch": 0.08357534688609229,
      "grad_norm": 2.0590176582336426,
      "learning_rate": 0.0005,
      "loss": 3.710510015487671,
      "step": 518
    },
    {
      "epoch": 0.08373668925459826,
      "grad_norm": 2.0119597911834717,
      "learning_rate": 0.0005,
      "loss": 3.8963820934295654,
      "step": 519
    },
    {
      "epoch": 0.08389803162310423,
      "grad_norm": 1.7133657932281494,
      "learning_rate": 0.0005,
      "loss": 3.3131377696990967,
      "step": 520
    },
    {
      "epoch": 0.0840593739916102,
      "grad_norm": 1.5602338314056396,
      "learning_rate": 0.0005,
      "loss": 3.889514923095703,
      "step": 521
    },
    {
      "epoch": 0.08422071636011616,
      "grad_norm": 1.536402702331543,
      "learning_rate": 0.0005,
      "loss": 3.502277374267578,
      "step": 522
    },
    {
      "epoch": 0.08438205872862213,
      "grad_norm": 1.5608576536178589,
      "learning_rate": 0.0005,
      "loss": 2.9576869010925293,
      "step": 523
    },
    {
      "epoch": 0.0845434010971281,
      "grad_norm": 2.2655880451202393,
      "learning_rate": 0.0005,
      "loss": 3.8602592945098877,
      "step": 524
    },
    {
      "epoch": 0.08470474346563407,
      "grad_norm": 1.8300869464874268,
      "learning_rate": 0.0005,
      "loss": 3.5671944618225098,
      "step": 525
    },
    {
      "epoch": 0.08486608583414004,
      "grad_norm": 2.3692193031311035,
      "learning_rate": 0.0005,
      "loss": 3.6121835708618164,
      "step": 526
    },
    {
      "epoch": 0.08502742820264601,
      "grad_norm": 1.4628742933273315,
      "learning_rate": 0.0005,
      "loss": 3.4507997035980225,
      "step": 527
    },
    {
      "epoch": 0.08518877057115198,
      "grad_norm": 1.397195816040039,
      "learning_rate": 0.0005,
      "loss": 3.806619167327881,
      "step": 528
    },
    {
      "epoch": 0.08535011293965795,
      "grad_norm": 1.7160848379135132,
      "learning_rate": 0.0005,
      "loss": 4.109884262084961,
      "step": 529
    },
    {
      "epoch": 0.08551145530816393,
      "grad_norm": 1.7899456024169922,
      "learning_rate": 0.0005,
      "loss": 3.5277113914489746,
      "step": 530
    },
    {
      "epoch": 0.0856727976766699,
      "grad_norm": 1.3459352254867554,
      "learning_rate": 0.0005,
      "loss": 3.0795559883117676,
      "step": 531
    },
    {
      "epoch": 0.08583414004517587,
      "grad_norm": 1.8092540502548218,
      "learning_rate": 0.0005,
      "loss": 4.35184907913208,
      "step": 532
    },
    {
      "epoch": 0.08599548241368184,
      "grad_norm": 1.8475300073623657,
      "learning_rate": 0.0005,
      "loss": 3.208369731903076,
      "step": 533
    },
    {
      "epoch": 0.08615682478218781,
      "grad_norm": 1.561145544052124,
      "learning_rate": 0.0005,
      "loss": 3.3895363807678223,
      "step": 534
    },
    {
      "epoch": 0.08631816715069378,
      "grad_norm": 1.574949026107788,
      "learning_rate": 0.0005,
      "loss": 3.5876412391662598,
      "step": 535
    },
    {
      "epoch": 0.08647950951919975,
      "grad_norm": 3.8335609436035156,
      "learning_rate": 0.0005,
      "loss": 3.597895860671997,
      "step": 536
    },
    {
      "epoch": 0.0866408518877057,
      "grad_norm": 1.5300650596618652,
      "learning_rate": 0.0005,
      "loss": 3.6732568740844727,
      "step": 537
    },
    {
      "epoch": 0.08680219425621168,
      "grad_norm": 1.543089509010315,
      "learning_rate": 0.0005,
      "loss": 3.8558568954467773,
      "step": 538
    },
    {
      "epoch": 0.08696353662471765,
      "grad_norm": 1.6991724967956543,
      "learning_rate": 0.0005,
      "loss": 3.9507617950439453,
      "step": 539
    },
    {
      "epoch": 0.08712487899322362,
      "grad_norm": 1.910149335861206,
      "learning_rate": 0.0005,
      "loss": 3.4920899868011475,
      "step": 540
    },
    {
      "epoch": 0.08728622136172959,
      "grad_norm": 1.9125733375549316,
      "learning_rate": 0.0005,
      "loss": 3.8009605407714844,
      "step": 541
    },
    {
      "epoch": 0.08744756373023556,
      "grad_norm": 1.670627474784851,
      "learning_rate": 0.0005,
      "loss": 3.5843443870544434,
      "step": 542
    },
    {
      "epoch": 0.08760890609874153,
      "grad_norm": 1.8695791959762573,
      "learning_rate": 0.0005,
      "loss": 3.794346809387207,
      "step": 543
    },
    {
      "epoch": 0.0877702484672475,
      "grad_norm": 1.6539111137390137,
      "learning_rate": 0.0005,
      "loss": 4.099869728088379,
      "step": 544
    },
    {
      "epoch": 0.08793159083575347,
      "grad_norm": 1.747591495513916,
      "learning_rate": 0.0005,
      "loss": 3.546466827392578,
      "step": 545
    },
    {
      "epoch": 0.08809293320425944,
      "grad_norm": 1.5926953554153442,
      "learning_rate": 0.0005,
      "loss": 3.1128993034362793,
      "step": 546
    },
    {
      "epoch": 0.08825427557276541,
      "grad_norm": 2.289315700531006,
      "learning_rate": 0.0005,
      "loss": 3.8800735473632812,
      "step": 547
    },
    {
      "epoch": 0.08841561794127138,
      "grad_norm": 2.396239995956421,
      "learning_rate": 0.0005,
      "loss": 3.365987777709961,
      "step": 548
    },
    {
      "epoch": 0.08857696030977735,
      "grad_norm": 1.4608325958251953,
      "learning_rate": 0.0005,
      "loss": 3.5027072429656982,
      "step": 549
    },
    {
      "epoch": 0.08873830267828332,
      "grad_norm": 1.5563955307006836,
      "learning_rate": 0.0005,
      "loss": 3.3488645553588867,
      "step": 550
    },
    {
      "epoch": 0.08889964504678928,
      "grad_norm": 1.5003201961517334,
      "learning_rate": 0.0005,
      "loss": 3.9457390308380127,
      "step": 551
    },
    {
      "epoch": 0.08906098741529525,
      "grad_norm": 3.5093624591827393,
      "learning_rate": 0.0005,
      "loss": 3.672919750213623,
      "step": 552
    },
    {
      "epoch": 0.08922232978380122,
      "grad_norm": 1.6257988214492798,
      "learning_rate": 0.0005,
      "loss": 3.413639545440674,
      "step": 553
    },
    {
      "epoch": 0.08938367215230719,
      "grad_norm": 1.761642575263977,
      "learning_rate": 0.0005,
      "loss": 3.5464982986450195,
      "step": 554
    },
    {
      "epoch": 0.08954501452081316,
      "grad_norm": 1.3519079685211182,
      "learning_rate": 0.0005,
      "loss": 3.510676622390747,
      "step": 555
    },
    {
      "epoch": 0.08970635688931913,
      "grad_norm": 1.3543999195098877,
      "learning_rate": 0.0005,
      "loss": 3.9597747325897217,
      "step": 556
    },
    {
      "epoch": 0.0898676992578251,
      "grad_norm": 1.2983002662658691,
      "learning_rate": 0.0005,
      "loss": 3.9326512813568115,
      "step": 557
    },
    {
      "epoch": 0.09002904162633107,
      "grad_norm": 1.7495406866073608,
      "learning_rate": 0.0005,
      "loss": 3.46524715423584,
      "step": 558
    },
    {
      "epoch": 0.09019038399483705,
      "grad_norm": 4.249914169311523,
      "learning_rate": 0.0005,
      "loss": 3.6437931060791016,
      "step": 559
    },
    {
      "epoch": 0.09035172636334302,
      "grad_norm": 1.6290159225463867,
      "learning_rate": 0.0005,
      "loss": 4.04512882232666,
      "step": 560
    },
    {
      "epoch": 0.09051306873184899,
      "grad_norm": 1.7505898475646973,
      "learning_rate": 0.0005,
      "loss": 3.7831459045410156,
      "step": 561
    },
    {
      "epoch": 0.09067441110035496,
      "grad_norm": 2.0622313022613525,
      "learning_rate": 0.0005,
      "loss": 3.597712278366089,
      "step": 562
    },
    {
      "epoch": 0.09083575346886093,
      "grad_norm": 1.9640132188796997,
      "learning_rate": 0.0005,
      "loss": 3.5922884941101074,
      "step": 563
    },
    {
      "epoch": 0.0909970958373669,
      "grad_norm": 1.6426328420639038,
      "learning_rate": 0.0005,
      "loss": 3.432982921600342,
      "step": 564
    },
    {
      "epoch": 0.09115843820587287,
      "grad_norm": 2.2980995178222656,
      "learning_rate": 0.0005,
      "loss": 3.265230655670166,
      "step": 565
    },
    {
      "epoch": 0.09131978057437883,
      "grad_norm": 1.584352970123291,
      "learning_rate": 0.0005,
      "loss": 4.020280838012695,
      "step": 566
    },
    {
      "epoch": 0.0914811229428848,
      "grad_norm": 2.2856390476226807,
      "learning_rate": 0.0005,
      "loss": 3.577633857727051,
      "step": 567
    },
    {
      "epoch": 0.09164246531139077,
      "grad_norm": 1.8659476041793823,
      "learning_rate": 0.0005,
      "loss": 3.366760730743408,
      "step": 568
    },
    {
      "epoch": 0.09180380767989674,
      "grad_norm": 1.4736436605453491,
      "learning_rate": 0.0005,
      "loss": 3.6580967903137207,
      "step": 569
    },
    {
      "epoch": 0.09196515004840271,
      "grad_norm": 2.115217447280884,
      "learning_rate": 0.0005,
      "loss": 4.321970462799072,
      "step": 570
    },
    {
      "epoch": 0.09212649241690868,
      "grad_norm": 1.484847903251648,
      "learning_rate": 0.0005,
      "loss": 3.405728816986084,
      "step": 571
    },
    {
      "epoch": 0.09228783478541465,
      "grad_norm": 2.118342161178589,
      "learning_rate": 0.0005,
      "loss": 3.5109620094299316,
      "step": 572
    },
    {
      "epoch": 0.09244917715392062,
      "grad_norm": 2.116482973098755,
      "learning_rate": 0.0005,
      "loss": 4.027048110961914,
      "step": 573
    },
    {
      "epoch": 0.09261051952242659,
      "grad_norm": 1.3959823846817017,
      "learning_rate": 0.0005,
      "loss": 3.7770445346832275,
      "step": 574
    },
    {
      "epoch": 0.09277186189093256,
      "grad_norm": 1.949000358581543,
      "learning_rate": 0.0005,
      "loss": 3.7401294708251953,
      "step": 575
    },
    {
      "epoch": 0.09293320425943853,
      "grad_norm": 1.5143311023712158,
      "learning_rate": 0.0005,
      "loss": 3.6580452919006348,
      "step": 576
    },
    {
      "epoch": 0.0930945466279445,
      "grad_norm": 2.070950984954834,
      "learning_rate": 0.0005,
      "loss": 3.465782642364502,
      "step": 577
    },
    {
      "epoch": 0.09325588899645047,
      "grad_norm": 1.494865894317627,
      "learning_rate": 0.0005,
      "loss": 3.5890092849731445,
      "step": 578
    },
    {
      "epoch": 0.09341723136495644,
      "grad_norm": 1.5756101608276367,
      "learning_rate": 0.0005,
      "loss": 3.732357978820801,
      "step": 579
    },
    {
      "epoch": 0.09357857373346241,
      "grad_norm": 1.3320037126541138,
      "learning_rate": 0.0005,
      "loss": 3.3831682205200195,
      "step": 580
    },
    {
      "epoch": 0.09373991610196837,
      "grad_norm": 1.9279625415802002,
      "learning_rate": 0.0005,
      "loss": 3.6700408458709717,
      "step": 581
    },
    {
      "epoch": 0.09390125847047434,
      "grad_norm": 1.426285743713379,
      "learning_rate": 0.0005,
      "loss": 3.682271957397461,
      "step": 582
    },
    {
      "epoch": 0.09406260083898031,
      "grad_norm": 1.2597018480300903,
      "learning_rate": 0.0005,
      "loss": 3.5298960208892822,
      "step": 583
    },
    {
      "epoch": 0.09422394320748628,
      "grad_norm": 1.7935094833374023,
      "learning_rate": 0.0005,
      "loss": 3.7588934898376465,
      "step": 584
    },
    {
      "epoch": 0.09438528557599225,
      "grad_norm": 1.2731043100357056,
      "learning_rate": 0.0005,
      "loss": 3.790924549102783,
      "step": 585
    },
    {
      "epoch": 0.09454662794449822,
      "grad_norm": 1.7626862525939941,
      "learning_rate": 0.0005,
      "loss": 3.6114554405212402,
      "step": 586
    },
    {
      "epoch": 0.0947079703130042,
      "grad_norm": 1.7737748622894287,
      "learning_rate": 0.0005,
      "loss": 3.5561892986297607,
      "step": 587
    },
    {
      "epoch": 0.09486931268151017,
      "grad_norm": 1.5825459957122803,
      "learning_rate": 0.0005,
      "loss": 4.003012657165527,
      "step": 588
    },
    {
      "epoch": 0.09503065505001614,
      "grad_norm": 1.7449301481246948,
      "learning_rate": 0.0005,
      "loss": 3.5702319145202637,
      "step": 589
    },
    {
      "epoch": 0.0951919974185221,
      "grad_norm": 2.1531572341918945,
      "learning_rate": 0.0005,
      "loss": 3.6021997928619385,
      "step": 590
    },
    {
      "epoch": 0.09535333978702808,
      "grad_norm": 1.2850607633590698,
      "learning_rate": 0.0005,
      "loss": 3.4416470527648926,
      "step": 591
    },
    {
      "epoch": 0.09551468215553405,
      "grad_norm": 1.8253064155578613,
      "learning_rate": 0.0005,
      "loss": 3.291110038757324,
      "step": 592
    },
    {
      "epoch": 0.09567602452404002,
      "grad_norm": 1.3507641553878784,
      "learning_rate": 0.0005,
      "loss": 3.332559585571289,
      "step": 593
    },
    {
      "epoch": 0.09583736689254599,
      "grad_norm": 1.8445383310317993,
      "learning_rate": 0.0005,
      "loss": 3.5949835777282715,
      "step": 594
    },
    {
      "epoch": 0.09599870926105195,
      "grad_norm": 2.0136196613311768,
      "learning_rate": 0.0005,
      "loss": 3.8746941089630127,
      "step": 595
    },
    {
      "epoch": 0.09616005162955792,
      "grad_norm": 1.809884786605835,
      "learning_rate": 0.0005,
      "loss": 3.4602270126342773,
      "step": 596
    },
    {
      "epoch": 0.09632139399806389,
      "grad_norm": 2.074324369430542,
      "learning_rate": 0.0005,
      "loss": 4.050209999084473,
      "step": 597
    },
    {
      "epoch": 0.09648273636656986,
      "grad_norm": 2.997457504272461,
      "learning_rate": 0.0005,
      "loss": 3.239490032196045,
      "step": 598
    },
    {
      "epoch": 0.09664407873507583,
      "grad_norm": 1.4116603136062622,
      "learning_rate": 0.0005,
      "loss": 4.008451461791992,
      "step": 599
    },
    {
      "epoch": 0.0968054211035818,
      "grad_norm": 2.3232340812683105,
      "learning_rate": 0.0005,
      "loss": 4.284806728363037,
      "step": 600
    },
    {
      "epoch": 0.09696676347208777,
      "grad_norm": 2.6560702323913574,
      "learning_rate": 0.0005,
      "loss": 3.8839409351348877,
      "step": 601
    },
    {
      "epoch": 0.09712810584059374,
      "grad_norm": 1.6913273334503174,
      "learning_rate": 0.0005,
      "loss": 3.6770052909851074,
      "step": 602
    },
    {
      "epoch": 0.09728944820909971,
      "grad_norm": 1.2328126430511475,
      "learning_rate": 0.0005,
      "loss": 3.938739538192749,
      "step": 603
    },
    {
      "epoch": 0.09745079057760568,
      "grad_norm": 1.5187004804611206,
      "learning_rate": 0.0005,
      "loss": 3.5243563652038574,
      "step": 604
    },
    {
      "epoch": 0.09761213294611165,
      "grad_norm": 2.2772557735443115,
      "learning_rate": 0.0005,
      "loss": 3.7534964084625244,
      "step": 605
    },
    {
      "epoch": 0.09777347531461762,
      "grad_norm": 1.5607309341430664,
      "learning_rate": 0.0005,
      "loss": 3.4031076431274414,
      "step": 606
    },
    {
      "epoch": 0.09793481768312359,
      "grad_norm": 1.4929018020629883,
      "learning_rate": 0.0005,
      "loss": 3.6803174018859863,
      "step": 607
    },
    {
      "epoch": 0.09809616005162956,
      "grad_norm": 1.3451220989227295,
      "learning_rate": 0.0005,
      "loss": 3.730201005935669,
      "step": 608
    },
    {
      "epoch": 0.09825750242013553,
      "grad_norm": 1.393361210823059,
      "learning_rate": 0.0005,
      "loss": 3.809596300125122,
      "step": 609
    },
    {
      "epoch": 0.09841884478864149,
      "grad_norm": 1.4220882654190063,
      "learning_rate": 0.0005,
      "loss": 3.704432249069214,
      "step": 610
    },
    {
      "epoch": 0.09858018715714746,
      "grad_norm": 1.4905002117156982,
      "learning_rate": 0.0005,
      "loss": 3.38814640045166,
      "step": 611
    },
    {
      "epoch": 0.09874152952565343,
      "grad_norm": 1.3277087211608887,
      "learning_rate": 0.0005,
      "loss": 3.1411454677581787,
      "step": 612
    },
    {
      "epoch": 0.0989028718941594,
      "grad_norm": 1.564682960510254,
      "learning_rate": 0.0005,
      "loss": 3.838010787963867,
      "step": 613
    },
    {
      "epoch": 0.09906421426266537,
      "grad_norm": 1.2384752035140991,
      "learning_rate": 0.0005,
      "loss": 3.8980443477630615,
      "step": 614
    },
    {
      "epoch": 0.09922555663117134,
      "grad_norm": 1.476803183555603,
      "learning_rate": 0.0005,
      "loss": 3.2764856815338135,
      "step": 615
    },
    {
      "epoch": 0.09938689899967731,
      "grad_norm": 1.6374719142913818,
      "learning_rate": 0.0005,
      "loss": 3.2748351097106934,
      "step": 616
    },
    {
      "epoch": 0.09954824136818329,
      "grad_norm": 1.52738618850708,
      "learning_rate": 0.0005,
      "loss": 3.619485378265381,
      "step": 617
    },
    {
      "epoch": 0.09970958373668926,
      "grad_norm": 1.512250304222107,
      "learning_rate": 0.0005,
      "loss": 4.199598789215088,
      "step": 618
    },
    {
      "epoch": 0.09987092610519523,
      "grad_norm": 1.4150855541229248,
      "learning_rate": 0.0005,
      "loss": 3.4897494316101074,
      "step": 619
    },
    {
      "epoch": 0.1000322684737012,
      "grad_norm": 2.001840829849243,
      "learning_rate": 0.0005,
      "loss": 4.036662578582764,
      "step": 620
    },
    {
      "epoch": 0.10019361084220717,
      "grad_norm": 1.6144965887069702,
      "learning_rate": 0.0005,
      "loss": 3.8471083641052246,
      "step": 621
    },
    {
      "epoch": 0.10035495321071314,
      "grad_norm": 1.8535898923873901,
      "learning_rate": 0.0005,
      "loss": 3.713714361190796,
      "step": 622
    },
    {
      "epoch": 0.10051629557921911,
      "grad_norm": 1.3852776288986206,
      "learning_rate": 0.0005,
      "loss": 3.843733549118042,
      "step": 623
    },
    {
      "epoch": 0.10067763794772508,
      "grad_norm": 1.989257574081421,
      "learning_rate": 0.0005,
      "loss": 3.904895305633545,
      "step": 624
    },
    {
      "epoch": 0.10083898031623104,
      "grad_norm": 1.6390717029571533,
      "learning_rate": 0.0005,
      "loss": 3.4966959953308105,
      "step": 625
    },
    {
      "epoch": 0.101000322684737,
      "grad_norm": 1.518842101097107,
      "learning_rate": 0.0005,
      "loss": 3.3194804191589355,
      "step": 626
    },
    {
      "epoch": 0.10116166505324298,
      "grad_norm": 1.3582837581634521,
      "learning_rate": 0.0005,
      "loss": 3.989997148513794,
      "step": 627
    },
    {
      "epoch": 0.10132300742174895,
      "grad_norm": 2.1163594722747803,
      "learning_rate": 0.0005,
      "loss": 3.5009653568267822,
      "step": 628
    },
    {
      "epoch": 0.10148434979025492,
      "grad_norm": 2.039806604385376,
      "learning_rate": 0.0005,
      "loss": 3.673875331878662,
      "step": 629
    },
    {
      "epoch": 0.10164569215876089,
      "grad_norm": 1.762290358543396,
      "learning_rate": 0.0005,
      "loss": 3.9078807830810547,
      "step": 630
    },
    {
      "epoch": 0.10180703452726686,
      "grad_norm": 1.6609859466552734,
      "learning_rate": 0.0005,
      "loss": 3.6829004287719727,
      "step": 631
    },
    {
      "epoch": 0.10196837689577283,
      "grad_norm": 2.0302023887634277,
      "learning_rate": 0.0005,
      "loss": 3.7699389457702637,
      "step": 632
    },
    {
      "epoch": 0.1021297192642788,
      "grad_norm": 1.7019327878952026,
      "learning_rate": 0.0005,
      "loss": 3.395679473876953,
      "step": 633
    },
    {
      "epoch": 0.10229106163278477,
      "grad_norm": 1.6133455038070679,
      "learning_rate": 0.0005,
      "loss": 3.2273871898651123,
      "step": 634
    },
    {
      "epoch": 0.10245240400129074,
      "grad_norm": 1.9630486965179443,
      "learning_rate": 0.0005,
      "loss": 3.273805618286133,
      "step": 635
    },
    {
      "epoch": 0.10261374636979671,
      "grad_norm": 1.575228214263916,
      "learning_rate": 0.0005,
      "loss": 3.6085245609283447,
      "step": 636
    },
    {
      "epoch": 0.10277508873830268,
      "grad_norm": 1.8548073768615723,
      "learning_rate": 0.0005,
      "loss": 3.703619956970215,
      "step": 637
    },
    {
      "epoch": 0.10293643110680865,
      "grad_norm": 1.6215606927871704,
      "learning_rate": 0.0005,
      "loss": 3.3200643062591553,
      "step": 638
    },
    {
      "epoch": 0.10309777347531461,
      "grad_norm": 1.6663475036621094,
      "learning_rate": 0.0005,
      "loss": 3.743480682373047,
      "step": 639
    },
    {
      "epoch": 0.10325911584382058,
      "grad_norm": 1.499160647392273,
      "learning_rate": 0.0005,
      "loss": 3.4871249198913574,
      "step": 640
    },
    {
      "epoch": 0.10342045821232655,
      "grad_norm": 1.7906975746154785,
      "learning_rate": 0.0005,
      "loss": 3.396268367767334,
      "step": 641
    },
    {
      "epoch": 0.10358180058083252,
      "grad_norm": 2.147569417953491,
      "learning_rate": 0.0005,
      "loss": 3.356290102005005,
      "step": 642
    },
    {
      "epoch": 0.1037431429493385,
      "grad_norm": 1.5272557735443115,
      "learning_rate": 0.0005,
      "loss": 3.4870572090148926,
      "step": 643
    },
    {
      "epoch": 0.10390448531784446,
      "grad_norm": 1.543114185333252,
      "learning_rate": 0.0005,
      "loss": 3.4929656982421875,
      "step": 644
    },
    {
      "epoch": 0.10406582768635043,
      "grad_norm": 1.3994650840759277,
      "learning_rate": 0.0005,
      "loss": 3.7221341133117676,
      "step": 645
    },
    {
      "epoch": 0.1042271700548564,
      "grad_norm": 1.6990724802017212,
      "learning_rate": 0.0005,
      "loss": 3.4772257804870605,
      "step": 646
    },
    {
      "epoch": 0.10438851242336238,
      "grad_norm": 1.574123501777649,
      "learning_rate": 0.0005,
      "loss": 3.2953715324401855,
      "step": 647
    },
    {
      "epoch": 0.10454985479186835,
      "grad_norm": 1.361518383026123,
      "learning_rate": 0.0005,
      "loss": 3.6101982593536377,
      "step": 648
    },
    {
      "epoch": 0.10471119716037432,
      "grad_norm": 1.57150399684906,
      "learning_rate": 0.0005,
      "loss": 3.2084269523620605,
      "step": 649
    },
    {
      "epoch": 0.10487253952888029,
      "grad_norm": 2.921816825866699,
      "learning_rate": 0.0005,
      "loss": 3.9084839820861816,
      "step": 650
    },
    {
      "epoch": 0.10503388189738626,
      "grad_norm": 1.7043753862380981,
      "learning_rate": 0.0005,
      "loss": 3.514220952987671,
      "step": 651
    },
    {
      "epoch": 0.10519522426589223,
      "grad_norm": 2.0047006607055664,
      "learning_rate": 0.0005,
      "loss": 3.4645376205444336,
      "step": 652
    },
    {
      "epoch": 0.1053565666343982,
      "grad_norm": 1.7603708505630493,
      "learning_rate": 0.0005,
      "loss": 3.5081186294555664,
      "step": 653
    },
    {
      "epoch": 0.10551790900290416,
      "grad_norm": 1.3695534467697144,
      "learning_rate": 0.0005,
      "loss": 3.7667465209960938,
      "step": 654
    },
    {
      "epoch": 0.10567925137141013,
      "grad_norm": 1.8658665418624878,
      "learning_rate": 0.0005,
      "loss": 3.6417179107666016,
      "step": 655
    },
    {
      "epoch": 0.1058405937399161,
      "grad_norm": 1.673029899597168,
      "learning_rate": 0.0005,
      "loss": 3.8428478240966797,
      "step": 656
    },
    {
      "epoch": 0.10600193610842207,
      "grad_norm": 1.2863669395446777,
      "learning_rate": 0.0005,
      "loss": 3.6688029766082764,
      "step": 657
    },
    {
      "epoch": 0.10616327847692804,
      "grad_norm": 1.7678155899047852,
      "learning_rate": 0.0005,
      "loss": 3.5549917221069336,
      "step": 658
    },
    {
      "epoch": 0.10632462084543401,
      "grad_norm": 1.9689021110534668,
      "learning_rate": 0.0005,
      "loss": 3.9144680500030518,
      "step": 659
    },
    {
      "epoch": 0.10648596321393998,
      "grad_norm": 1.5181857347488403,
      "learning_rate": 0.0005,
      "loss": 3.831001043319702,
      "step": 660
    },
    {
      "epoch": 0.10664730558244595,
      "grad_norm": 1.6332857608795166,
      "learning_rate": 0.0005,
      "loss": 3.6389048099517822,
      "step": 661
    },
    {
      "epoch": 0.10680864795095192,
      "grad_norm": 1.6173419952392578,
      "learning_rate": 0.0005,
      "loss": 3.500802993774414,
      "step": 662
    },
    {
      "epoch": 0.10696999031945789,
      "grad_norm": 2.620269775390625,
      "learning_rate": 0.0005,
      "loss": 3.9047622680664062,
      "step": 663
    },
    {
      "epoch": 0.10713133268796386,
      "grad_norm": 1.4282728433609009,
      "learning_rate": 0.0005,
      "loss": 3.808380126953125,
      "step": 664
    },
    {
      "epoch": 0.10729267505646983,
      "grad_norm": 1.4602264165878296,
      "learning_rate": 0.0005,
      "loss": 3.3569955825805664,
      "step": 665
    },
    {
      "epoch": 0.1074540174249758,
      "grad_norm": 1.4232652187347412,
      "learning_rate": 0.0005,
      "loss": 3.4234015941619873,
      "step": 666
    },
    {
      "epoch": 0.10761535979348177,
      "grad_norm": 1.8405320644378662,
      "learning_rate": 0.0005,
      "loss": 3.8232858180999756,
      "step": 667
    },
    {
      "epoch": 0.10777670216198774,
      "grad_norm": 1.809605598449707,
      "learning_rate": 0.0005,
      "loss": 3.737903356552124,
      "step": 668
    },
    {
      "epoch": 0.1079380445304937,
      "grad_norm": 1.5916411876678467,
      "learning_rate": 0.0005,
      "loss": 3.5633249282836914,
      "step": 669
    },
    {
      "epoch": 0.10809938689899967,
      "grad_norm": 1.4563355445861816,
      "learning_rate": 0.0005,
      "loss": 3.906338691711426,
      "step": 670
    },
    {
      "epoch": 0.10826072926750564,
      "grad_norm": 2.2909865379333496,
      "learning_rate": 0.0005,
      "loss": 3.5298843383789062,
      "step": 671
    },
    {
      "epoch": 0.10842207163601161,
      "grad_norm": 1.5812084674835205,
      "learning_rate": 0.0005,
      "loss": 3.6794192790985107,
      "step": 672
    },
    {
      "epoch": 0.10858341400451758,
      "grad_norm": 1.416994571685791,
      "learning_rate": 0.0005,
      "loss": 3.8237547874450684,
      "step": 673
    },
    {
      "epoch": 0.10874475637302355,
      "grad_norm": 1.753266453742981,
      "learning_rate": 0.0005,
      "loss": 3.2612152099609375,
      "step": 674
    },
    {
      "epoch": 0.10890609874152953,
      "grad_norm": 10.773433685302734,
      "learning_rate": 0.0005,
      "loss": 3.467538833618164,
      "step": 675
    },
    {
      "epoch": 0.1090674411100355,
      "grad_norm": 1.9460309743881226,
      "learning_rate": 0.0005,
      "loss": 3.2119226455688477,
      "step": 676
    },
    {
      "epoch": 0.10922878347854147,
      "grad_norm": 1.8510522842407227,
      "learning_rate": 0.0005,
      "loss": 3.5033507347106934,
      "step": 677
    },
    {
      "epoch": 0.10939012584704744,
      "grad_norm": 1.8676763772964478,
      "learning_rate": 0.0005,
      "loss": 3.6965994834899902,
      "step": 678
    },
    {
      "epoch": 0.10955146821555341,
      "grad_norm": 3.1014676094055176,
      "learning_rate": 0.0005,
      "loss": 3.5026681423187256,
      "step": 679
    },
    {
      "epoch": 0.10971281058405938,
      "grad_norm": 1.9413912296295166,
      "learning_rate": 0.0005,
      "loss": 3.550840377807617,
      "step": 680
    },
    {
      "epoch": 0.10987415295256535,
      "grad_norm": 1.608410120010376,
      "learning_rate": 0.0005,
      "loss": 4.025226593017578,
      "step": 681
    },
    {
      "epoch": 0.11003549532107132,
      "grad_norm": 1.4736360311508179,
      "learning_rate": 0.0005,
      "loss": 3.2506895065307617,
      "step": 682
    },
    {
      "epoch": 0.11019683768957728,
      "grad_norm": 2.2749218940734863,
      "learning_rate": 0.0005,
      "loss": 4.135654926300049,
      "step": 683
    },
    {
      "epoch": 0.11035818005808325,
      "grad_norm": 1.8973143100738525,
      "learning_rate": 0.0005,
      "loss": 3.9582266807556152,
      "step": 684
    },
    {
      "epoch": 0.11051952242658922,
      "grad_norm": 1.6463476419448853,
      "learning_rate": 0.0005,
      "loss": 3.501798152923584,
      "step": 685
    },
    {
      "epoch": 0.11068086479509519,
      "grad_norm": 1.690322756767273,
      "learning_rate": 0.0005,
      "loss": 3.4315247535705566,
      "step": 686
    },
    {
      "epoch": 0.11084220716360116,
      "grad_norm": 1.842950463294983,
      "learning_rate": 0.0005,
      "loss": 3.9228129386901855,
      "step": 687
    },
    {
      "epoch": 0.11100354953210713,
      "grad_norm": 1.70718514919281,
      "learning_rate": 0.0005,
      "loss": 3.659071922302246,
      "step": 688
    },
    {
      "epoch": 0.1111648919006131,
      "grad_norm": 1.5132725238800049,
      "learning_rate": 0.0005,
      "loss": 4.2338080406188965,
      "step": 689
    },
    {
      "epoch": 0.11132623426911907,
      "grad_norm": 1.7526116371154785,
      "learning_rate": 0.0005,
      "loss": 3.093094825744629,
      "step": 690
    },
    {
      "epoch": 0.11148757663762504,
      "grad_norm": 2.379584550857544,
      "learning_rate": 0.0005,
      "loss": 3.5191071033477783,
      "step": 691
    },
    {
      "epoch": 0.11164891900613101,
      "grad_norm": 1.6983720064163208,
      "learning_rate": 0.0005,
      "loss": 3.7057712078094482,
      "step": 692
    },
    {
      "epoch": 0.11181026137463698,
      "grad_norm": 1.7082191705703735,
      "learning_rate": 0.0005,
      "loss": 3.141120433807373,
      "step": 693
    },
    {
      "epoch": 0.11197160374314295,
      "grad_norm": 1.5397616624832153,
      "learning_rate": 0.0005,
      "loss": 3.563905954360962,
      "step": 694
    },
    {
      "epoch": 0.11213294611164892,
      "grad_norm": 1.7321102619171143,
      "learning_rate": 0.0005,
      "loss": 3.5807301998138428,
      "step": 695
    },
    {
      "epoch": 0.1122942884801549,
      "grad_norm": 2.205608606338501,
      "learning_rate": 0.0005,
      "loss": 3.6610357761383057,
      "step": 696
    },
    {
      "epoch": 0.11245563084866086,
      "grad_norm": 1.257646918296814,
      "learning_rate": 0.0005,
      "loss": 2.816258430480957,
      "step": 697
    },
    {
      "epoch": 0.11261697321716682,
      "grad_norm": 2.1306159496307373,
      "learning_rate": 0.0005,
      "loss": 2.946949005126953,
      "step": 698
    },
    {
      "epoch": 0.11277831558567279,
      "grad_norm": 1.4372401237487793,
      "learning_rate": 0.0005,
      "loss": 3.4605765342712402,
      "step": 699
    },
    {
      "epoch": 0.11293965795417876,
      "grad_norm": 2.026036262512207,
      "learning_rate": 0.0005,
      "loss": 3.3763539791107178,
      "step": 700
    },
    {
      "epoch": 0.11310100032268473,
      "grad_norm": 1.9275014400482178,
      "learning_rate": 0.0005,
      "loss": 3.9231863021850586,
      "step": 701
    },
    {
      "epoch": 0.1132623426911907,
      "grad_norm": 1.5754579305648804,
      "learning_rate": 0.0005,
      "loss": 3.8819191455841064,
      "step": 702
    },
    {
      "epoch": 0.11342368505969667,
      "grad_norm": 2.1792376041412354,
      "learning_rate": 0.0005,
      "loss": 3.6689648628234863,
      "step": 703
    },
    {
      "epoch": 0.11358502742820265,
      "grad_norm": 1.2766265869140625,
      "learning_rate": 0.0005,
      "loss": 3.6160223484039307,
      "step": 704
    },
    {
      "epoch": 0.11374636979670862,
      "grad_norm": 1.433862328529358,
      "learning_rate": 0.0005,
      "loss": 2.9531805515289307,
      "step": 705
    },
    {
      "epoch": 0.11390771216521459,
      "grad_norm": 1.673296570777893,
      "learning_rate": 0.0005,
      "loss": 3.480828285217285,
      "step": 706
    },
    {
      "epoch": 0.11406905453372056,
      "grad_norm": 1.9420677423477173,
      "learning_rate": 0.0005,
      "loss": 3.9338951110839844,
      "step": 707
    },
    {
      "epoch": 0.11423039690222653,
      "grad_norm": 1.495200753211975,
      "learning_rate": 0.0005,
      "loss": 3.7429330348968506,
      "step": 708
    },
    {
      "epoch": 0.1143917392707325,
      "grad_norm": 1.829703450202942,
      "learning_rate": 0.0005,
      "loss": 3.8970563411712646,
      "step": 709
    },
    {
      "epoch": 0.11455308163923847,
      "grad_norm": 2.4345922470092773,
      "learning_rate": 0.0005,
      "loss": 3.7599966526031494,
      "step": 710
    },
    {
      "epoch": 0.11471442400774444,
      "grad_norm": 2.144434690475464,
      "learning_rate": 0.0005,
      "loss": 3.9889676570892334,
      "step": 711
    },
    {
      "epoch": 0.11487576637625041,
      "grad_norm": 3.413947582244873,
      "learning_rate": 0.0005,
      "loss": 3.5142629146575928,
      "step": 712
    },
    {
      "epoch": 0.11503710874475637,
      "grad_norm": 1.3637521266937256,
      "learning_rate": 0.0005,
      "loss": 3.840996265411377,
      "step": 713
    },
    {
      "epoch": 0.11519845111326234,
      "grad_norm": 1.2567284107208252,
      "learning_rate": 0.0005,
      "loss": 3.5343899726867676,
      "step": 714
    },
    {
      "epoch": 0.11535979348176831,
      "grad_norm": 1.526837944984436,
      "learning_rate": 0.0005,
      "loss": 3.6606085300445557,
      "step": 715
    },
    {
      "epoch": 0.11552113585027428,
      "grad_norm": 1.3729662895202637,
      "learning_rate": 0.0005,
      "loss": 3.3844032287597656,
      "step": 716
    },
    {
      "epoch": 0.11568247821878025,
      "grad_norm": 1.6497611999511719,
      "learning_rate": 0.0005,
      "loss": 3.7438628673553467,
      "step": 717
    },
    {
      "epoch": 0.11584382058728622,
      "grad_norm": 1.4780858755111694,
      "learning_rate": 0.0005,
      "loss": 3.282573938369751,
      "step": 718
    },
    {
      "epoch": 0.11600516295579219,
      "grad_norm": 1.2900711297988892,
      "learning_rate": 0.0005,
      "loss": 3.5955123901367188,
      "step": 719
    },
    {
      "epoch": 0.11616650532429816,
      "grad_norm": 1.941322684288025,
      "learning_rate": 0.0005,
      "loss": 3.6472134590148926,
      "step": 720
    },
    {
      "epoch": 0.11632784769280413,
      "grad_norm": 2.2677438259124756,
      "learning_rate": 0.0005,
      "loss": 3.325273036956787,
      "step": 721
    },
    {
      "epoch": 0.1164891900613101,
      "grad_norm": 1.7121100425720215,
      "learning_rate": 0.0005,
      "loss": 3.3880605697631836,
      "step": 722
    },
    {
      "epoch": 0.11665053242981607,
      "grad_norm": 1.6114319562911987,
      "learning_rate": 0.0005,
      "loss": 3.5403811931610107,
      "step": 723
    },
    {
      "epoch": 0.11681187479832204,
      "grad_norm": 1.7386704683303833,
      "learning_rate": 0.0005,
      "loss": 3.816895008087158,
      "step": 724
    },
    {
      "epoch": 0.11697321716682801,
      "grad_norm": 1.3603243827819824,
      "learning_rate": 0.0005,
      "loss": 3.5007476806640625,
      "step": 725
    },
    {
      "epoch": 0.11713455953533398,
      "grad_norm": 1.3034660816192627,
      "learning_rate": 0.0005,
      "loss": 3.7465128898620605,
      "step": 726
    },
    {
      "epoch": 0.11729590190383996,
      "grad_norm": 1.8711638450622559,
      "learning_rate": 0.0005,
      "loss": 3.831336736679077,
      "step": 727
    },
    {
      "epoch": 0.11745724427234591,
      "grad_norm": 2.90805721282959,
      "learning_rate": 0.0005,
      "loss": 3.255676507949829,
      "step": 728
    },
    {
      "epoch": 0.11761858664085188,
      "grad_norm": 1.8624333143234253,
      "learning_rate": 0.0005,
      "loss": 3.043123722076416,
      "step": 729
    },
    {
      "epoch": 0.11777992900935785,
      "grad_norm": 2.985616445541382,
      "learning_rate": 0.0005,
      "loss": 3.6255149841308594,
      "step": 730
    },
    {
      "epoch": 0.11794127137786382,
      "grad_norm": 1.6458585262298584,
      "learning_rate": 0.0005,
      "loss": 3.6871023178100586,
      "step": 731
    },
    {
      "epoch": 0.1181026137463698,
      "grad_norm": 1.4183112382888794,
      "learning_rate": 0.0005,
      "loss": 3.3129489421844482,
      "step": 732
    },
    {
      "epoch": 0.11826395611487577,
      "grad_norm": 1.623807430267334,
      "learning_rate": 0.0005,
      "loss": 3.4482481479644775,
      "step": 733
    },
    {
      "epoch": 0.11842529848338174,
      "grad_norm": 1.491934061050415,
      "learning_rate": 0.0005,
      "loss": 3.7350471019744873,
      "step": 734
    },
    {
      "epoch": 0.1185866408518877,
      "grad_norm": 1.4209892749786377,
      "learning_rate": 0.0005,
      "loss": 3.831543445587158,
      "step": 735
    },
    {
      "epoch": 0.11874798322039368,
      "grad_norm": 1.995377779006958,
      "learning_rate": 0.0005,
      "loss": 3.279115915298462,
      "step": 736
    },
    {
      "epoch": 0.11890932558889965,
      "grad_norm": 1.428926706314087,
      "learning_rate": 0.0005,
      "loss": 3.3944993019104004,
      "step": 737
    },
    {
      "epoch": 0.11907066795740562,
      "grad_norm": 1.6638051271438599,
      "learning_rate": 0.0005,
      "loss": 3.5130515098571777,
      "step": 738
    },
    {
      "epoch": 0.11923201032591159,
      "grad_norm": 1.896909475326538,
      "learning_rate": 0.0005,
      "loss": 3.770157814025879,
      "step": 739
    },
    {
      "epoch": 0.11939335269441756,
      "grad_norm": 2.0261261463165283,
      "learning_rate": 0.0005,
      "loss": 3.5996737480163574,
      "step": 740
    },
    {
      "epoch": 0.11955469506292353,
      "grad_norm": 1.6223160028457642,
      "learning_rate": 0.0005,
      "loss": 3.5046844482421875,
      "step": 741
    },
    {
      "epoch": 0.11971603743142949,
      "grad_norm": 1.6577415466308594,
      "learning_rate": 0.0005,
      "loss": 3.426295042037964,
      "step": 742
    },
    {
      "epoch": 0.11987737979993546,
      "grad_norm": 1.5803258419036865,
      "learning_rate": 0.0005,
      "loss": 3.0301871299743652,
      "step": 743
    },
    {
      "epoch": 0.12003872216844143,
      "grad_norm": 1.6452780961990356,
      "learning_rate": 0.0005,
      "loss": 3.768505334854126,
      "step": 744
    },
    {
      "epoch": 0.1202000645369474,
      "grad_norm": 1.516381025314331,
      "learning_rate": 0.0005,
      "loss": 3.781809091567993,
      "step": 745
    },
    {
      "epoch": 0.12036140690545337,
      "grad_norm": 1.757521390914917,
      "learning_rate": 0.0005,
      "loss": 3.552809953689575,
      "step": 746
    },
    {
      "epoch": 0.12052274927395934,
      "grad_norm": 1.2513681650161743,
      "learning_rate": 0.0005,
      "loss": 3.515418291091919,
      "step": 747
    },
    {
      "epoch": 0.12068409164246531,
      "grad_norm": 1.5127712488174438,
      "learning_rate": 0.0005,
      "loss": 3.381349802017212,
      "step": 748
    },
    {
      "epoch": 0.12084543401097128,
      "grad_norm": 1.431350827217102,
      "learning_rate": 0.0005,
      "loss": 3.5632028579711914,
      "step": 749
    },
    {
      "epoch": 0.12100677637947725,
      "grad_norm": 1.458520770072937,
      "learning_rate": 0.0005,
      "loss": 3.7520456314086914,
      "step": 750
    },
    {
      "epoch": 0.12116811874798322,
      "grad_norm": 1.7683721780776978,
      "learning_rate": 0.0005,
      "loss": 3.626919984817505,
      "step": 751
    },
    {
      "epoch": 0.12132946111648919,
      "grad_norm": 1.5293382406234741,
      "learning_rate": 0.0005,
      "loss": 3.6118836402893066,
      "step": 752
    },
    {
      "epoch": 0.12149080348499516,
      "grad_norm": 2.2204701900482178,
      "learning_rate": 0.0005,
      "loss": 3.613513946533203,
      "step": 753
    },
    {
      "epoch": 0.12165214585350113,
      "grad_norm": 2.3012430667877197,
      "learning_rate": 0.0005,
      "loss": 3.687070369720459,
      "step": 754
    },
    {
      "epoch": 0.1218134882220071,
      "grad_norm": 1.6849340200424194,
      "learning_rate": 0.0005,
      "loss": 3.858905076980591,
      "step": 755
    },
    {
      "epoch": 0.12197483059051308,
      "grad_norm": 1.498904824256897,
      "learning_rate": 0.0005,
      "loss": 3.3328258991241455,
      "step": 756
    },
    {
      "epoch": 0.12213617295901903,
      "grad_norm": 1.5799874067306519,
      "learning_rate": 0.0005,
      "loss": 3.098871946334839,
      "step": 757
    },
    {
      "epoch": 0.122297515327525,
      "grad_norm": 2.6529529094696045,
      "learning_rate": 0.0005,
      "loss": 3.580103635787964,
      "step": 758
    },
    {
      "epoch": 0.12245885769603097,
      "grad_norm": 1.7314424514770508,
      "learning_rate": 0.0005,
      "loss": 3.951362371444702,
      "step": 759
    },
    {
      "epoch": 0.12262020006453694,
      "grad_norm": 3.0530784130096436,
      "learning_rate": 0.0005,
      "loss": 3.376746892929077,
      "step": 760
    },
    {
      "epoch": 0.12278154243304291,
      "grad_norm": 1.941941499710083,
      "learning_rate": 0.0005,
      "loss": 3.8482937812805176,
      "step": 761
    },
    {
      "epoch": 0.12294288480154889,
      "grad_norm": 1.6782170534133911,
      "learning_rate": 0.0005,
      "loss": 4.044387340545654,
      "step": 762
    },
    {
      "epoch": 0.12310422717005486,
      "grad_norm": 1.2677851915359497,
      "learning_rate": 0.0005,
      "loss": 3.227829933166504,
      "step": 763
    },
    {
      "epoch": 0.12326556953856083,
      "grad_norm": 1.2089550495147705,
      "learning_rate": 0.0005,
      "loss": 3.8102493286132812,
      "step": 764
    },
    {
      "epoch": 0.1234269119070668,
      "grad_norm": 1.5285348892211914,
      "learning_rate": 0.0005,
      "loss": 4.191930770874023,
      "step": 765
    },
    {
      "epoch": 0.12358825427557277,
      "grad_norm": 1.418165922164917,
      "learning_rate": 0.0005,
      "loss": 4.145941734313965,
      "step": 766
    },
    {
      "epoch": 0.12374959664407874,
      "grad_norm": 1.4954966306686401,
      "learning_rate": 0.0005,
      "loss": 3.385709762573242,
      "step": 767
    },
    {
      "epoch": 0.12391093901258471,
      "grad_norm": 1.3804562091827393,
      "learning_rate": 0.0005,
      "loss": 3.4121882915496826,
      "step": 768
    },
    {
      "epoch": 0.12407228138109068,
      "grad_norm": 1.4898186922073364,
      "learning_rate": 0.0005,
      "loss": 3.7483532428741455,
      "step": 769
    },
    {
      "epoch": 0.12423362374959665,
      "grad_norm": 1.720102310180664,
      "learning_rate": 0.0005,
      "loss": 3.505009174346924,
      "step": 770
    },
    {
      "epoch": 0.12439496611810262,
      "grad_norm": 2.3228728771209717,
      "learning_rate": 0.0005,
      "loss": 3.530769109725952,
      "step": 771
    },
    {
      "epoch": 0.12455630848660858,
      "grad_norm": 1.3990052938461304,
      "learning_rate": 0.0005,
      "loss": 3.4108142852783203,
      "step": 772
    },
    {
      "epoch": 0.12471765085511455,
      "grad_norm": 1.458388328552246,
      "learning_rate": 0.0005,
      "loss": 3.700734853744507,
      "step": 773
    },
    {
      "epoch": 0.12487899322362052,
      "grad_norm": 1.6071608066558838,
      "learning_rate": 0.0005,
      "loss": 3.536651134490967,
      "step": 774
    },
    {
      "epoch": 0.1250403355921265,
      "grad_norm": 1.8028730154037476,
      "learning_rate": 0.0005,
      "loss": 3.6978180408477783,
      "step": 775
    },
    {
      "epoch": 0.12520167796063247,
      "grad_norm": 1.2061434984207153,
      "learning_rate": 0.0005,
      "loss": 3.7061519622802734,
      "step": 776
    },
    {
      "epoch": 0.12536302032913843,
      "grad_norm": 2.042872190475464,
      "learning_rate": 0.0005,
      "loss": 3.9202394485473633,
      "step": 777
    },
    {
      "epoch": 0.12552436269764441,
      "grad_norm": 1.5952022075653076,
      "learning_rate": 0.0005,
      "loss": 3.577633857727051,
      "step": 778
    },
    {
      "epoch": 0.12568570506615037,
      "grad_norm": 1.6751669645309448,
      "learning_rate": 0.0005,
      "loss": 3.3786301612854004,
      "step": 779
    },
    {
      "epoch": 0.12584704743465633,
      "grad_norm": 1.7737646102905273,
      "learning_rate": 0.0005,
      "loss": 4.122894763946533,
      "step": 780
    },
    {
      "epoch": 0.1260083898031623,
      "grad_norm": 1.6178051233291626,
      "learning_rate": 0.0005,
      "loss": 3.4551820755004883,
      "step": 781
    },
    {
      "epoch": 0.12616973217166827,
      "grad_norm": 1.597787618637085,
      "learning_rate": 0.0005,
      "loss": 3.881776809692383,
      "step": 782
    },
    {
      "epoch": 0.12633107454017425,
      "grad_norm": 1.6388062238693237,
      "learning_rate": 0.0005,
      "loss": 4.081183433532715,
      "step": 783
    },
    {
      "epoch": 0.1264924169086802,
      "grad_norm": 1.596035122871399,
      "learning_rate": 0.0005,
      "loss": 3.765995502471924,
      "step": 784
    },
    {
      "epoch": 0.1266537592771862,
      "grad_norm": 2.004638671875,
      "learning_rate": 0.0005,
      "loss": 4.096636772155762,
      "step": 785
    },
    {
      "epoch": 0.12681510164569215,
      "grad_norm": 1.4459642171859741,
      "learning_rate": 0.0005,
      "loss": 3.33119797706604,
      "step": 786
    },
    {
      "epoch": 0.12697644401419814,
      "grad_norm": 1.242882251739502,
      "learning_rate": 0.0005,
      "loss": 3.5069189071655273,
      "step": 787
    },
    {
      "epoch": 0.1271377863827041,
      "grad_norm": 1.3100908994674683,
      "learning_rate": 0.0005,
      "loss": 3.6641526222229004,
      "step": 788
    },
    {
      "epoch": 0.12729912875121008,
      "grad_norm": 1.5502554178237915,
      "learning_rate": 0.0005,
      "loss": 3.3860535621643066,
      "step": 789
    },
    {
      "epoch": 0.12746047111971603,
      "grad_norm": 1.7881311178207397,
      "learning_rate": 0.0005,
      "loss": 3.7863636016845703,
      "step": 790
    },
    {
      "epoch": 0.12762181348822202,
      "grad_norm": 1.487774133682251,
      "learning_rate": 0.0005,
      "loss": 3.66518235206604,
      "step": 791
    },
    {
      "epoch": 0.12778315585672798,
      "grad_norm": 1.4973691701889038,
      "learning_rate": 0.0005,
      "loss": 3.922816753387451,
      "step": 792
    },
    {
      "epoch": 0.12794449822523396,
      "grad_norm": 1.8439723253250122,
      "learning_rate": 0.0005,
      "loss": 3.5129103660583496,
      "step": 793
    },
    {
      "epoch": 0.12810584059373992,
      "grad_norm": 1.7774319648742676,
      "learning_rate": 0.0005,
      "loss": 3.9353017807006836,
      "step": 794
    },
    {
      "epoch": 0.12826718296224587,
      "grad_norm": 1.6239296197891235,
      "learning_rate": 0.0005,
      "loss": 3.609863758087158,
      "step": 795
    },
    {
      "epoch": 0.12842852533075186,
      "grad_norm": 1.1920322179794312,
      "learning_rate": 0.0005,
      "loss": 3.9077820777893066,
      "step": 796
    },
    {
      "epoch": 0.12858986769925781,
      "grad_norm": 1.9473388195037842,
      "learning_rate": 0.0005,
      "loss": 3.46964168548584,
      "step": 797
    },
    {
      "epoch": 0.1287512100677638,
      "grad_norm": 1.330611228942871,
      "learning_rate": 0.0005,
      "loss": 3.6634039878845215,
      "step": 798
    },
    {
      "epoch": 0.12891255243626976,
      "grad_norm": 1.1996945142745972,
      "learning_rate": 0.0005,
      "loss": 3.746753454208374,
      "step": 799
    },
    {
      "epoch": 0.12907389480477574,
      "grad_norm": 1.2095850706100464,
      "learning_rate": 0.0005,
      "loss": 3.5981931686401367,
      "step": 800
    },
    {
      "epoch": 0.1292352371732817,
      "grad_norm": 1.647594928741455,
      "learning_rate": 0.0005,
      "loss": 3.309957981109619,
      "step": 801
    },
    {
      "epoch": 0.12939657954178768,
      "grad_norm": 1.4594805240631104,
      "learning_rate": 0.0005,
      "loss": 3.3303322792053223,
      "step": 802
    },
    {
      "epoch": 0.12955792191029364,
      "grad_norm": 1.6483070850372314,
      "learning_rate": 0.0005,
      "loss": 3.6949691772460938,
      "step": 803
    },
    {
      "epoch": 0.12971926427879962,
      "grad_norm": 1.670823574066162,
      "learning_rate": 0.0005,
      "loss": 3.397998094558716,
      "step": 804
    },
    {
      "epoch": 0.12988060664730558,
      "grad_norm": 1.8095999956130981,
      "learning_rate": 0.0005,
      "loss": 3.4727213382720947,
      "step": 805
    },
    {
      "epoch": 0.13004194901581156,
      "grad_norm": 1.8851726055145264,
      "learning_rate": 0.0005,
      "loss": 3.538837432861328,
      "step": 806
    },
    {
      "epoch": 0.13020329138431752,
      "grad_norm": 3.09908390045166,
      "learning_rate": 0.0005,
      "loss": 3.955477476119995,
      "step": 807
    },
    {
      "epoch": 0.13036463375282348,
      "grad_norm": 1.7193939685821533,
      "learning_rate": 0.0005,
      "loss": 3.7371249198913574,
      "step": 808
    },
    {
      "epoch": 0.13052597612132946,
      "grad_norm": 1.3917896747589111,
      "learning_rate": 0.0005,
      "loss": 3.3348746299743652,
      "step": 809
    },
    {
      "epoch": 0.13068731848983542,
      "grad_norm": 5.029810428619385,
      "learning_rate": 0.0005,
      "loss": 3.17476487159729,
      "step": 810
    },
    {
      "epoch": 0.1308486608583414,
      "grad_norm": 1.7548437118530273,
      "learning_rate": 0.0005,
      "loss": 2.9642996788024902,
      "step": 811
    },
    {
      "epoch": 0.13101000322684736,
      "grad_norm": 1.965108871459961,
      "learning_rate": 0.0005,
      "loss": 3.715148448944092,
      "step": 812
    },
    {
      "epoch": 0.13117134559535334,
      "grad_norm": 1.9814929962158203,
      "learning_rate": 0.0005,
      "loss": 3.4277803897857666,
      "step": 813
    },
    {
      "epoch": 0.1313326879638593,
      "grad_norm": 1.9255709648132324,
      "learning_rate": 0.0005,
      "loss": 3.523958206176758,
      "step": 814
    },
    {
      "epoch": 0.13149403033236529,
      "grad_norm": 1.2582411766052246,
      "learning_rate": 0.0005,
      "loss": 3.714118480682373,
      "step": 815
    },
    {
      "epoch": 0.13165537270087124,
      "grad_norm": 1.746360421180725,
      "learning_rate": 0.0005,
      "loss": 3.3096652030944824,
      "step": 816
    },
    {
      "epoch": 0.13181671506937723,
      "grad_norm": 1.7183223962783813,
      "learning_rate": 0.0005,
      "loss": 3.3262267112731934,
      "step": 817
    },
    {
      "epoch": 0.13197805743788318,
      "grad_norm": 1.6819164752960205,
      "learning_rate": 0.0005,
      "loss": 3.276188373565674,
      "step": 818
    },
    {
      "epoch": 0.13213939980638917,
      "grad_norm": 1.3272889852523804,
      "learning_rate": 0.0005,
      "loss": 3.3610475063323975,
      "step": 819
    },
    {
      "epoch": 0.13230074217489513,
      "grad_norm": 1.552462100982666,
      "learning_rate": 0.0005,
      "loss": 3.5476794242858887,
      "step": 820
    },
    {
      "epoch": 0.1324620845434011,
      "grad_norm": 1.393373727798462,
      "learning_rate": 0.0005,
      "loss": 3.756437301635742,
      "step": 821
    },
    {
      "epoch": 0.13262342691190707,
      "grad_norm": 1.5245839357376099,
      "learning_rate": 0.0005,
      "loss": 3.7943460941314697,
      "step": 822
    },
    {
      "epoch": 0.13278476928041302,
      "grad_norm": 1.261261224746704,
      "learning_rate": 0.0005,
      "loss": 3.778229236602783,
      "step": 823
    },
    {
      "epoch": 0.132946111648919,
      "grad_norm": 1.432022213935852,
      "learning_rate": 0.0005,
      "loss": 3.6689021587371826,
      "step": 824
    },
    {
      "epoch": 0.13310745401742496,
      "grad_norm": 1.4199793338775635,
      "learning_rate": 0.0005,
      "loss": 3.5761566162109375,
      "step": 825
    },
    {
      "epoch": 0.13326879638593095,
      "grad_norm": 1.755669116973877,
      "learning_rate": 0.0005,
      "loss": 3.8138809204101562,
      "step": 826
    },
    {
      "epoch": 0.1334301387544369,
      "grad_norm": 1.7608661651611328,
      "learning_rate": 0.0005,
      "loss": 3.562490224838257,
      "step": 827
    },
    {
      "epoch": 0.1335914811229429,
      "grad_norm": 17.46319580078125,
      "learning_rate": 0.0005,
      "loss": 3.559154987335205,
      "step": 828
    },
    {
      "epoch": 0.13375282349144885,
      "grad_norm": 1.4868404865264893,
      "learning_rate": 0.0005,
      "loss": 3.6110143661499023,
      "step": 829
    },
    {
      "epoch": 0.13391416585995483,
      "grad_norm": 1.5748826265335083,
      "learning_rate": 0.0005,
      "loss": 3.721944808959961,
      "step": 830
    },
    {
      "epoch": 0.1340755082284608,
      "grad_norm": 1.824604868888855,
      "learning_rate": 0.0005,
      "loss": 3.4709620475769043,
      "step": 831
    },
    {
      "epoch": 0.13423685059696677,
      "grad_norm": 1.6895760297775269,
      "learning_rate": 0.0005,
      "loss": 3.3240866661071777,
      "step": 832
    },
    {
      "epoch": 0.13439819296547273,
      "grad_norm": 1.539539098739624,
      "learning_rate": 0.0005,
      "loss": 3.379225730895996,
      "step": 833
    },
    {
      "epoch": 0.1345595353339787,
      "grad_norm": 1.4918872117996216,
      "learning_rate": 0.0005,
      "loss": 3.50476336479187,
      "step": 834
    },
    {
      "epoch": 0.13472087770248467,
      "grad_norm": 1.484108328819275,
      "learning_rate": 0.0005,
      "loss": 4.003046035766602,
      "step": 835
    },
    {
      "epoch": 0.13488222007099065,
      "grad_norm": 1.4288097620010376,
      "learning_rate": 0.0005,
      "loss": 3.3139469623565674,
      "step": 836
    },
    {
      "epoch": 0.1350435624394966,
      "grad_norm": 1.7283989191055298,
      "learning_rate": 0.0005,
      "loss": 3.583401679992676,
      "step": 837
    },
    {
      "epoch": 0.13520490480800257,
      "grad_norm": 1.933901071548462,
      "learning_rate": 0.0005,
      "loss": 3.573788642883301,
      "step": 838
    },
    {
      "epoch": 0.13536624717650855,
      "grad_norm": 1.3437460660934448,
      "learning_rate": 0.0005,
      "loss": 3.466886520385742,
      "step": 839
    },
    {
      "epoch": 0.1355275895450145,
      "grad_norm": 1.7231711149215698,
      "learning_rate": 0.0005,
      "loss": 3.4997918605804443,
      "step": 840
    },
    {
      "epoch": 0.1356889319135205,
      "grad_norm": 2.2514779567718506,
      "learning_rate": 0.0005,
      "loss": 3.3013267517089844,
      "step": 841
    },
    {
      "epoch": 0.13585027428202645,
      "grad_norm": 1.48702073097229,
      "learning_rate": 0.0005,
      "loss": 3.221001148223877,
      "step": 842
    },
    {
      "epoch": 0.13601161665053244,
      "grad_norm": 1.61322820186615,
      "learning_rate": 0.0005,
      "loss": 3.306002616882324,
      "step": 843
    },
    {
      "epoch": 0.1361729590190384,
      "grad_norm": 1.5445975065231323,
      "learning_rate": 0.0005,
      "loss": 3.8556721210479736,
      "step": 844
    },
    {
      "epoch": 0.13633430138754438,
      "grad_norm": 1.4747039079666138,
      "learning_rate": 0.0005,
      "loss": 3.7116475105285645,
      "step": 845
    },
    {
      "epoch": 0.13649564375605033,
      "grad_norm": 1.7202606201171875,
      "learning_rate": 0.0005,
      "loss": 3.2447218894958496,
      "step": 846
    },
    {
      "epoch": 0.13665698612455632,
      "grad_norm": 1.5640136003494263,
      "learning_rate": 0.0005,
      "loss": 3.4208877086639404,
      "step": 847
    },
    {
      "epoch": 0.13681832849306227,
      "grad_norm": 1.8231593370437622,
      "learning_rate": 0.0005,
      "loss": 3.8414056301116943,
      "step": 848
    },
    {
      "epoch": 0.13697967086156826,
      "grad_norm": 1.564974069595337,
      "learning_rate": 0.0005,
      "loss": 3.7660975456237793,
      "step": 849
    },
    {
      "epoch": 0.13714101323007422,
      "grad_norm": 1.4165441989898682,
      "learning_rate": 0.0005,
      "loss": 3.7834067344665527,
      "step": 850
    },
    {
      "epoch": 0.1373023555985802,
      "grad_norm": 4.667461395263672,
      "learning_rate": 0.0005,
      "loss": 4.657867431640625,
      "step": 851
    },
    {
      "epoch": 0.13746369796708616,
      "grad_norm": 1.6188974380493164,
      "learning_rate": 0.0005,
      "loss": 3.2623825073242188,
      "step": 852
    },
    {
      "epoch": 0.1376250403355921,
      "grad_norm": 1.6293781995773315,
      "learning_rate": 0.0005,
      "loss": 3.7336299419403076,
      "step": 853
    },
    {
      "epoch": 0.1377863827040981,
      "grad_norm": 3.745065450668335,
      "learning_rate": 0.0005,
      "loss": 3.682766914367676,
      "step": 854
    },
    {
      "epoch": 0.13794772507260405,
      "grad_norm": 1.3759311437606812,
      "learning_rate": 0.0005,
      "loss": 3.702366352081299,
      "step": 855
    },
    {
      "epoch": 0.13810906744111004,
      "grad_norm": 1.5285018682479858,
      "learning_rate": 0.0005,
      "loss": 3.336226224899292,
      "step": 856
    },
    {
      "epoch": 0.138270409809616,
      "grad_norm": 1.3901724815368652,
      "learning_rate": 0.0005,
      "loss": 4.120444297790527,
      "step": 857
    },
    {
      "epoch": 0.13843175217812198,
      "grad_norm": 1.5157381296157837,
      "learning_rate": 0.0005,
      "loss": 3.596156358718872,
      "step": 858
    },
    {
      "epoch": 0.13859309454662794,
      "grad_norm": 4.4763264656066895,
      "learning_rate": 0.0005,
      "loss": 3.749938488006592,
      "step": 859
    },
    {
      "epoch": 0.13875443691513392,
      "grad_norm": 2.3771579265594482,
      "learning_rate": 0.0005,
      "loss": 3.75701904296875,
      "step": 860
    },
    {
      "epoch": 0.13891577928363988,
      "grad_norm": 1.393933892250061,
      "learning_rate": 0.0005,
      "loss": 3.6234662532806396,
      "step": 861
    },
    {
      "epoch": 0.13907712165214586,
      "grad_norm": 1.439180612564087,
      "learning_rate": 0.0005,
      "loss": 3.5447838306427,
      "step": 862
    },
    {
      "epoch": 0.13923846402065182,
      "grad_norm": 1.4210115671157837,
      "learning_rate": 0.0005,
      "loss": 3.748605966567993,
      "step": 863
    },
    {
      "epoch": 0.1393998063891578,
      "grad_norm": 2.48468279838562,
      "learning_rate": 0.0005,
      "loss": 3.7636585235595703,
      "step": 864
    },
    {
      "epoch": 0.13956114875766376,
      "grad_norm": 1.7228893041610718,
      "learning_rate": 0.0005,
      "loss": 3.178345203399658,
      "step": 865
    },
    {
      "epoch": 0.13972249112616975,
      "grad_norm": 2.0168073177337646,
      "learning_rate": 0.0005,
      "loss": 3.485015869140625,
      "step": 866
    },
    {
      "epoch": 0.1398838334946757,
      "grad_norm": 1.5587278604507446,
      "learning_rate": 0.0005,
      "loss": 3.712435722351074,
      "step": 867
    },
    {
      "epoch": 0.14004517586318166,
      "grad_norm": 1.2164067029953003,
      "learning_rate": 0.0005,
      "loss": 3.120537757873535,
      "step": 868
    },
    {
      "epoch": 0.14020651823168764,
      "grad_norm": 1.4195889234542847,
      "learning_rate": 0.0005,
      "loss": 3.440117835998535,
      "step": 869
    },
    {
      "epoch": 0.1403678606001936,
      "grad_norm": 1.5270063877105713,
      "learning_rate": 0.0005,
      "loss": 3.8141660690307617,
      "step": 870
    },
    {
      "epoch": 0.14052920296869958,
      "grad_norm": 1.6922060251235962,
      "learning_rate": 0.0005,
      "loss": 3.438610076904297,
      "step": 871
    },
    {
      "epoch": 0.14069054533720554,
      "grad_norm": 1.5541877746582031,
      "learning_rate": 0.0005,
      "loss": 3.2465662956237793,
      "step": 872
    },
    {
      "epoch": 0.14085188770571153,
      "grad_norm": 1.5422232151031494,
      "learning_rate": 0.0005,
      "loss": 3.1972780227661133,
      "step": 873
    },
    {
      "epoch": 0.14101323007421748,
      "grad_norm": 1.4411174058914185,
      "learning_rate": 0.0005,
      "loss": 3.3112058639526367,
      "step": 874
    },
    {
      "epoch": 0.14117457244272347,
      "grad_norm": 1.5205059051513672,
      "learning_rate": 0.0005,
      "loss": 3.4515020847320557,
      "step": 875
    },
    {
      "epoch": 0.14133591481122942,
      "grad_norm": 1.5687346458435059,
      "learning_rate": 0.0005,
      "loss": 3.272386312484741,
      "step": 876
    },
    {
      "epoch": 0.1414972571797354,
      "grad_norm": 1.6951626539230347,
      "learning_rate": 0.0005,
      "loss": 3.8557732105255127,
      "step": 877
    },
    {
      "epoch": 0.14165859954824137,
      "grad_norm": 1.3426250219345093,
      "learning_rate": 0.0005,
      "loss": 3.4440665245056152,
      "step": 878
    },
    {
      "epoch": 0.14181994191674735,
      "grad_norm": 1.6219878196716309,
      "learning_rate": 0.0005,
      "loss": 3.2630138397216797,
      "step": 879
    },
    {
      "epoch": 0.1419812842852533,
      "grad_norm": 2.1419177055358887,
      "learning_rate": 0.0005,
      "loss": 3.665806531906128,
      "step": 880
    },
    {
      "epoch": 0.1421426266537593,
      "grad_norm": 2.0957834720611572,
      "learning_rate": 0.0005,
      "loss": 4.145696640014648,
      "step": 881
    },
    {
      "epoch": 0.14230396902226525,
      "grad_norm": 2.219362497329712,
      "learning_rate": 0.0005,
      "loss": 3.390934944152832,
      "step": 882
    },
    {
      "epoch": 0.1424653113907712,
      "grad_norm": 1.2425482273101807,
      "learning_rate": 0.0005,
      "loss": 4.0453643798828125,
      "step": 883
    },
    {
      "epoch": 0.1426266537592772,
      "grad_norm": 1.4673327207565308,
      "learning_rate": 0.0005,
      "loss": 3.4534101486206055,
      "step": 884
    },
    {
      "epoch": 0.14278799612778315,
      "grad_norm": 1.569749116897583,
      "learning_rate": 0.0005,
      "loss": 3.1553266048431396,
      "step": 885
    },
    {
      "epoch": 0.14294933849628913,
      "grad_norm": 1.514225959777832,
      "learning_rate": 0.0005,
      "loss": 3.2551302909851074,
      "step": 886
    },
    {
      "epoch": 0.1431106808647951,
      "grad_norm": 1.38337242603302,
      "learning_rate": 0.0005,
      "loss": 3.2722859382629395,
      "step": 887
    },
    {
      "epoch": 0.14327202323330107,
      "grad_norm": 1.5350255966186523,
      "learning_rate": 0.0005,
      "loss": 3.306922674179077,
      "step": 888
    },
    {
      "epoch": 0.14343336560180703,
      "grad_norm": 2.1089670658111572,
      "learning_rate": 0.0005,
      "loss": 3.5600316524505615,
      "step": 889
    },
    {
      "epoch": 0.143594707970313,
      "grad_norm": 1.635948896408081,
      "learning_rate": 0.0005,
      "loss": 3.5272364616394043,
      "step": 890
    },
    {
      "epoch": 0.14375605033881897,
      "grad_norm": 1.8033466339111328,
      "learning_rate": 0.0005,
      "loss": 2.9883809089660645,
      "step": 891
    },
    {
      "epoch": 0.14391739270732495,
      "grad_norm": 1.737288475036621,
      "learning_rate": 0.0005,
      "loss": 3.5016255378723145,
      "step": 892
    },
    {
      "epoch": 0.1440787350758309,
      "grad_norm": 1.3755000829696655,
      "learning_rate": 0.0005,
      "loss": 3.8264660835266113,
      "step": 893
    },
    {
      "epoch": 0.1442400774443369,
      "grad_norm": 1.9098230600357056,
      "learning_rate": 0.0005,
      "loss": 3.5962367057800293,
      "step": 894
    },
    {
      "epoch": 0.14440141981284285,
      "grad_norm": 1.739730954170227,
      "learning_rate": 0.0005,
      "loss": 3.8707079887390137,
      "step": 895
    },
    {
      "epoch": 0.1445627621813488,
      "grad_norm": 1.5100637674331665,
      "learning_rate": 0.0005,
      "loss": 3.355877161026001,
      "step": 896
    },
    {
      "epoch": 0.1447241045498548,
      "grad_norm": 1.4743331670761108,
      "learning_rate": 0.0005,
      "loss": 3.6528236865997314,
      "step": 897
    },
    {
      "epoch": 0.14488544691836075,
      "grad_norm": 1.6530674695968628,
      "learning_rate": 0.0005,
      "loss": 3.4201114177703857,
      "step": 898
    },
    {
      "epoch": 0.14504678928686673,
      "grad_norm": 1.32415771484375,
      "learning_rate": 0.0005,
      "loss": 3.75655198097229,
      "step": 899
    },
    {
      "epoch": 0.1452081316553727,
      "grad_norm": 1.5974175930023193,
      "learning_rate": 0.0005,
      "loss": 3.9036126136779785,
      "step": 900
    },
    {
      "epoch": 0.14536947402387868,
      "grad_norm": 1.11509108543396,
      "learning_rate": 0.0005,
      "loss": 3.8676557540893555,
      "step": 901
    },
    {
      "epoch": 0.14553081639238463,
      "grad_norm": 1.3995288610458374,
      "learning_rate": 0.0005,
      "loss": 3.3873403072357178,
      "step": 902
    },
    {
      "epoch": 0.14569215876089062,
      "grad_norm": 1.3228325843811035,
      "learning_rate": 0.0005,
      "loss": 3.998408317565918,
      "step": 903
    },
    {
      "epoch": 0.14585350112939657,
      "grad_norm": 1.3391789197921753,
      "learning_rate": 0.0005,
      "loss": 3.6667022705078125,
      "step": 904
    },
    {
      "epoch": 0.14601484349790256,
      "grad_norm": 1.683939814567566,
      "learning_rate": 0.0005,
      "loss": 3.5936813354492188,
      "step": 905
    },
    {
      "epoch": 0.14617618586640851,
      "grad_norm": 1.668615698814392,
      "learning_rate": 0.0005,
      "loss": 3.82525372505188,
      "step": 906
    },
    {
      "epoch": 0.1463375282349145,
      "grad_norm": 1.5009410381317139,
      "learning_rate": 0.0005,
      "loss": 3.780808687210083,
      "step": 907
    },
    {
      "epoch": 0.14649887060342046,
      "grad_norm": 1.6092071533203125,
      "learning_rate": 0.0005,
      "loss": 3.178295373916626,
      "step": 908
    },
    {
      "epoch": 0.14666021297192644,
      "grad_norm": 1.293674349784851,
      "learning_rate": 0.0005,
      "loss": 3.635409355163574,
      "step": 909
    },
    {
      "epoch": 0.1468215553404324,
      "grad_norm": 1.643517255783081,
      "learning_rate": 0.0005,
      "loss": 3.3410863876342773,
      "step": 910
    },
    {
      "epoch": 0.14698289770893835,
      "grad_norm": 1.2056348323822021,
      "learning_rate": 0.0005,
      "loss": 3.7187116146087646,
      "step": 911
    },
    {
      "epoch": 0.14714424007744434,
      "grad_norm": 2.8605306148529053,
      "learning_rate": 0.0005,
      "loss": 3.7692112922668457,
      "step": 912
    },
    {
      "epoch": 0.1473055824459503,
      "grad_norm": 1.3898807764053345,
      "learning_rate": 0.0005,
      "loss": 2.966289758682251,
      "step": 913
    },
    {
      "epoch": 0.14746692481445628,
      "grad_norm": 1.6349345445632935,
      "learning_rate": 0.0005,
      "loss": 3.646000623703003,
      "step": 914
    },
    {
      "epoch": 0.14762826718296224,
      "grad_norm": 1.797984004020691,
      "learning_rate": 0.0005,
      "loss": 3.799182415008545,
      "step": 915
    },
    {
      "epoch": 0.14778960955146822,
      "grad_norm": 1.4224592447280884,
      "learning_rate": 0.0005,
      "loss": 3.8296937942504883,
      "step": 916
    },
    {
      "epoch": 0.14795095191997418,
      "grad_norm": 1.7340494394302368,
      "learning_rate": 0.0005,
      "loss": 3.000173568725586,
      "step": 917
    },
    {
      "epoch": 0.14811229428848016,
      "grad_norm": 1.9698902368545532,
      "learning_rate": 0.0005,
      "loss": 3.609966516494751,
      "step": 918
    },
    {
      "epoch": 0.14827363665698612,
      "grad_norm": 1.7655057907104492,
      "learning_rate": 0.0005,
      "loss": 3.2412326335906982,
      "step": 919
    },
    {
      "epoch": 0.1484349790254921,
      "grad_norm": 1.3669871091842651,
      "learning_rate": 0.0005,
      "loss": 3.074787139892578,
      "step": 920
    },
    {
      "epoch": 0.14859632139399806,
      "grad_norm": 2.3543660640716553,
      "learning_rate": 0.0005,
      "loss": 3.7281525135040283,
      "step": 921
    },
    {
      "epoch": 0.14875766376250404,
      "grad_norm": 2.0077872276306152,
      "learning_rate": 0.0005,
      "loss": 3.4539895057678223,
      "step": 922
    },
    {
      "epoch": 0.14891900613101,
      "grad_norm": 1.4948302507400513,
      "learning_rate": 0.0005,
      "loss": 3.7638001441955566,
      "step": 923
    },
    {
      "epoch": 0.14908034849951599,
      "grad_norm": 1.6121330261230469,
      "learning_rate": 0.0005,
      "loss": 3.2690818309783936,
      "step": 924
    },
    {
      "epoch": 0.14924169086802194,
      "grad_norm": 1.582528829574585,
      "learning_rate": 0.0005,
      "loss": 3.1509876251220703,
      "step": 925
    },
    {
      "epoch": 0.1494030332365279,
      "grad_norm": 1.2329472303390503,
      "learning_rate": 0.0005,
      "loss": 3.2819440364837646,
      "step": 926
    },
    {
      "epoch": 0.14956437560503388,
      "grad_norm": 1.5682544708251953,
      "learning_rate": 0.0005,
      "loss": 3.5550143718719482,
      "step": 927
    },
    {
      "epoch": 0.14972571797353984,
      "grad_norm": 1.4086750745773315,
      "learning_rate": 0.0005,
      "loss": 3.4318132400512695,
      "step": 928
    },
    {
      "epoch": 0.14988706034204582,
      "grad_norm": 1.3997164964675903,
      "learning_rate": 0.0005,
      "loss": 3.626451253890991,
      "step": 929
    },
    {
      "epoch": 0.15004840271055178,
      "grad_norm": 1.715841293334961,
      "learning_rate": 0.0005,
      "loss": 3.6098856925964355,
      "step": 930
    },
    {
      "epoch": 0.15020974507905777,
      "grad_norm": 1.8333563804626465,
      "learning_rate": 0.0005,
      "loss": 3.332181930541992,
      "step": 931
    },
    {
      "epoch": 0.15037108744756372,
      "grad_norm": 1.9552700519561768,
      "learning_rate": 0.0005,
      "loss": 2.9092235565185547,
      "step": 932
    },
    {
      "epoch": 0.1505324298160697,
      "grad_norm": 1.6218507289886475,
      "learning_rate": 0.0005,
      "loss": 3.2810990810394287,
      "step": 933
    },
    {
      "epoch": 0.15069377218457566,
      "grad_norm": 1.7448561191558838,
      "learning_rate": 0.0005,
      "loss": 3.168628692626953,
      "step": 934
    },
    {
      "epoch": 0.15085511455308165,
      "grad_norm": 1.6998567581176758,
      "learning_rate": 0.0005,
      "loss": 3.3323352336883545,
      "step": 935
    },
    {
      "epoch": 0.1510164569215876,
      "grad_norm": 1.4951517581939697,
      "learning_rate": 0.0005,
      "loss": 3.704010486602783,
      "step": 936
    },
    {
      "epoch": 0.1511777992900936,
      "grad_norm": 1.451547622680664,
      "learning_rate": 0.0005,
      "loss": 3.5507588386535645,
      "step": 937
    },
    {
      "epoch": 0.15133914165859955,
      "grad_norm": 1.7412854433059692,
      "learning_rate": 0.0005,
      "loss": 2.8525900840759277,
      "step": 938
    },
    {
      "epoch": 0.15150048402710553,
      "grad_norm": 1.434116244316101,
      "learning_rate": 0.0005,
      "loss": 3.6309688091278076,
      "step": 939
    },
    {
      "epoch": 0.1516618263956115,
      "grad_norm": 1.2791624069213867,
      "learning_rate": 0.0005,
      "loss": 3.702418804168701,
      "step": 940
    },
    {
      "epoch": 0.15182316876411744,
      "grad_norm": 1.4097989797592163,
      "learning_rate": 0.0005,
      "loss": 3.490683078765869,
      "step": 941
    },
    {
      "epoch": 0.15198451113262343,
      "grad_norm": 1.5152442455291748,
      "learning_rate": 0.0005,
      "loss": 3.457589626312256,
      "step": 942
    },
    {
      "epoch": 0.15214585350112939,
      "grad_norm": 1.5662288665771484,
      "learning_rate": 0.0005,
      "loss": 3.5072624683380127,
      "step": 943
    },
    {
      "epoch": 0.15230719586963537,
      "grad_norm": 1.487629771232605,
      "learning_rate": 0.0005,
      "loss": 3.5859978199005127,
      "step": 944
    },
    {
      "epoch": 0.15246853823814133,
      "grad_norm": 2.1752309799194336,
      "learning_rate": 0.0005,
      "loss": 3.4865312576293945,
      "step": 945
    },
    {
      "epoch": 0.1526298806066473,
      "grad_norm": 1.441523790359497,
      "learning_rate": 0.0005,
      "loss": 3.753878355026245,
      "step": 946
    },
    {
      "epoch": 0.15279122297515327,
      "grad_norm": 1.5011526346206665,
      "learning_rate": 0.0005,
      "loss": 2.980473518371582,
      "step": 947
    },
    {
      "epoch": 0.15295256534365925,
      "grad_norm": 3.142535924911499,
      "learning_rate": 0.0005,
      "loss": 3.59281325340271,
      "step": 948
    },
    {
      "epoch": 0.1531139077121652,
      "grad_norm": 1.517052412033081,
      "learning_rate": 0.0005,
      "loss": 3.08837890625,
      "step": 949
    },
    {
      "epoch": 0.1532752500806712,
      "grad_norm": 1.2462271451950073,
      "learning_rate": 0.0005,
      "loss": 3.837411880493164,
      "step": 950
    },
    {
      "epoch": 0.15343659244917715,
      "grad_norm": 5.1802167892456055,
      "learning_rate": 0.0005,
      "loss": 3.5071215629577637,
      "step": 951
    },
    {
      "epoch": 0.15359793481768313,
      "grad_norm": 4.706263065338135,
      "learning_rate": 0.0005,
      "loss": 3.4145865440368652,
      "step": 952
    },
    {
      "epoch": 0.1537592771861891,
      "grad_norm": 1.4776742458343506,
      "learning_rate": 0.0005,
      "loss": 3.636805295944214,
      "step": 953
    },
    {
      "epoch": 0.15392061955469508,
      "grad_norm": 1.5422662496566772,
      "learning_rate": 0.0005,
      "loss": 3.4464566707611084,
      "step": 954
    },
    {
      "epoch": 0.15408196192320103,
      "grad_norm": 4.018926620483398,
      "learning_rate": 0.0005,
      "loss": 4.034543991088867,
      "step": 955
    },
    {
      "epoch": 0.154243304291707,
      "grad_norm": 1.4676035642623901,
      "learning_rate": 0.0005,
      "loss": 3.477749824523926,
      "step": 956
    },
    {
      "epoch": 0.15440464666021297,
      "grad_norm": 1.4513214826583862,
      "learning_rate": 0.0005,
      "loss": 3.4334897994995117,
      "step": 957
    },
    {
      "epoch": 0.15456598902871893,
      "grad_norm": 1.846684217453003,
      "learning_rate": 0.0005,
      "loss": 3.4196622371673584,
      "step": 958
    },
    {
      "epoch": 0.15472733139722492,
      "grad_norm": 1.5684551000595093,
      "learning_rate": 0.0005,
      "loss": 3.2817068099975586,
      "step": 959
    },
    {
      "epoch": 0.15488867376573087,
      "grad_norm": 1.880596399307251,
      "learning_rate": 0.0005,
      "loss": 3.7830026149749756,
      "step": 960
    },
    {
      "epoch": 0.15505001613423686,
      "grad_norm": 1.3032448291778564,
      "learning_rate": 0.0005,
      "loss": 3.433438777923584,
      "step": 961
    },
    {
      "epoch": 0.1552113585027428,
      "grad_norm": 1.2846095561981201,
      "learning_rate": 0.0005,
      "loss": 3.520292282104492,
      "step": 962
    },
    {
      "epoch": 0.1553727008712488,
      "grad_norm": 1.4457677602767944,
      "learning_rate": 0.0005,
      "loss": 3.526456356048584,
      "step": 963
    },
    {
      "epoch": 0.15553404323975475,
      "grad_norm": 1.7629698514938354,
      "learning_rate": 0.0005,
      "loss": 3.277785062789917,
      "step": 964
    },
    {
      "epoch": 0.15569538560826074,
      "grad_norm": 1.645277738571167,
      "learning_rate": 0.0005,
      "loss": 3.435235023498535,
      "step": 965
    },
    {
      "epoch": 0.1558567279767667,
      "grad_norm": 1.3726584911346436,
      "learning_rate": 0.0005,
      "loss": 3.5051321983337402,
      "step": 966
    },
    {
      "epoch": 0.15601807034527268,
      "grad_norm": 1.7316443920135498,
      "learning_rate": 0.0005,
      "loss": 3.1042537689208984,
      "step": 967
    },
    {
      "epoch": 0.15617941271377864,
      "grad_norm": 1.5993821620941162,
      "learning_rate": 0.0005,
      "loss": 4.055567264556885,
      "step": 968
    },
    {
      "epoch": 0.15634075508228462,
      "grad_norm": 1.6762806177139282,
      "learning_rate": 0.0005,
      "loss": 3.4722769260406494,
      "step": 969
    },
    {
      "epoch": 0.15650209745079058,
      "grad_norm": 1.2544318437576294,
      "learning_rate": 0.0005,
      "loss": 3.5360825061798096,
      "step": 970
    },
    {
      "epoch": 0.15666343981929653,
      "grad_norm": 1.4434998035430908,
      "learning_rate": 0.0005,
      "loss": 3.167495012283325,
      "step": 971
    },
    {
      "epoch": 0.15682478218780252,
      "grad_norm": 1.312690019607544,
      "learning_rate": 0.0005,
      "loss": 3.835216522216797,
      "step": 972
    },
    {
      "epoch": 0.15698612455630848,
      "grad_norm": 2.0168354511260986,
      "learning_rate": 0.0005,
      "loss": 3.4454822540283203,
      "step": 973
    },
    {
      "epoch": 0.15714746692481446,
      "grad_norm": 1.893154263496399,
      "learning_rate": 0.0005,
      "loss": 3.758387804031372,
      "step": 974
    },
    {
      "epoch": 0.15730880929332042,
      "grad_norm": 1.7745466232299805,
      "learning_rate": 0.0005,
      "loss": 3.5217936038970947,
      "step": 975
    },
    {
      "epoch": 0.1574701516618264,
      "grad_norm": 1.4710280895233154,
      "learning_rate": 0.0005,
      "loss": 3.801091194152832,
      "step": 976
    },
    {
      "epoch": 0.15763149403033236,
      "grad_norm": 1.3201106786727905,
      "learning_rate": 0.0005,
      "loss": 3.63175106048584,
      "step": 977
    },
    {
      "epoch": 0.15779283639883834,
      "grad_norm": 1.4380452632904053,
      "learning_rate": 0.0005,
      "loss": 3.265134334564209,
      "step": 978
    },
    {
      "epoch": 0.1579541787673443,
      "grad_norm": 1.4422982931137085,
      "learning_rate": 0.0005,
      "loss": 2.9795331954956055,
      "step": 979
    },
    {
      "epoch": 0.15811552113585028,
      "grad_norm": 1.42436683177948,
      "learning_rate": 0.0005,
      "loss": 3.3184549808502197,
      "step": 980
    },
    {
      "epoch": 0.15827686350435624,
      "grad_norm": 1.8848508596420288,
      "learning_rate": 0.0005,
      "loss": 3.713099479675293,
      "step": 981
    },
    {
      "epoch": 0.15843820587286223,
      "grad_norm": 5.019906520843506,
      "learning_rate": 0.0005,
      "loss": 3.4153048992156982,
      "step": 982
    },
    {
      "epoch": 0.15859954824136818,
      "grad_norm": 2.3197085857391357,
      "learning_rate": 0.0005,
      "loss": 3.39668607711792,
      "step": 983
    },
    {
      "epoch": 0.15876089060987417,
      "grad_norm": 1.8292871713638306,
      "learning_rate": 0.0005,
      "loss": 2.9972448348999023,
      "step": 984
    },
    {
      "epoch": 0.15892223297838012,
      "grad_norm": 2.025080442428589,
      "learning_rate": 0.0005,
      "loss": 3.3644821643829346,
      "step": 985
    },
    {
      "epoch": 0.15908357534688608,
      "grad_norm": 1.683134913444519,
      "learning_rate": 0.0005,
      "loss": 3.4546027183532715,
      "step": 986
    },
    {
      "epoch": 0.15924491771539206,
      "grad_norm": 1.6760954856872559,
      "learning_rate": 0.0005,
      "loss": 3.277340888977051,
      "step": 987
    },
    {
      "epoch": 0.15940626008389802,
      "grad_norm": 1.3169841766357422,
      "learning_rate": 0.0005,
      "loss": 3.3744049072265625,
      "step": 988
    },
    {
      "epoch": 0.159567602452404,
      "grad_norm": 1.3928868770599365,
      "learning_rate": 0.0005,
      "loss": 3.8871073722839355,
      "step": 989
    },
    {
      "epoch": 0.15972894482090996,
      "grad_norm": 1.6644752025604248,
      "learning_rate": 0.0005,
      "loss": 3.6169509887695312,
      "step": 990
    },
    {
      "epoch": 0.15989028718941595,
      "grad_norm": 1.829301118850708,
      "learning_rate": 0.0005,
      "loss": 3.534182548522949,
      "step": 991
    },
    {
      "epoch": 0.1600516295579219,
      "grad_norm": 1.7453938722610474,
      "learning_rate": 0.0005,
      "loss": 3.246546745300293,
      "step": 992
    },
    {
      "epoch": 0.1602129719264279,
      "grad_norm": 1.629512071609497,
      "learning_rate": 0.0005,
      "loss": 3.4506301879882812,
      "step": 993
    },
    {
      "epoch": 0.16037431429493385,
      "grad_norm": 1.3530992269515991,
      "learning_rate": 0.0005,
      "loss": 3.6934256553649902,
      "step": 994
    },
    {
      "epoch": 0.16053565666343983,
      "grad_norm": 1.380924940109253,
      "learning_rate": 0.0005,
      "loss": 3.8660316467285156,
      "step": 995
    },
    {
      "epoch": 0.1606969990319458,
      "grad_norm": 1.5875873565673828,
      "learning_rate": 0.0005,
      "loss": 3.7842934131622314,
      "step": 996
    },
    {
      "epoch": 0.16085834140045177,
      "grad_norm": 1.314467191696167,
      "learning_rate": 0.0005,
      "loss": 3.396366596221924,
      "step": 997
    },
    {
      "epoch": 0.16101968376895773,
      "grad_norm": 1.7313719987869263,
      "learning_rate": 0.0005,
      "loss": 3.1294808387756348,
      "step": 998
    },
    {
      "epoch": 0.16118102613746368,
      "grad_norm": 1.4627434015274048,
      "learning_rate": 0.0005,
      "loss": 3.702561855316162,
      "step": 999
    },
    {
      "epoch": 0.16134236850596967,
      "grad_norm": 1.3650438785552979,
      "learning_rate": 0.0005,
      "loss": 3.679373264312744,
      "step": 1000
    },
    {
      "epoch": 0.16150371087447563,
      "grad_norm": 1.3172519207000732,
      "learning_rate": 0.0005,
      "loss": 3.4626307487487793,
      "step": 1001
    },
    {
      "epoch": 0.1616650532429816,
      "grad_norm": 1.4563027620315552,
      "learning_rate": 0.0005,
      "loss": 3.2040462493896484,
      "step": 1002
    },
    {
      "epoch": 0.16182639561148757,
      "grad_norm": 1.2230184078216553,
      "learning_rate": 0.0005,
      "loss": 3.0450806617736816,
      "step": 1003
    },
    {
      "epoch": 0.16198773797999355,
      "grad_norm": 1.315061092376709,
      "learning_rate": 0.0005,
      "loss": 3.249149799346924,
      "step": 1004
    },
    {
      "epoch": 0.1621490803484995,
      "grad_norm": 1.5761613845825195,
      "learning_rate": 0.0005,
      "loss": 3.3633522987365723,
      "step": 1005
    },
    {
      "epoch": 0.1623104227170055,
      "grad_norm": 1.9986296892166138,
      "learning_rate": 0.0005,
      "loss": 3.226944923400879,
      "step": 1006
    },
    {
      "epoch": 0.16247176508551145,
      "grad_norm": 1.6243953704833984,
      "learning_rate": 0.0005,
      "loss": 2.985332489013672,
      "step": 1007
    },
    {
      "epoch": 0.16263310745401743,
      "grad_norm": 1.583056926727295,
      "learning_rate": 0.0005,
      "loss": 3.793983221054077,
      "step": 1008
    },
    {
      "epoch": 0.1627944498225234,
      "grad_norm": 2.015859842300415,
      "learning_rate": 0.0005,
      "loss": 3.472769260406494,
      "step": 1009
    },
    {
      "epoch": 0.16295579219102937,
      "grad_norm": 1.6550801992416382,
      "learning_rate": 0.0005,
      "loss": 3.18782901763916,
      "step": 1010
    },
    {
      "epoch": 0.16311713455953533,
      "grad_norm": 2.0841002464294434,
      "learning_rate": 0.0005,
      "loss": 3.5523667335510254,
      "step": 1011
    },
    {
      "epoch": 0.16327847692804132,
      "grad_norm": 2.0053040981292725,
      "learning_rate": 0.0005,
      "loss": 3.286858081817627,
      "step": 1012
    },
    {
      "epoch": 0.16343981929654727,
      "grad_norm": 1.8609944581985474,
      "learning_rate": 0.0005,
      "loss": 3.4670894145965576,
      "step": 1013
    },
    {
      "epoch": 0.16360116166505323,
      "grad_norm": 1.6774260997772217,
      "learning_rate": 0.0005,
      "loss": 3.4629175662994385,
      "step": 1014
    },
    {
      "epoch": 0.16376250403355921,
      "grad_norm": 1.8282699584960938,
      "learning_rate": 0.0005,
      "loss": 3.3781418800354004,
      "step": 1015
    },
    {
      "epoch": 0.16392384640206517,
      "grad_norm": 1.4510159492492676,
      "learning_rate": 0.0005,
      "loss": 3.334105968475342,
      "step": 1016
    },
    {
      "epoch": 0.16408518877057116,
      "grad_norm": 1.5453253984451294,
      "learning_rate": 0.0005,
      "loss": 3.352839231491089,
      "step": 1017
    },
    {
      "epoch": 0.1642465311390771,
      "grad_norm": 1.5542526245117188,
      "learning_rate": 0.0005,
      "loss": 3.1499557495117188,
      "step": 1018
    },
    {
      "epoch": 0.1644078735075831,
      "grad_norm": 1.3556009531021118,
      "learning_rate": 0.0005,
      "loss": 3.543549060821533,
      "step": 1019
    },
    {
      "epoch": 0.16456921587608905,
      "grad_norm": 1.5948736667633057,
      "learning_rate": 0.0005,
      "loss": 3.8867249488830566,
      "step": 1020
    },
    {
      "epoch": 0.16473055824459504,
      "grad_norm": 1.353843092918396,
      "learning_rate": 0.0005,
      "loss": 3.2580862045288086,
      "step": 1021
    },
    {
      "epoch": 0.164891900613101,
      "grad_norm": 2.08532977104187,
      "learning_rate": 0.0005,
      "loss": 2.986912488937378,
      "step": 1022
    },
    {
      "epoch": 0.16505324298160698,
      "grad_norm": 1.418404221534729,
      "learning_rate": 0.0005,
      "loss": 3.5830941200256348,
      "step": 1023
    },
    {
      "epoch": 0.16521458535011294,
      "grad_norm": 1.5340937376022339,
      "learning_rate": 0.0005,
      "loss": 3.355778217315674,
      "step": 1024
    },
    {
      "epoch": 0.16537592771861892,
      "grad_norm": 5.275727272033691,
      "learning_rate": 0.0005,
      "loss": 3.656609058380127,
      "step": 1025
    },
    {
      "epoch": 0.16553727008712488,
      "grad_norm": 1.4793697595596313,
      "learning_rate": 0.0005,
      "loss": 3.4674038887023926,
      "step": 1026
    },
    {
      "epoch": 0.16569861245563086,
      "grad_norm": 1.6704833507537842,
      "learning_rate": 0.0005,
      "loss": 3.1608896255493164,
      "step": 1027
    },
    {
      "epoch": 0.16585995482413682,
      "grad_norm": 1.5258033275604248,
      "learning_rate": 0.0005,
      "loss": 3.8679962158203125,
      "step": 1028
    },
    {
      "epoch": 0.16602129719264277,
      "grad_norm": 1.4159893989562988,
      "learning_rate": 0.0005,
      "loss": 3.1694717407226562,
      "step": 1029
    },
    {
      "epoch": 0.16618263956114876,
      "grad_norm": 1.6453522443771362,
      "learning_rate": 0.0005,
      "loss": 3.72320556640625,
      "step": 1030
    },
    {
      "epoch": 0.16634398192965472,
      "grad_norm": 1.6521408557891846,
      "learning_rate": 0.0005,
      "loss": 3.5877773761749268,
      "step": 1031
    },
    {
      "epoch": 0.1665053242981607,
      "grad_norm": 1.2480117082595825,
      "learning_rate": 0.0005,
      "loss": 3.6028976440429688,
      "step": 1032
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 3.029247760772705,
      "learning_rate": 0.0005,
      "loss": 3.201857805252075,
      "step": 1033
    },
    {
      "epoch": 0.16682800903517264,
      "grad_norm": 1.3577995300292969,
      "learning_rate": 0.0005,
      "loss": 3.743372917175293,
      "step": 1034
    },
    {
      "epoch": 0.1669893514036786,
      "grad_norm": 1.680625081062317,
      "learning_rate": 0.0005,
      "loss": 3.1599998474121094,
      "step": 1035
    },
    {
      "epoch": 0.16715069377218458,
      "grad_norm": 1.6168800592422485,
      "learning_rate": 0.0005,
      "loss": 3.5870249271392822,
      "step": 1036
    },
    {
      "epoch": 0.16731203614069054,
      "grad_norm": 1.5445683002471924,
      "learning_rate": 0.0005,
      "loss": 3.1790595054626465,
      "step": 1037
    },
    {
      "epoch": 0.16747337850919652,
      "grad_norm": 1.7072439193725586,
      "learning_rate": 0.0005,
      "loss": 3.9868032932281494,
      "step": 1038
    },
    {
      "epoch": 0.16763472087770248,
      "grad_norm": 1.790108323097229,
      "learning_rate": 0.0005,
      "loss": 3.4819188117980957,
      "step": 1039
    },
    {
      "epoch": 0.16779606324620847,
      "grad_norm": 1.9772247076034546,
      "learning_rate": 0.0005,
      "loss": 3.72056245803833,
      "step": 1040
    },
    {
      "epoch": 0.16795740561471442,
      "grad_norm": 1.6606727838516235,
      "learning_rate": 0.0005,
      "loss": 3.222121477127075,
      "step": 1041
    },
    {
      "epoch": 0.1681187479832204,
      "grad_norm": 1.4662407636642456,
      "learning_rate": 0.0005,
      "loss": 3.335338830947876,
      "step": 1042
    },
    {
      "epoch": 0.16828009035172636,
      "grad_norm": 1.3152977228164673,
      "learning_rate": 0.0005,
      "loss": 3.2783968448638916,
      "step": 1043
    },
    {
      "epoch": 0.16844143272023232,
      "grad_norm": 1.5094579458236694,
      "learning_rate": 0.0005,
      "loss": 2.940613269805908,
      "step": 1044
    },
    {
      "epoch": 0.1686027750887383,
      "grad_norm": 1.687294363975525,
      "learning_rate": 0.0005,
      "loss": 3.683750867843628,
      "step": 1045
    },
    {
      "epoch": 0.16876411745724426,
      "grad_norm": 1.4172444343566895,
      "learning_rate": 0.0005,
      "loss": 3.324422597885132,
      "step": 1046
    },
    {
      "epoch": 0.16892545982575025,
      "grad_norm": 1.5201691389083862,
      "learning_rate": 0.0005,
      "loss": 3.3266801834106445,
      "step": 1047
    },
    {
      "epoch": 0.1690868021942562,
      "grad_norm": 1.390182614326477,
      "learning_rate": 0.0005,
      "loss": 3.702974557876587,
      "step": 1048
    },
    {
      "epoch": 0.1692481445627622,
      "grad_norm": 1.6757354736328125,
      "learning_rate": 0.0005,
      "loss": 4.048693656921387,
      "step": 1049
    },
    {
      "epoch": 0.16940948693126814,
      "grad_norm": 1.4374892711639404,
      "learning_rate": 0.0005,
      "loss": 3.327867031097412,
      "step": 1050
    },
    {
      "epoch": 0.16957082929977413,
      "grad_norm": 1.7176018953323364,
      "learning_rate": 0.0005,
      "loss": 3.003253698348999,
      "step": 1051
    },
    {
      "epoch": 0.16973217166828009,
      "grad_norm": 1.8300288915634155,
      "learning_rate": 0.0005,
      "loss": 3.5669984817504883,
      "step": 1052
    },
    {
      "epoch": 0.16989351403678607,
      "grad_norm": 1.728633999824524,
      "learning_rate": 0.0005,
      "loss": 4.192070484161377,
      "step": 1053
    },
    {
      "epoch": 0.17005485640529203,
      "grad_norm": 1.710809588432312,
      "learning_rate": 0.0005,
      "loss": 3.2241649627685547,
      "step": 1054
    },
    {
      "epoch": 0.170216198773798,
      "grad_norm": 1.4422967433929443,
      "learning_rate": 0.0005,
      "loss": 3.4503989219665527,
      "step": 1055
    },
    {
      "epoch": 0.17037754114230397,
      "grad_norm": 1.6663556098937988,
      "learning_rate": 0.0005,
      "loss": 3.8640618324279785,
      "step": 1056
    },
    {
      "epoch": 0.17053888351080995,
      "grad_norm": 1.5389951467514038,
      "learning_rate": 0.0005,
      "loss": 3.4503517150878906,
      "step": 1057
    },
    {
      "epoch": 0.1707002258793159,
      "grad_norm": 1.478868007659912,
      "learning_rate": 0.0005,
      "loss": 3.5576083660125732,
      "step": 1058
    },
    {
      "epoch": 0.17086156824782187,
      "grad_norm": 1.767317295074463,
      "learning_rate": 0.0005,
      "loss": 3.221419095993042,
      "step": 1059
    },
    {
      "epoch": 0.17102291061632785,
      "grad_norm": 1.6354708671569824,
      "learning_rate": 0.0005,
      "loss": 3.679783582687378,
      "step": 1060
    },
    {
      "epoch": 0.1711842529848338,
      "grad_norm": 1.471105933189392,
      "learning_rate": 0.0005,
      "loss": 3.435887336730957,
      "step": 1061
    },
    {
      "epoch": 0.1713455953533398,
      "grad_norm": 1.312441110610962,
      "learning_rate": 0.0005,
      "loss": 3.6000890731811523,
      "step": 1062
    },
    {
      "epoch": 0.17150693772184575,
      "grad_norm": 1.4927682876586914,
      "learning_rate": 0.0005,
      "loss": 3.6478443145751953,
      "step": 1063
    },
    {
      "epoch": 0.17166828009035173,
      "grad_norm": 1.7386655807495117,
      "learning_rate": 0.0005,
      "loss": 3.0454437732696533,
      "step": 1064
    },
    {
      "epoch": 0.1718296224588577,
      "grad_norm": 1.148919939994812,
      "learning_rate": 0.0005,
      "loss": 3.615166187286377,
      "step": 1065
    },
    {
      "epoch": 0.17199096482736367,
      "grad_norm": 2.0515193939208984,
      "learning_rate": 0.0005,
      "loss": 3.247175693511963,
      "step": 1066
    },
    {
      "epoch": 0.17215230719586963,
      "grad_norm": 1.247416377067566,
      "learning_rate": 0.0005,
      "loss": 3.4178450107574463,
      "step": 1067
    },
    {
      "epoch": 0.17231364956437561,
      "grad_norm": 1.5754410028457642,
      "learning_rate": 0.0005,
      "loss": 2.9394664764404297,
      "step": 1068
    },
    {
      "epoch": 0.17247499193288157,
      "grad_norm": 1.4697198867797852,
      "learning_rate": 0.0005,
      "loss": 3.486159086227417,
      "step": 1069
    },
    {
      "epoch": 0.17263633430138756,
      "grad_norm": 7.693874835968018,
      "learning_rate": 0.0005,
      "loss": 3.304206609725952,
      "step": 1070
    },
    {
      "epoch": 0.1727976766698935,
      "grad_norm": 1.4402050971984863,
      "learning_rate": 0.0005,
      "loss": 3.625030040740967,
      "step": 1071
    },
    {
      "epoch": 0.1729590190383995,
      "grad_norm": 1.2728826999664307,
      "learning_rate": 0.0005,
      "loss": 3.5926766395568848,
      "step": 1072
    },
    {
      "epoch": 0.17312036140690545,
      "grad_norm": 1.3197442293167114,
      "learning_rate": 0.0005,
      "loss": 3.3790059089660645,
      "step": 1073
    },
    {
      "epoch": 0.1732817037754114,
      "grad_norm": 1.3868862390518188,
      "learning_rate": 0.0005,
      "loss": 3.6719629764556885,
      "step": 1074
    },
    {
      "epoch": 0.1734430461439174,
      "grad_norm": 1.6307554244995117,
      "learning_rate": 0.0005,
      "loss": 3.341008186340332,
      "step": 1075
    },
    {
      "epoch": 0.17360438851242335,
      "grad_norm": 1.8073911666870117,
      "learning_rate": 0.0005,
      "loss": 3.583996295928955,
      "step": 1076
    },
    {
      "epoch": 0.17376573088092934,
      "grad_norm": 2.0327706336975098,
      "learning_rate": 0.0005,
      "loss": 3.6246323585510254,
      "step": 1077
    },
    {
      "epoch": 0.1739270732494353,
      "grad_norm": 2.4541919231414795,
      "learning_rate": 0.0005,
      "loss": 3.211897850036621,
      "step": 1078
    },
    {
      "epoch": 0.17408841561794128,
      "grad_norm": 1.8760188817977905,
      "learning_rate": 0.0005,
      "loss": 3.548492908477783,
      "step": 1079
    },
    {
      "epoch": 0.17424975798644723,
      "grad_norm": 1.4473953247070312,
      "learning_rate": 0.0005,
      "loss": 3.1265158653259277,
      "step": 1080
    },
    {
      "epoch": 0.17441110035495322,
      "grad_norm": 1.5357568264007568,
      "learning_rate": 0.0005,
      "loss": 3.213207721710205,
      "step": 1081
    },
    {
      "epoch": 0.17457244272345918,
      "grad_norm": 1.6159693002700806,
      "learning_rate": 0.0005,
      "loss": 3.1948444843292236,
      "step": 1082
    },
    {
      "epoch": 0.17473378509196516,
      "grad_norm": 1.2243324518203735,
      "learning_rate": 0.0005,
      "loss": 3.234656572341919,
      "step": 1083
    },
    {
      "epoch": 0.17489512746047112,
      "grad_norm": 1.7358818054199219,
      "learning_rate": 0.0005,
      "loss": 3.3359761238098145,
      "step": 1084
    },
    {
      "epoch": 0.1750564698289771,
      "grad_norm": 1.3549562692642212,
      "learning_rate": 0.0005,
      "loss": 3.8626151084899902,
      "step": 1085
    },
    {
      "epoch": 0.17521781219748306,
      "grad_norm": 1.3268153667449951,
      "learning_rate": 0.0005,
      "loss": 3.5838840007781982,
      "step": 1086
    },
    {
      "epoch": 0.17537915456598901,
      "grad_norm": 1.8619263172149658,
      "learning_rate": 0.0005,
      "loss": 3.6752023696899414,
      "step": 1087
    },
    {
      "epoch": 0.175540496934495,
      "grad_norm": 1.3367191553115845,
      "learning_rate": 0.0005,
      "loss": 3.4148287773132324,
      "step": 1088
    },
    {
      "epoch": 0.17570183930300096,
      "grad_norm": 1.2303630113601685,
      "learning_rate": 0.0005,
      "loss": 3.621647596359253,
      "step": 1089
    },
    {
      "epoch": 0.17586318167150694,
      "grad_norm": 1.4025585651397705,
      "learning_rate": 0.0005,
      "loss": 3.3710813522338867,
      "step": 1090
    },
    {
      "epoch": 0.1760245240400129,
      "grad_norm": 1.3625519275665283,
      "learning_rate": 0.0005,
      "loss": 3.256371259689331,
      "step": 1091
    },
    {
      "epoch": 0.17618586640851888,
      "grad_norm": 1.4983922243118286,
      "learning_rate": 0.0005,
      "loss": 3.0891594886779785,
      "step": 1092
    },
    {
      "epoch": 0.17634720877702484,
      "grad_norm": 1.3808057308197021,
      "learning_rate": 0.0005,
      "loss": 3.60524845123291,
      "step": 1093
    },
    {
      "epoch": 0.17650855114553082,
      "grad_norm": 2.2209391593933105,
      "learning_rate": 0.0005,
      "loss": 3.795259952545166,
      "step": 1094
    },
    {
      "epoch": 0.17666989351403678,
      "grad_norm": 2.140918016433716,
      "learning_rate": 0.0005,
      "loss": 3.694790840148926,
      "step": 1095
    },
    {
      "epoch": 0.17683123588254276,
      "grad_norm": 1.5289558172225952,
      "learning_rate": 0.0005,
      "loss": 3.500901699066162,
      "step": 1096
    },
    {
      "epoch": 0.17699257825104872,
      "grad_norm": 1.640810489654541,
      "learning_rate": 0.0005,
      "loss": 3.2795093059539795,
      "step": 1097
    },
    {
      "epoch": 0.1771539206195547,
      "grad_norm": 1.8776100873947144,
      "learning_rate": 0.0005,
      "loss": 3.503514528274536,
      "step": 1098
    },
    {
      "epoch": 0.17731526298806066,
      "grad_norm": 1.9800257682800293,
      "learning_rate": 0.0005,
      "loss": 3.361422300338745,
      "step": 1099
    },
    {
      "epoch": 0.17747660535656665,
      "grad_norm": 1.4252591133117676,
      "learning_rate": 0.0005,
      "loss": 3.8917713165283203,
      "step": 1100
    },
    {
      "epoch": 0.1776379477250726,
      "grad_norm": 1.3642007112503052,
      "learning_rate": 0.0005,
      "loss": 3.5740180015563965,
      "step": 1101
    },
    {
      "epoch": 0.17779929009357856,
      "grad_norm": 1.5108851194381714,
      "learning_rate": 0.0005,
      "loss": 3.510035514831543,
      "step": 1102
    },
    {
      "epoch": 0.17796063246208454,
      "grad_norm": 1.377938151359558,
      "learning_rate": 0.0005,
      "loss": 3.5879287719726562,
      "step": 1103
    },
    {
      "epoch": 0.1781219748305905,
      "grad_norm": 1.5333082675933838,
      "learning_rate": 0.0005,
      "loss": 2.6687350273132324,
      "step": 1104
    },
    {
      "epoch": 0.17828331719909649,
      "grad_norm": 1.7040499448776245,
      "learning_rate": 0.0005,
      "loss": 3.201301097869873,
      "step": 1105
    },
    {
      "epoch": 0.17844465956760244,
      "grad_norm": 1.5491331815719604,
      "learning_rate": 0.0005,
      "loss": 3.7520532608032227,
      "step": 1106
    },
    {
      "epoch": 0.17860600193610843,
      "grad_norm": 1.8534948825836182,
      "learning_rate": 0.0005,
      "loss": 3.615110397338867,
      "step": 1107
    },
    {
      "epoch": 0.17876734430461438,
      "grad_norm": 1.3667815923690796,
      "learning_rate": 0.0005,
      "loss": 3.102832555770874,
      "step": 1108
    },
    {
      "epoch": 0.17892868667312037,
      "grad_norm": 1.5451066493988037,
      "learning_rate": 0.0005,
      "loss": 3.250577211380005,
      "step": 1109
    },
    {
      "epoch": 0.17909002904162633,
      "grad_norm": 1.1882050037384033,
      "learning_rate": 0.0005,
      "loss": 3.339308977127075,
      "step": 1110
    },
    {
      "epoch": 0.1792513714101323,
      "grad_norm": 1.7834583520889282,
      "learning_rate": 0.0005,
      "loss": 3.2004752159118652,
      "step": 1111
    },
    {
      "epoch": 0.17941271377863827,
      "grad_norm": 1.61403489112854,
      "learning_rate": 0.0005,
      "loss": 3.6312286853790283,
      "step": 1112
    },
    {
      "epoch": 0.17957405614714425,
      "grad_norm": 1.5212866067886353,
      "learning_rate": 0.0005,
      "loss": 3.7098231315612793,
      "step": 1113
    },
    {
      "epoch": 0.1797353985156502,
      "grad_norm": 2.514091730117798,
      "learning_rate": 0.0005,
      "loss": 3.3970277309417725,
      "step": 1114
    },
    {
      "epoch": 0.1798967408841562,
      "grad_norm": 1.896327257156372,
      "learning_rate": 0.0005,
      "loss": 3.554124355316162,
      "step": 1115
    },
    {
      "epoch": 0.18005808325266215,
      "grad_norm": 3.195127010345459,
      "learning_rate": 0.0005,
      "loss": 3.710808038711548,
      "step": 1116
    },
    {
      "epoch": 0.1802194256211681,
      "grad_norm": 2.115048408508301,
      "learning_rate": 0.0005,
      "loss": 4.183918476104736,
      "step": 1117
    },
    {
      "epoch": 0.1803807679896741,
      "grad_norm": 1.3856277465820312,
      "learning_rate": 0.0005,
      "loss": 3.6681740283966064,
      "step": 1118
    },
    {
      "epoch": 0.18054211035818005,
      "grad_norm": 1.289044737815857,
      "learning_rate": 0.0005,
      "loss": 3.487999677658081,
      "step": 1119
    },
    {
      "epoch": 0.18070345272668603,
      "grad_norm": 1.5158277750015259,
      "learning_rate": 0.0005,
      "loss": 3.6686272621154785,
      "step": 1120
    },
    {
      "epoch": 0.180864795095192,
      "grad_norm": 1.58295476436615,
      "learning_rate": 0.0005,
      "loss": 3.2345540523529053,
      "step": 1121
    },
    {
      "epoch": 0.18102613746369797,
      "grad_norm": 1.5972776412963867,
      "learning_rate": 0.0005,
      "loss": 3.13674259185791,
      "step": 1122
    },
    {
      "epoch": 0.18118747983220393,
      "grad_norm": 1.7590560913085938,
      "learning_rate": 0.0005,
      "loss": 3.548859119415283,
      "step": 1123
    },
    {
      "epoch": 0.1813488222007099,
      "grad_norm": 2.8922674655914307,
      "learning_rate": 0.0005,
      "loss": 3.049187183380127,
      "step": 1124
    },
    {
      "epoch": 0.18151016456921587,
      "grad_norm": 2.259303092956543,
      "learning_rate": 0.0005,
      "loss": 3.5610716342926025,
      "step": 1125
    },
    {
      "epoch": 0.18167150693772185,
      "grad_norm": 1.3471400737762451,
      "learning_rate": 0.0005,
      "loss": 3.0957906246185303,
      "step": 1126
    },
    {
      "epoch": 0.1818328493062278,
      "grad_norm": 1.4482152462005615,
      "learning_rate": 0.0005,
      "loss": 2.6895220279693604,
      "step": 1127
    },
    {
      "epoch": 0.1819941916747338,
      "grad_norm": 1.2522006034851074,
      "learning_rate": 0.0005,
      "loss": 3.680912971496582,
      "step": 1128
    },
    {
      "epoch": 0.18215553404323975,
      "grad_norm": 1.9214448928833008,
      "learning_rate": 0.0005,
      "loss": 3.567924976348877,
      "step": 1129
    },
    {
      "epoch": 0.18231687641174574,
      "grad_norm": 1.4277902841567993,
      "learning_rate": 0.0005,
      "loss": 3.885589838027954,
      "step": 1130
    },
    {
      "epoch": 0.1824782187802517,
      "grad_norm": 1.5767422914505005,
      "learning_rate": 0.0005,
      "loss": 3.3263041973114014,
      "step": 1131
    },
    {
      "epoch": 0.18263956114875765,
      "grad_norm": 1.5730009078979492,
      "learning_rate": 0.0005,
      "loss": 3.5558128356933594,
      "step": 1132
    },
    {
      "epoch": 0.18280090351726364,
      "grad_norm": 1.1740294694900513,
      "learning_rate": 0.0005,
      "loss": 3.9902498722076416,
      "step": 1133
    },
    {
      "epoch": 0.1829622458857696,
      "grad_norm": 1.3526417016983032,
      "learning_rate": 0.0005,
      "loss": 3.666027545928955,
      "step": 1134
    },
    {
      "epoch": 0.18312358825427558,
      "grad_norm": 1.2932697534561157,
      "learning_rate": 0.0005,
      "loss": 3.4100537300109863,
      "step": 1135
    },
    {
      "epoch": 0.18328493062278153,
      "grad_norm": 1.495993971824646,
      "learning_rate": 0.0005,
      "loss": 2.9528770446777344,
      "step": 1136
    },
    {
      "epoch": 0.18344627299128752,
      "grad_norm": 1.2452890872955322,
      "learning_rate": 0.0005,
      "loss": 3.0294389724731445,
      "step": 1137
    },
    {
      "epoch": 0.18360761535979347,
      "grad_norm": 1.3803703784942627,
      "learning_rate": 0.0005,
      "loss": 3.4391098022460938,
      "step": 1138
    },
    {
      "epoch": 0.18376895772829946,
      "grad_norm": 1.396335244178772,
      "learning_rate": 0.0005,
      "loss": 3.5709052085876465,
      "step": 1139
    },
    {
      "epoch": 0.18393030009680542,
      "grad_norm": 1.2722753286361694,
      "learning_rate": 0.0005,
      "loss": 3.4232327938079834,
      "step": 1140
    },
    {
      "epoch": 0.1840916424653114,
      "grad_norm": 1.5015695095062256,
      "learning_rate": 0.0005,
      "loss": 3.5281591415405273,
      "step": 1141
    },
    {
      "epoch": 0.18425298483381736,
      "grad_norm": 1.4802627563476562,
      "learning_rate": 0.0005,
      "loss": 3.7794790267944336,
      "step": 1142
    },
    {
      "epoch": 0.18441432720232334,
      "grad_norm": 1.6341785192489624,
      "learning_rate": 0.0005,
      "loss": 3.4359803199768066,
      "step": 1143
    },
    {
      "epoch": 0.1845756695708293,
      "grad_norm": 1.412300944328308,
      "learning_rate": 0.0005,
      "loss": 3.288750648498535,
      "step": 1144
    },
    {
      "epoch": 0.18473701193933528,
      "grad_norm": 1.7271381616592407,
      "learning_rate": 0.0005,
      "loss": 3.591161012649536,
      "step": 1145
    },
    {
      "epoch": 0.18489835430784124,
      "grad_norm": 1.5939985513687134,
      "learning_rate": 0.0005,
      "loss": 3.3773016929626465,
      "step": 1146
    },
    {
      "epoch": 0.1850596966763472,
      "grad_norm": 1.6246888637542725,
      "learning_rate": 0.0005,
      "loss": 3.287684440612793,
      "step": 1147
    },
    {
      "epoch": 0.18522103904485318,
      "grad_norm": 1.797770380973816,
      "learning_rate": 0.0005,
      "loss": 3.3752989768981934,
      "step": 1148
    },
    {
      "epoch": 0.18538238141335914,
      "grad_norm": 1.4926823377609253,
      "learning_rate": 0.0005,
      "loss": 3.5659167766571045,
      "step": 1149
    },
    {
      "epoch": 0.18554372378186512,
      "grad_norm": 1.4612767696380615,
      "learning_rate": 0.0005,
      "loss": 3.156111240386963,
      "step": 1150
    },
    {
      "epoch": 0.18570506615037108,
      "grad_norm": 1.3367953300476074,
      "learning_rate": 0.0005,
      "loss": 3.362626314163208,
      "step": 1151
    },
    {
      "epoch": 0.18586640851887706,
      "grad_norm": 1.5253289937973022,
      "learning_rate": 0.0005,
      "loss": 3.729149341583252,
      "step": 1152
    },
    {
      "epoch": 0.18602775088738302,
      "grad_norm": 2.131052017211914,
      "learning_rate": 0.0005,
      "loss": 3.317887306213379,
      "step": 1153
    },
    {
      "epoch": 0.186189093255889,
      "grad_norm": 1.491410493850708,
      "learning_rate": 0.0005,
      "loss": 3.517404556274414,
      "step": 1154
    },
    {
      "epoch": 0.18635043562439496,
      "grad_norm": 1.2425233125686646,
      "learning_rate": 0.0005,
      "loss": 3.5237863063812256,
      "step": 1155
    },
    {
      "epoch": 0.18651177799290095,
      "grad_norm": 1.5345606803894043,
      "learning_rate": 0.0005,
      "loss": 3.163459539413452,
      "step": 1156
    },
    {
      "epoch": 0.1866731203614069,
      "grad_norm": 1.4929392337799072,
      "learning_rate": 0.0005,
      "loss": 3.136995553970337,
      "step": 1157
    },
    {
      "epoch": 0.1868344627299129,
      "grad_norm": 1.3519033193588257,
      "learning_rate": 0.0005,
      "loss": 3.4237232208251953,
      "step": 1158
    },
    {
      "epoch": 0.18699580509841884,
      "grad_norm": 1.5213418006896973,
      "learning_rate": 0.0005,
      "loss": 3.4509854316711426,
      "step": 1159
    },
    {
      "epoch": 0.18715714746692483,
      "grad_norm": 1.7469913959503174,
      "learning_rate": 0.0005,
      "loss": 3.006502628326416,
      "step": 1160
    },
    {
      "epoch": 0.18731848983543078,
      "grad_norm": 1.3247501850128174,
      "learning_rate": 0.0005,
      "loss": 2.7189974784851074,
      "step": 1161
    },
    {
      "epoch": 0.18747983220393674,
      "grad_norm": 1.4597324132919312,
      "learning_rate": 0.0005,
      "loss": 3.404283046722412,
      "step": 1162
    },
    {
      "epoch": 0.18764117457244273,
      "grad_norm": 1.4198683500289917,
      "learning_rate": 0.0005,
      "loss": 3.3403608798980713,
      "step": 1163
    },
    {
      "epoch": 0.18780251694094868,
      "grad_norm": 2.237436294555664,
      "learning_rate": 0.0005,
      "loss": 3.392488956451416,
      "step": 1164
    },
    {
      "epoch": 0.18796385930945467,
      "grad_norm": 1.4838796854019165,
      "learning_rate": 0.0005,
      "loss": 3.048924684524536,
      "step": 1165
    },
    {
      "epoch": 0.18812520167796062,
      "grad_norm": 1.6439318656921387,
      "learning_rate": 0.0005,
      "loss": 3.72078800201416,
      "step": 1166
    },
    {
      "epoch": 0.1882865440464666,
      "grad_norm": 1.2605773210525513,
      "learning_rate": 0.0005,
      "loss": 3.2849459648132324,
      "step": 1167
    },
    {
      "epoch": 0.18844788641497257,
      "grad_norm": 1.9005483388900757,
      "learning_rate": 0.0005,
      "loss": 3.0839531421661377,
      "step": 1168
    },
    {
      "epoch": 0.18860922878347855,
      "grad_norm": 1.3531014919281006,
      "learning_rate": 0.0005,
      "loss": 3.3666367530822754,
      "step": 1169
    },
    {
      "epoch": 0.1887705711519845,
      "grad_norm": 1.8001304864883423,
      "learning_rate": 0.0005,
      "loss": 3.385270595550537,
      "step": 1170
    },
    {
      "epoch": 0.1889319135204905,
      "grad_norm": 1.8345674276351929,
      "learning_rate": 0.0005,
      "loss": 3.2349047660827637,
      "step": 1171
    },
    {
      "epoch": 0.18909325588899645,
      "grad_norm": 1.6694319248199463,
      "learning_rate": 0.0005,
      "loss": 2.8328003883361816,
      "step": 1172
    },
    {
      "epoch": 0.18925459825750243,
      "grad_norm": 1.321489930152893,
      "learning_rate": 0.0005,
      "loss": 3.2820076942443848,
      "step": 1173
    },
    {
      "epoch": 0.1894159406260084,
      "grad_norm": 1.6890617609024048,
      "learning_rate": 0.0005,
      "loss": 3.28718900680542,
      "step": 1174
    },
    {
      "epoch": 0.18957728299451435,
      "grad_norm": 1.4347622394561768,
      "learning_rate": 0.0005,
      "loss": 3.2182626724243164,
      "step": 1175
    },
    {
      "epoch": 0.18973862536302033,
      "grad_norm": 3.567615032196045,
      "learning_rate": 0.0005,
      "loss": 3.577493667602539,
      "step": 1176
    },
    {
      "epoch": 0.1898999677315263,
      "grad_norm": 1.404252052307129,
      "learning_rate": 0.0005,
      "loss": 3.538600444793701,
      "step": 1177
    },
    {
      "epoch": 0.19006131010003227,
      "grad_norm": 1.3723360300064087,
      "learning_rate": 0.0005,
      "loss": 3.20896315574646,
      "step": 1178
    },
    {
      "epoch": 0.19022265246853823,
      "grad_norm": 1.4242022037506104,
      "learning_rate": 0.0005,
      "loss": 3.357758045196533,
      "step": 1179
    },
    {
      "epoch": 0.1903839948370442,
      "grad_norm": 1.2564054727554321,
      "learning_rate": 0.0005,
      "loss": 3.2982969284057617,
      "step": 1180
    },
    {
      "epoch": 0.19054533720555017,
      "grad_norm": 1.3628102540969849,
      "learning_rate": 0.0005,
      "loss": 3.173430919647217,
      "step": 1181
    },
    {
      "epoch": 0.19070667957405615,
      "grad_norm": 1.9942412376403809,
      "learning_rate": 0.0005,
      "loss": 3.4910669326782227,
      "step": 1182
    },
    {
      "epoch": 0.1908680219425621,
      "grad_norm": 1.3904420137405396,
      "learning_rate": 0.0005,
      "loss": 3.4319639205932617,
      "step": 1183
    },
    {
      "epoch": 0.1910293643110681,
      "grad_norm": 1.1890426874160767,
      "learning_rate": 0.0005,
      "loss": 3.367300510406494,
      "step": 1184
    },
    {
      "epoch": 0.19119070667957405,
      "grad_norm": 1.5045664310455322,
      "learning_rate": 0.0005,
      "loss": 3.115601062774658,
      "step": 1185
    },
    {
      "epoch": 0.19135204904808004,
      "grad_norm": 1.3279104232788086,
      "learning_rate": 0.0005,
      "loss": 3.4613256454467773,
      "step": 1186
    },
    {
      "epoch": 0.191513391416586,
      "grad_norm": 1.4112173318862915,
      "learning_rate": 0.0005,
      "loss": 3.0923171043395996,
      "step": 1187
    },
    {
      "epoch": 0.19167473378509198,
      "grad_norm": 2.220172166824341,
      "learning_rate": 0.0005,
      "loss": 3.6258046627044678,
      "step": 1188
    },
    {
      "epoch": 0.19183607615359793,
      "grad_norm": 1.6453125476837158,
      "learning_rate": 0.0005,
      "loss": 2.913100481033325,
      "step": 1189
    },
    {
      "epoch": 0.1919974185221039,
      "grad_norm": 1.5700929164886475,
      "learning_rate": 0.0005,
      "loss": 3.644486427307129,
      "step": 1190
    },
    {
      "epoch": 0.19215876089060988,
      "grad_norm": 1.3260059356689453,
      "learning_rate": 0.0005,
      "loss": 3.5104432106018066,
      "step": 1191
    },
    {
      "epoch": 0.19232010325911583,
      "grad_norm": 1.3710871934890747,
      "learning_rate": 0.0005,
      "loss": 3.364729642868042,
      "step": 1192
    },
    {
      "epoch": 0.19248144562762182,
      "grad_norm": 1.7722489833831787,
      "learning_rate": 0.0005,
      "loss": 3.2442421913146973,
      "step": 1193
    },
    {
      "epoch": 0.19264278799612777,
      "grad_norm": 1.4280465841293335,
      "learning_rate": 0.0005,
      "loss": 3.26546311378479,
      "step": 1194
    },
    {
      "epoch": 0.19280413036463376,
      "grad_norm": 1.9887065887451172,
      "learning_rate": 0.0005,
      "loss": 3.611560344696045,
      "step": 1195
    },
    {
      "epoch": 0.19296547273313971,
      "grad_norm": 1.689757227897644,
      "learning_rate": 0.0005,
      "loss": 3.710063934326172,
      "step": 1196
    },
    {
      "epoch": 0.1931268151016457,
      "grad_norm": 1.359716534614563,
      "learning_rate": 0.0005,
      "loss": 3.5886595249176025,
      "step": 1197
    },
    {
      "epoch": 0.19328815747015166,
      "grad_norm": 1.6765846014022827,
      "learning_rate": 0.0005,
      "loss": 3.488220691680908,
      "step": 1198
    },
    {
      "epoch": 0.19344949983865764,
      "grad_norm": 1.367083191871643,
      "learning_rate": 0.0005,
      "loss": 3.6274185180664062,
      "step": 1199
    },
    {
      "epoch": 0.1936108422071636,
      "grad_norm": 1.4583523273468018,
      "learning_rate": 0.0005,
      "loss": 3.492990493774414,
      "step": 1200
    },
    {
      "epoch": 0.19377218457566958,
      "grad_norm": 1.738855242729187,
      "learning_rate": 0.0005,
      "loss": 3.3642120361328125,
      "step": 1201
    },
    {
      "epoch": 0.19393352694417554,
      "grad_norm": 1.1961830854415894,
      "learning_rate": 0.0005,
      "loss": 2.689415693283081,
      "step": 1202
    },
    {
      "epoch": 0.19409486931268152,
      "grad_norm": 2.4630372524261475,
      "learning_rate": 0.0005,
      "loss": 3.464242458343506,
      "step": 1203
    },
    {
      "epoch": 0.19425621168118748,
      "grad_norm": 1.294940710067749,
      "learning_rate": 0.0005,
      "loss": 3.2834792137145996,
      "step": 1204
    },
    {
      "epoch": 0.19441755404969344,
      "grad_norm": 1.5283191204071045,
      "learning_rate": 0.0005,
      "loss": 3.1527397632598877,
      "step": 1205
    },
    {
      "epoch": 0.19457889641819942,
      "grad_norm": 1.3809964656829834,
      "learning_rate": 0.0005,
      "loss": 3.6390886306762695,
      "step": 1206
    },
    {
      "epoch": 0.19474023878670538,
      "grad_norm": 1.8008835315704346,
      "learning_rate": 0.0005,
      "loss": 3.612445831298828,
      "step": 1207
    },
    {
      "epoch": 0.19490158115521136,
      "grad_norm": 1.7513196468353271,
      "learning_rate": 0.0005,
      "loss": 3.515315055847168,
      "step": 1208
    },
    {
      "epoch": 0.19506292352371732,
      "grad_norm": 1.9432076215744019,
      "learning_rate": 0.0005,
      "loss": 3.5900020599365234,
      "step": 1209
    },
    {
      "epoch": 0.1952242658922233,
      "grad_norm": 1.4490911960601807,
      "learning_rate": 0.0005,
      "loss": 3.2824320793151855,
      "step": 1210
    },
    {
      "epoch": 0.19538560826072926,
      "grad_norm": 1.37918221950531,
      "learning_rate": 0.0005,
      "loss": 4.0615739822387695,
      "step": 1211
    },
    {
      "epoch": 0.19554695062923524,
      "grad_norm": 1.5453191995620728,
      "learning_rate": 0.0005,
      "loss": 3.292595386505127,
      "step": 1212
    },
    {
      "epoch": 0.1957082929977412,
      "grad_norm": 2.312279462814331,
      "learning_rate": 0.0005,
      "loss": 2.796891689300537,
      "step": 1213
    },
    {
      "epoch": 0.19586963536624719,
      "grad_norm": 1.4235353469848633,
      "learning_rate": 0.0005,
      "loss": 3.354851722717285,
      "step": 1214
    },
    {
      "epoch": 0.19603097773475314,
      "grad_norm": 1.3579100370407104,
      "learning_rate": 0.0005,
      "loss": 3.567014694213867,
      "step": 1215
    },
    {
      "epoch": 0.19619232010325913,
      "grad_norm": 1.3356722593307495,
      "learning_rate": 0.0005,
      "loss": 3.161695957183838,
      "step": 1216
    },
    {
      "epoch": 0.19635366247176508,
      "grad_norm": 1.45870041847229,
      "learning_rate": 0.0005,
      "loss": 3.8308892250061035,
      "step": 1217
    },
    {
      "epoch": 0.19651500484027107,
      "grad_norm": 3.0640878677368164,
      "learning_rate": 0.0005,
      "loss": 3.255476474761963,
      "step": 1218
    },
    {
      "epoch": 0.19667634720877702,
      "grad_norm": 1.5286750793457031,
      "learning_rate": 0.0005,
      "loss": 3.4036378860473633,
      "step": 1219
    },
    {
      "epoch": 0.19683768957728298,
      "grad_norm": 1.8052996397018433,
      "learning_rate": 0.0005,
      "loss": 3.2690749168395996,
      "step": 1220
    },
    {
      "epoch": 0.19699903194578897,
      "grad_norm": 1.245418906211853,
      "learning_rate": 0.0005,
      "loss": 3.563668966293335,
      "step": 1221
    },
    {
      "epoch": 0.19716037431429492,
      "grad_norm": 1.580676555633545,
      "learning_rate": 0.0005,
      "loss": 3.5202605724334717,
      "step": 1222
    },
    {
      "epoch": 0.1973217166828009,
      "grad_norm": 1.783717155456543,
      "learning_rate": 0.0005,
      "loss": 3.3320720195770264,
      "step": 1223
    },
    {
      "epoch": 0.19748305905130686,
      "grad_norm": 1.7993993759155273,
      "learning_rate": 0.0005,
      "loss": 3.2473254203796387,
      "step": 1224
    },
    {
      "epoch": 0.19764440141981285,
      "grad_norm": 1.8955678939819336,
      "learning_rate": 0.0005,
      "loss": 3.502549409866333,
      "step": 1225
    },
    {
      "epoch": 0.1978057437883188,
      "grad_norm": 1.4811326265335083,
      "learning_rate": 0.0005,
      "loss": 3.3414697647094727,
      "step": 1226
    },
    {
      "epoch": 0.1979670861568248,
      "grad_norm": 1.6741176843643188,
      "learning_rate": 0.0005,
      "loss": 3.550067186355591,
      "step": 1227
    },
    {
      "epoch": 0.19812842852533075,
      "grad_norm": 1.3829326629638672,
      "learning_rate": 0.0005,
      "loss": 3.415571689605713,
      "step": 1228
    },
    {
      "epoch": 0.19828977089383673,
      "grad_norm": 1.5550212860107422,
      "learning_rate": 0.0005,
      "loss": 3.164062738418579,
      "step": 1229
    },
    {
      "epoch": 0.1984511132623427,
      "grad_norm": 1.705682635307312,
      "learning_rate": 0.0005,
      "loss": 3.612593173980713,
      "step": 1230
    },
    {
      "epoch": 0.19861245563084867,
      "grad_norm": 1.3803033828735352,
      "learning_rate": 0.0005,
      "loss": 3.114748001098633,
      "step": 1231
    },
    {
      "epoch": 0.19877379799935463,
      "grad_norm": 1.3318060636520386,
      "learning_rate": 0.0005,
      "loss": 3.5824382305145264,
      "step": 1232
    },
    {
      "epoch": 0.1989351403678606,
      "grad_norm": 1.2917876243591309,
      "learning_rate": 0.0005,
      "loss": 3.4232685565948486,
      "step": 1233
    },
    {
      "epoch": 0.19909648273636657,
      "grad_norm": 1.5482836961746216,
      "learning_rate": 0.0005,
      "loss": 3.6247124671936035,
      "step": 1234
    },
    {
      "epoch": 0.19925782510487253,
      "grad_norm": 1.2882307767868042,
      "learning_rate": 0.0005,
      "loss": 3.2661619186401367,
      "step": 1235
    },
    {
      "epoch": 0.1994191674733785,
      "grad_norm": 1.5887383222579956,
      "learning_rate": 0.0005,
      "loss": 3.1018519401550293,
      "step": 1236
    },
    {
      "epoch": 0.19958050984188447,
      "grad_norm": 1.6878678798675537,
      "learning_rate": 0.0005,
      "loss": 3.062251567840576,
      "step": 1237
    },
    {
      "epoch": 0.19974185221039045,
      "grad_norm": 1.3605506420135498,
      "learning_rate": 0.0005,
      "loss": 3.6110451221466064,
      "step": 1238
    },
    {
      "epoch": 0.1999031945788964,
      "grad_norm": 1.5609475374221802,
      "learning_rate": 0.0005,
      "loss": 3.558605432510376,
      "step": 1239
    },
    {
      "epoch": 0.2000645369474024,
      "grad_norm": 1.3045730590820312,
      "learning_rate": 0.0005,
      "loss": 3.273226022720337,
      "step": 1240
    },
    {
      "epoch": 0.20022587931590835,
      "grad_norm": 1.2598376274108887,
      "learning_rate": 0.0005,
      "loss": 3.792297124862671,
      "step": 1241
    },
    {
      "epoch": 0.20038722168441434,
      "grad_norm": 1.797809362411499,
      "learning_rate": 0.0005,
      "loss": 3.060579776763916,
      "step": 1242
    },
    {
      "epoch": 0.2005485640529203,
      "grad_norm": 1.5041255950927734,
      "learning_rate": 0.0005,
      "loss": 3.5171685218811035,
      "step": 1243
    },
    {
      "epoch": 0.20070990642142628,
      "grad_norm": 1.7037465572357178,
      "learning_rate": 0.0005,
      "loss": 3.57312273979187,
      "step": 1244
    },
    {
      "epoch": 0.20087124878993223,
      "grad_norm": 1.4010252952575684,
      "learning_rate": 0.0005,
      "loss": 3.579796552658081,
      "step": 1245
    },
    {
      "epoch": 0.20103259115843822,
      "grad_norm": 1.2783128023147583,
      "learning_rate": 0.0005,
      "loss": 3.1125221252441406,
      "step": 1246
    },
    {
      "epoch": 0.20119393352694417,
      "grad_norm": 1.4929826259613037,
      "learning_rate": 0.0005,
      "loss": 3.60792875289917,
      "step": 1247
    },
    {
      "epoch": 0.20135527589545016,
      "grad_norm": 1.1050077676773071,
      "learning_rate": 0.0005,
      "loss": 3.146650791168213,
      "step": 1248
    },
    {
      "epoch": 0.20151661826395612,
      "grad_norm": 1.224799394607544,
      "learning_rate": 0.0005,
      "loss": 3.4169833660125732,
      "step": 1249
    },
    {
      "epoch": 0.20167796063246207,
      "grad_norm": 1.279704213142395,
      "learning_rate": 0.0005,
      "loss": 3.405984878540039,
      "step": 1250
    },
    {
      "epoch": 0.20183930300096806,
      "grad_norm": 1.1171066761016846,
      "learning_rate": 0.0005,
      "loss": 3.224693536758423,
      "step": 1251
    },
    {
      "epoch": 0.202000645369474,
      "grad_norm": 1.2324926853179932,
      "learning_rate": 0.0005,
      "loss": 3.15446138381958,
      "step": 1252
    },
    {
      "epoch": 0.20216198773798,
      "grad_norm": 1.5479930639266968,
      "learning_rate": 0.0005,
      "loss": 3.4092679023742676,
      "step": 1253
    },
    {
      "epoch": 0.20232333010648595,
      "grad_norm": 1.5488427877426147,
      "learning_rate": 0.0005,
      "loss": 3.429424285888672,
      "step": 1254
    },
    {
      "epoch": 0.20248467247499194,
      "grad_norm": 1.5518611669540405,
      "learning_rate": 0.0005,
      "loss": 3.36434006690979,
      "step": 1255
    },
    {
      "epoch": 0.2026460148434979,
      "grad_norm": 1.6923985481262207,
      "learning_rate": 0.0005,
      "loss": 3.545681953430176,
      "step": 1256
    },
    {
      "epoch": 0.20280735721200388,
      "grad_norm": 1.6283818483352661,
      "learning_rate": 0.0005,
      "loss": 3.400412082672119,
      "step": 1257
    },
    {
      "epoch": 0.20296869958050984,
      "grad_norm": 1.1984890699386597,
      "learning_rate": 0.0005,
      "loss": 3.830540895462036,
      "step": 1258
    },
    {
      "epoch": 0.20313004194901582,
      "grad_norm": 1.9830329418182373,
      "learning_rate": 0.0005,
      "loss": 3.1450536251068115,
      "step": 1259
    },
    {
      "epoch": 0.20329138431752178,
      "grad_norm": 1.3529812097549438,
      "learning_rate": 0.0005,
      "loss": 3.5902023315429688,
      "step": 1260
    },
    {
      "epoch": 0.20345272668602776,
      "grad_norm": 1.2355464696884155,
      "learning_rate": 0.0005,
      "loss": 3.473071575164795,
      "step": 1261
    },
    {
      "epoch": 0.20361406905453372,
      "grad_norm": 1.2492282390594482,
      "learning_rate": 0.0005,
      "loss": 3.5999951362609863,
      "step": 1262
    },
    {
      "epoch": 0.2037754114230397,
      "grad_norm": 1.7630183696746826,
      "learning_rate": 0.0005,
      "loss": 3.18977689743042,
      "step": 1263
    },
    {
      "epoch": 0.20393675379154566,
      "grad_norm": 1.7358052730560303,
      "learning_rate": 0.0005,
      "loss": 3.1354029178619385,
      "step": 1264
    },
    {
      "epoch": 0.20409809616005162,
      "grad_norm": 1.724717617034912,
      "learning_rate": 0.0005,
      "loss": 3.4450197219848633,
      "step": 1265
    },
    {
      "epoch": 0.2042594385285576,
      "grad_norm": 1.6236096620559692,
      "learning_rate": 0.0005,
      "loss": 3.0978071689605713,
      "step": 1266
    },
    {
      "epoch": 0.20442078089706356,
      "grad_norm": 1.5150796175003052,
      "learning_rate": 0.0005,
      "loss": 3.360694646835327,
      "step": 1267
    },
    {
      "epoch": 0.20458212326556954,
      "grad_norm": 1.5252103805541992,
      "learning_rate": 0.0005,
      "loss": 2.960894823074341,
      "step": 1268
    },
    {
      "epoch": 0.2047434656340755,
      "grad_norm": 1.2289553880691528,
      "learning_rate": 0.0005,
      "loss": 3.7284812927246094,
      "step": 1269
    },
    {
      "epoch": 0.20490480800258148,
      "grad_norm": 1.5813864469528198,
      "learning_rate": 0.0005,
      "loss": 3.008594036102295,
      "step": 1270
    },
    {
      "epoch": 0.20506615037108744,
      "grad_norm": 1.9989118576049805,
      "learning_rate": 0.0005,
      "loss": 3.1733055114746094,
      "step": 1271
    },
    {
      "epoch": 0.20522749273959343,
      "grad_norm": 1.348676085472107,
      "learning_rate": 0.0005,
      "loss": 3.2254016399383545,
      "step": 1272
    },
    {
      "epoch": 0.20538883510809938,
      "grad_norm": 1.303061842918396,
      "learning_rate": 0.0005,
      "loss": 3.563703775405884,
      "step": 1273
    },
    {
      "epoch": 0.20555017747660537,
      "grad_norm": 1.3208316564559937,
      "learning_rate": 0.0005,
      "loss": 3.2998578548431396,
      "step": 1274
    },
    {
      "epoch": 0.20571151984511132,
      "grad_norm": 1.3861973285675049,
      "learning_rate": 0.0005,
      "loss": 3.034788131713867,
      "step": 1275
    },
    {
      "epoch": 0.2058728622136173,
      "grad_norm": 1.5575860738754272,
      "learning_rate": 0.0005,
      "loss": 3.4143118858337402,
      "step": 1276
    },
    {
      "epoch": 0.20603420458212326,
      "grad_norm": 1.3598648309707642,
      "learning_rate": 0.0005,
      "loss": 3.3954427242279053,
      "step": 1277
    },
    {
      "epoch": 0.20619554695062922,
      "grad_norm": 1.1379501819610596,
      "learning_rate": 0.0005,
      "loss": 3.61873197555542,
      "step": 1278
    },
    {
      "epoch": 0.2063568893191352,
      "grad_norm": 1.5199607610702515,
      "learning_rate": 0.0005,
      "loss": 3.1337084770202637,
      "step": 1279
    },
    {
      "epoch": 0.20651823168764116,
      "grad_norm": 2.2184183597564697,
      "learning_rate": 0.0005,
      "loss": 3.487215995788574,
      "step": 1280
    },
    {
      "epoch": 0.20667957405614715,
      "grad_norm": 1.3960191011428833,
      "learning_rate": 0.0005,
      "loss": 3.4527270793914795,
      "step": 1281
    },
    {
      "epoch": 0.2068409164246531,
      "grad_norm": 1.1075679063796997,
      "learning_rate": 0.0005,
      "loss": 3.4819176197052,
      "step": 1282
    },
    {
      "epoch": 0.2070022587931591,
      "grad_norm": 1.2508809566497803,
      "learning_rate": 0.0005,
      "loss": 3.1561427116394043,
      "step": 1283
    },
    {
      "epoch": 0.20716360116166505,
      "grad_norm": 1.9716461896896362,
      "learning_rate": 0.0005,
      "loss": 3.2601590156555176,
      "step": 1284
    },
    {
      "epoch": 0.20732494353017103,
      "grad_norm": 1.3705044984817505,
      "learning_rate": 0.0005,
      "loss": 3.4946930408477783,
      "step": 1285
    },
    {
      "epoch": 0.207486285898677,
      "grad_norm": 2.1344220638275146,
      "learning_rate": 0.0005,
      "loss": 3.1404550075531006,
      "step": 1286
    },
    {
      "epoch": 0.20764762826718297,
      "grad_norm": 1.625028133392334,
      "learning_rate": 0.0005,
      "loss": 3.437394142150879,
      "step": 1287
    },
    {
      "epoch": 0.20780897063568893,
      "grad_norm": 1.7724497318267822,
      "learning_rate": 0.0005,
      "loss": 3.9370551109313965,
      "step": 1288
    },
    {
      "epoch": 0.2079703130041949,
      "grad_norm": 1.747758388519287,
      "learning_rate": 0.0005,
      "loss": 3.285104751586914,
      "step": 1289
    },
    {
      "epoch": 0.20813165537270087,
      "grad_norm": 2.078756093978882,
      "learning_rate": 0.0005,
      "loss": 3.421161651611328,
      "step": 1290
    },
    {
      "epoch": 0.20829299774120685,
      "grad_norm": 1.6350436210632324,
      "learning_rate": 0.0005,
      "loss": 3.2776360511779785,
      "step": 1291
    },
    {
      "epoch": 0.2084543401097128,
      "grad_norm": 1.739914894104004,
      "learning_rate": 0.0005,
      "loss": 3.3026137351989746,
      "step": 1292
    },
    {
      "epoch": 0.20861568247821877,
      "grad_norm": 1.4220658540725708,
      "learning_rate": 0.0005,
      "loss": 3.27169132232666,
      "step": 1293
    },
    {
      "epoch": 0.20877702484672475,
      "grad_norm": 1.9781392812728882,
      "learning_rate": 0.0005,
      "loss": 3.4289464950561523,
      "step": 1294
    },
    {
      "epoch": 0.2089383672152307,
      "grad_norm": 1.5922200679779053,
      "learning_rate": 0.0005,
      "loss": 3.406639814376831,
      "step": 1295
    },
    {
      "epoch": 0.2090997095837367,
      "grad_norm": 1.2873175144195557,
      "learning_rate": 0.0005,
      "loss": 3.1434450149536133,
      "step": 1296
    },
    {
      "epoch": 0.20926105195224265,
      "grad_norm": 1.7758984565734863,
      "learning_rate": 0.0005,
      "loss": 3.384260654449463,
      "step": 1297
    },
    {
      "epoch": 0.20942239432074863,
      "grad_norm": 1.2524131536483765,
      "learning_rate": 0.0005,
      "loss": 3.44458270072937,
      "step": 1298
    },
    {
      "epoch": 0.2095837366892546,
      "grad_norm": 1.7130112648010254,
      "learning_rate": 0.0005,
      "loss": 2.7879598140716553,
      "step": 1299
    },
    {
      "epoch": 0.20974507905776058,
      "grad_norm": 1.4304330348968506,
      "learning_rate": 0.0005,
      "loss": 3.337681770324707,
      "step": 1300
    },
    {
      "epoch": 0.20990642142626653,
      "grad_norm": 1.707759141921997,
      "learning_rate": 0.0005,
      "loss": 4.074671745300293,
      "step": 1301
    },
    {
      "epoch": 0.21006776379477252,
      "grad_norm": 1.2921181917190552,
      "learning_rate": 0.0005,
      "loss": 3.3771350383758545,
      "step": 1302
    },
    {
      "epoch": 0.21022910616327847,
      "grad_norm": 1.453909993171692,
      "learning_rate": 0.0005,
      "loss": 3.79317045211792,
      "step": 1303
    },
    {
      "epoch": 0.21039044853178446,
      "grad_norm": 1.4531275033950806,
      "learning_rate": 0.0005,
      "loss": 3.4417126178741455,
      "step": 1304
    },
    {
      "epoch": 0.21055179090029041,
      "grad_norm": 2.1200175285339355,
      "learning_rate": 0.0005,
      "loss": 3.3902857303619385,
      "step": 1305
    },
    {
      "epoch": 0.2107131332687964,
      "grad_norm": 1.908875584602356,
      "learning_rate": 0.0005,
      "loss": 3.5003628730773926,
      "step": 1306
    },
    {
      "epoch": 0.21087447563730236,
      "grad_norm": 1.6475284099578857,
      "learning_rate": 0.0005,
      "loss": 3.461798667907715,
      "step": 1307
    },
    {
      "epoch": 0.2110358180058083,
      "grad_norm": 1.6909599304199219,
      "learning_rate": 0.0005,
      "loss": 3.355648994445801,
      "step": 1308
    },
    {
      "epoch": 0.2111971603743143,
      "grad_norm": 1.1293801069259644,
      "learning_rate": 0.0005,
      "loss": 3.424459218978882,
      "step": 1309
    },
    {
      "epoch": 0.21135850274282025,
      "grad_norm": 1.208646297454834,
      "learning_rate": 0.0005,
      "loss": 3.5919387340545654,
      "step": 1310
    },
    {
      "epoch": 0.21151984511132624,
      "grad_norm": 1.476381540298462,
      "learning_rate": 0.0005,
      "loss": 3.260326862335205,
      "step": 1311
    },
    {
      "epoch": 0.2116811874798322,
      "grad_norm": 1.709276556968689,
      "learning_rate": 0.0005,
      "loss": 3.2899813652038574,
      "step": 1312
    },
    {
      "epoch": 0.21184252984833818,
      "grad_norm": 1.8443045616149902,
      "learning_rate": 0.0005,
      "loss": 3.0546793937683105,
      "step": 1313
    },
    {
      "epoch": 0.21200387221684414,
      "grad_norm": 1.7521775960922241,
      "learning_rate": 0.0005,
      "loss": 3.1615071296691895,
      "step": 1314
    },
    {
      "epoch": 0.21216521458535012,
      "grad_norm": 1.4425710439682007,
      "learning_rate": 0.0005,
      "loss": 3.236999750137329,
      "step": 1315
    },
    {
      "epoch": 0.21232655695385608,
      "grad_norm": 2.202556610107422,
      "learning_rate": 0.0005,
      "loss": 3.444796562194824,
      "step": 1316
    },
    {
      "epoch": 0.21248789932236206,
      "grad_norm": 2.0754921436309814,
      "learning_rate": 0.0005,
      "loss": 3.2639806270599365,
      "step": 1317
    },
    {
      "epoch": 0.21264924169086802,
      "grad_norm": 1.912126898765564,
      "learning_rate": 0.0005,
      "loss": 3.214205265045166,
      "step": 1318
    },
    {
      "epoch": 0.212810584059374,
      "grad_norm": 1.8969011306762695,
      "learning_rate": 0.0005,
      "loss": 3.340076446533203,
      "step": 1319
    },
    {
      "epoch": 0.21297192642787996,
      "grad_norm": 1.6880797147750854,
      "learning_rate": 0.0005,
      "loss": 3.5594773292541504,
      "step": 1320
    },
    {
      "epoch": 0.21313326879638594,
      "grad_norm": 1.5958276987075806,
      "learning_rate": 0.0005,
      "loss": 3.1018853187561035,
      "step": 1321
    },
    {
      "epoch": 0.2132946111648919,
      "grad_norm": 1.592896819114685,
      "learning_rate": 0.0005,
      "loss": 3.2597246170043945,
      "step": 1322
    },
    {
      "epoch": 0.21345595353339786,
      "grad_norm": 1.4888449907302856,
      "learning_rate": 0.0005,
      "loss": 3.4442644119262695,
      "step": 1323
    },
    {
      "epoch": 0.21361729590190384,
      "grad_norm": 1.4493335485458374,
      "learning_rate": 0.0005,
      "loss": 2.9280343055725098,
      "step": 1324
    },
    {
      "epoch": 0.2137786382704098,
      "grad_norm": 2.1003177165985107,
      "learning_rate": 0.0005,
      "loss": 3.2955150604248047,
      "step": 1325
    },
    {
      "epoch": 0.21393998063891578,
      "grad_norm": 1.3650448322296143,
      "learning_rate": 0.0005,
      "loss": 3.3795902729034424,
      "step": 1326
    },
    {
      "epoch": 0.21410132300742174,
      "grad_norm": 2.307969093322754,
      "learning_rate": 0.0005,
      "loss": 3.52213716506958,
      "step": 1327
    },
    {
      "epoch": 0.21426266537592772,
      "grad_norm": 1.5600459575653076,
      "learning_rate": 0.0005,
      "loss": 2.9142370223999023,
      "step": 1328
    },
    {
      "epoch": 0.21442400774443368,
      "grad_norm": 1.5253950357437134,
      "learning_rate": 0.0005,
      "loss": 3.446159601211548,
      "step": 1329
    },
    {
      "epoch": 0.21458535011293967,
      "grad_norm": 1.5270100831985474,
      "learning_rate": 0.0005,
      "loss": 3.2281408309936523,
      "step": 1330
    },
    {
      "epoch": 0.21474669248144562,
      "grad_norm": 1.3934580087661743,
      "learning_rate": 0.0005,
      "loss": 3.182157039642334,
      "step": 1331
    },
    {
      "epoch": 0.2149080348499516,
      "grad_norm": 1.2689803838729858,
      "learning_rate": 0.0005,
      "loss": 3.45965838432312,
      "step": 1332
    },
    {
      "epoch": 0.21506937721845756,
      "grad_norm": 1.3494845628738403,
      "learning_rate": 0.0005,
      "loss": 3.2922260761260986,
      "step": 1333
    },
    {
      "epoch": 0.21523071958696355,
      "grad_norm": 1.5299220085144043,
      "learning_rate": 0.0005,
      "loss": 3.0341668128967285,
      "step": 1334
    },
    {
      "epoch": 0.2153920619554695,
      "grad_norm": 2.0390777587890625,
      "learning_rate": 0.0005,
      "loss": 3.687213182449341,
      "step": 1335
    },
    {
      "epoch": 0.2155534043239755,
      "grad_norm": 1.3109655380249023,
      "learning_rate": 0.0005,
      "loss": 3.4686827659606934,
      "step": 1336
    },
    {
      "epoch": 0.21571474669248145,
      "grad_norm": 1.6463855504989624,
      "learning_rate": 0.0005,
      "loss": 3.594871997833252,
      "step": 1337
    },
    {
      "epoch": 0.2158760890609874,
      "grad_norm": 1.555261254310608,
      "learning_rate": 0.0005,
      "loss": 3.4444918632507324,
      "step": 1338
    },
    {
      "epoch": 0.2160374314294934,
      "grad_norm": 1.7355554103851318,
      "learning_rate": 0.0005,
      "loss": 3.1383042335510254,
      "step": 1339
    },
    {
      "epoch": 0.21619877379799934,
      "grad_norm": 2.2868542671203613,
      "learning_rate": 0.0005,
      "loss": 2.976884365081787,
      "step": 1340
    },
    {
      "epoch": 0.21636011616650533,
      "grad_norm": 1.467675805091858,
      "learning_rate": 0.0005,
      "loss": 2.972228527069092,
      "step": 1341
    },
    {
      "epoch": 0.21652145853501129,
      "grad_norm": 1.3930333852767944,
      "learning_rate": 0.0005,
      "loss": 3.2119009494781494,
      "step": 1342
    },
    {
      "epoch": 0.21668280090351727,
      "grad_norm": 1.2166067361831665,
      "learning_rate": 0.0005,
      "loss": 3.371392250061035,
      "step": 1343
    },
    {
      "epoch": 0.21684414327202323,
      "grad_norm": 1.336699366569519,
      "learning_rate": 0.0005,
      "loss": 3.2497096061706543,
      "step": 1344
    },
    {
      "epoch": 0.2170054856405292,
      "grad_norm": 1.343985915184021,
      "learning_rate": 0.0005,
      "loss": 3.014490842819214,
      "step": 1345
    },
    {
      "epoch": 0.21716682800903517,
      "grad_norm": 1.3971655368804932,
      "learning_rate": 0.0005,
      "loss": 3.1060566902160645,
      "step": 1346
    },
    {
      "epoch": 0.21732817037754115,
      "grad_norm": 1.1677148342132568,
      "learning_rate": 0.0005,
      "loss": 3.3256068229675293,
      "step": 1347
    },
    {
      "epoch": 0.2174895127460471,
      "grad_norm": 1.7012039422988892,
      "learning_rate": 0.0005,
      "loss": 3.3272287845611572,
      "step": 1348
    },
    {
      "epoch": 0.2176508551145531,
      "grad_norm": 1.2840014696121216,
      "learning_rate": 0.0005,
      "loss": 3.3154079914093018,
      "step": 1349
    },
    {
      "epoch": 0.21781219748305905,
      "grad_norm": 1.6180421113967896,
      "learning_rate": 0.0005,
      "loss": 3.343627452850342,
      "step": 1350
    },
    {
      "epoch": 0.21797353985156503,
      "grad_norm": 1.7273566722869873,
      "learning_rate": 0.0005,
      "loss": 3.1544907093048096,
      "step": 1351
    },
    {
      "epoch": 0.218134882220071,
      "grad_norm": 1.836459755897522,
      "learning_rate": 0.0005,
      "loss": 2.624182939529419,
      "step": 1352
    },
    {
      "epoch": 0.21829622458857695,
      "grad_norm": 1.8182722330093384,
      "learning_rate": 0.0005,
      "loss": 3.1355137825012207,
      "step": 1353
    },
    {
      "epoch": 0.21845756695708293,
      "grad_norm": 1.3388434648513794,
      "learning_rate": 0.0005,
      "loss": 3.8046529293060303,
      "step": 1354
    },
    {
      "epoch": 0.2186189093255889,
      "grad_norm": 1.4244011640548706,
      "learning_rate": 0.0005,
      "loss": 3.396315097808838,
      "step": 1355
    },
    {
      "epoch": 0.21878025169409487,
      "grad_norm": 1.3575325012207031,
      "learning_rate": 0.0005,
      "loss": 3.013068199157715,
      "step": 1356
    },
    {
      "epoch": 0.21894159406260083,
      "grad_norm": 1.6997175216674805,
      "learning_rate": 0.0005,
      "loss": 2.9402520656585693,
      "step": 1357
    },
    {
      "epoch": 0.21910293643110682,
      "grad_norm": 1.7344063520431519,
      "learning_rate": 0.0005,
      "loss": 3.7642195224761963,
      "step": 1358
    },
    {
      "epoch": 0.21926427879961277,
      "grad_norm": 1.432507872581482,
      "learning_rate": 0.0005,
      "loss": 2.8923754692077637,
      "step": 1359
    },
    {
      "epoch": 0.21942562116811876,
      "grad_norm": 1.405187964439392,
      "learning_rate": 0.0005,
      "loss": 3.0657479763031006,
      "step": 1360
    },
    {
      "epoch": 0.2195869635366247,
      "grad_norm": 1.285197377204895,
      "learning_rate": 0.0005,
      "loss": 3.6160759925842285,
      "step": 1361
    },
    {
      "epoch": 0.2197483059051307,
      "grad_norm": 1.6828148365020752,
      "learning_rate": 0.0005,
      "loss": 3.5200841426849365,
      "step": 1362
    },
    {
      "epoch": 0.21990964827363665,
      "grad_norm": 1.464192509651184,
      "learning_rate": 0.0005,
      "loss": 3.4536550045013428,
      "step": 1363
    },
    {
      "epoch": 0.22007099064214264,
      "grad_norm": 1.7671561241149902,
      "learning_rate": 0.0005,
      "loss": 2.5329527854919434,
      "step": 1364
    },
    {
      "epoch": 0.2202323330106486,
      "grad_norm": 1.231062889099121,
      "learning_rate": 0.0005,
      "loss": 3.4277291297912598,
      "step": 1365
    },
    {
      "epoch": 0.22039367537915455,
      "grad_norm": 1.4840476512908936,
      "learning_rate": 0.0005,
      "loss": 3.1997525691986084,
      "step": 1366
    },
    {
      "epoch": 0.22055501774766054,
      "grad_norm": 1.4235132932662964,
      "learning_rate": 0.0005,
      "loss": 3.2712955474853516,
      "step": 1367
    },
    {
      "epoch": 0.2207163601161665,
      "grad_norm": 4.516407012939453,
      "learning_rate": 0.0005,
      "loss": 3.4041125774383545,
      "step": 1368
    },
    {
      "epoch": 0.22087770248467248,
      "grad_norm": 1.499579906463623,
      "learning_rate": 0.0005,
      "loss": 3.7302651405334473,
      "step": 1369
    },
    {
      "epoch": 0.22103904485317843,
      "grad_norm": 1.2922184467315674,
      "learning_rate": 0.0005,
      "loss": 3.5534627437591553,
      "step": 1370
    },
    {
      "epoch": 0.22120038722168442,
      "grad_norm": 1.2455153465270996,
      "learning_rate": 0.0005,
      "loss": 3.1135025024414062,
      "step": 1371
    },
    {
      "epoch": 0.22136172959019038,
      "grad_norm": 1.7381795644760132,
      "learning_rate": 0.0005,
      "loss": 3.209228754043579,
      "step": 1372
    },
    {
      "epoch": 0.22152307195869636,
      "grad_norm": 1.4980778694152832,
      "learning_rate": 0.0005,
      "loss": 3.6612653732299805,
      "step": 1373
    },
    {
      "epoch": 0.22168441432720232,
      "grad_norm": 1.6269490718841553,
      "learning_rate": 0.0005,
      "loss": 3.036959171295166,
      "step": 1374
    },
    {
      "epoch": 0.2218457566957083,
      "grad_norm": 1.222651720046997,
      "learning_rate": 0.0005,
      "loss": 3.847292184829712,
      "step": 1375
    },
    {
      "epoch": 0.22200709906421426,
      "grad_norm": 1.2235585451126099,
      "learning_rate": 0.0005,
      "loss": 3.261967658996582,
      "step": 1376
    },
    {
      "epoch": 0.22216844143272024,
      "grad_norm": 1.27597177028656,
      "learning_rate": 0.0005,
      "loss": 2.794475793838501,
      "step": 1377
    },
    {
      "epoch": 0.2223297838012262,
      "grad_norm": 1.7160930633544922,
      "learning_rate": 0.0005,
      "loss": 3.614581823348999,
      "step": 1378
    },
    {
      "epoch": 0.22249112616973218,
      "grad_norm": 1.8946716785430908,
      "learning_rate": 0.0005,
      "loss": 3.247948169708252,
      "step": 1379
    },
    {
      "epoch": 0.22265246853823814,
      "grad_norm": 1.175439715385437,
      "learning_rate": 0.0005,
      "loss": 3.901627540588379,
      "step": 1380
    },
    {
      "epoch": 0.2228138109067441,
      "grad_norm": 1.7963333129882812,
      "learning_rate": 0.0005,
      "loss": 3.547520160675049,
      "step": 1381
    },
    {
      "epoch": 0.22297515327525008,
      "grad_norm": 1.5620269775390625,
      "learning_rate": 0.0005,
      "loss": 3.2786052227020264,
      "step": 1382
    },
    {
      "epoch": 0.22313649564375604,
      "grad_norm": 1.473816156387329,
      "learning_rate": 0.0005,
      "loss": 3.2171738147735596,
      "step": 1383
    },
    {
      "epoch": 0.22329783801226202,
      "grad_norm": 1.5430747270584106,
      "learning_rate": 0.0005,
      "loss": 3.296379566192627,
      "step": 1384
    },
    {
      "epoch": 0.22345918038076798,
      "grad_norm": 3.1370201110839844,
      "learning_rate": 0.0005,
      "loss": 3.298135280609131,
      "step": 1385
    },
    {
      "epoch": 0.22362052274927396,
      "grad_norm": 1.55245041847229,
      "learning_rate": 0.0005,
      "loss": 2.8352742195129395,
      "step": 1386
    },
    {
      "epoch": 0.22378186511777992,
      "grad_norm": 1.9635465145111084,
      "learning_rate": 0.0005,
      "loss": 3.2734522819519043,
      "step": 1387
    },
    {
      "epoch": 0.2239432074862859,
      "grad_norm": 1.7224321365356445,
      "learning_rate": 0.0005,
      "loss": 2.699103832244873,
      "step": 1388
    },
    {
      "epoch": 0.22410454985479186,
      "grad_norm": 1.4926546812057495,
      "learning_rate": 0.0005,
      "loss": 3.390611171722412,
      "step": 1389
    },
    {
      "epoch": 0.22426589222329785,
      "grad_norm": 1.6669706106185913,
      "learning_rate": 0.0005,
      "loss": 3.693054437637329,
      "step": 1390
    },
    {
      "epoch": 0.2244272345918038,
      "grad_norm": 1.292091727256775,
      "learning_rate": 0.0005,
      "loss": 3.363966226577759,
      "step": 1391
    },
    {
      "epoch": 0.2245885769603098,
      "grad_norm": 1.387843370437622,
      "learning_rate": 0.0005,
      "loss": 3.209176540374756,
      "step": 1392
    },
    {
      "epoch": 0.22474991932881574,
      "grad_norm": 1.7554208040237427,
      "learning_rate": 0.0005,
      "loss": 3.3614659309387207,
      "step": 1393
    },
    {
      "epoch": 0.22491126169732173,
      "grad_norm": 1.6256310939788818,
      "learning_rate": 0.0005,
      "loss": 3.666872978210449,
      "step": 1394
    },
    {
      "epoch": 0.22507260406582769,
      "grad_norm": 1.4347399473190308,
      "learning_rate": 0.0005,
      "loss": 3.0650861263275146,
      "step": 1395
    },
    {
      "epoch": 0.22523394643433364,
      "grad_norm": 1.687424659729004,
      "learning_rate": 0.0005,
      "loss": 3.5377397537231445,
      "step": 1396
    },
    {
      "epoch": 0.22539528880283963,
      "grad_norm": 1.4191049337387085,
      "learning_rate": 0.0005,
      "loss": 3.4494881629943848,
      "step": 1397
    },
    {
      "epoch": 0.22555663117134558,
      "grad_norm": 1.6494141817092896,
      "learning_rate": 0.0005,
      "loss": 3.2871856689453125,
      "step": 1398
    },
    {
      "epoch": 0.22571797353985157,
      "grad_norm": 1.3933930397033691,
      "learning_rate": 0.0005,
      "loss": 3.0761613845825195,
      "step": 1399
    },
    {
      "epoch": 0.22587931590835753,
      "grad_norm": 1.4005918502807617,
      "learning_rate": 0.0005,
      "loss": 3.411816120147705,
      "step": 1400
    },
    {
      "epoch": 0.2260406582768635,
      "grad_norm": 1.1981772184371948,
      "learning_rate": 0.0005,
      "loss": 2.8465006351470947,
      "step": 1401
    },
    {
      "epoch": 0.22620200064536947,
      "grad_norm": 1.785845160484314,
      "learning_rate": 0.0005,
      "loss": 3.3680458068847656,
      "step": 1402
    },
    {
      "epoch": 0.22636334301387545,
      "grad_norm": 1.5406097173690796,
      "learning_rate": 0.0005,
      "loss": 3.126621961593628,
      "step": 1403
    },
    {
      "epoch": 0.2265246853823814,
      "grad_norm": 1.2737159729003906,
      "learning_rate": 0.0005,
      "loss": 3.1526248455047607,
      "step": 1404
    },
    {
      "epoch": 0.2266860277508874,
      "grad_norm": 1.6483606100082397,
      "learning_rate": 0.0005,
      "loss": 3.123751640319824,
      "step": 1405
    },
    {
      "epoch": 0.22684737011939335,
      "grad_norm": 1.4289630651474,
      "learning_rate": 0.0005,
      "loss": 2.9533987045288086,
      "step": 1406
    },
    {
      "epoch": 0.22700871248789933,
      "grad_norm": 3.961014986038208,
      "learning_rate": 0.0005,
      "loss": 2.2932496070861816,
      "step": 1407
    },
    {
      "epoch": 0.2271700548564053,
      "grad_norm": 2.616243839263916,
      "learning_rate": 0.0005,
      "loss": 3.2312910556793213,
      "step": 1408
    },
    {
      "epoch": 0.22733139722491127,
      "grad_norm": 1.4847335815429688,
      "learning_rate": 0.0005,
      "loss": 3.462677478790283,
      "step": 1409
    },
    {
      "epoch": 0.22749273959341723,
      "grad_norm": 1.623164415359497,
      "learning_rate": 0.0005,
      "loss": 3.1358320713043213,
      "step": 1410
    },
    {
      "epoch": 0.2276540819619232,
      "grad_norm": 1.2061331272125244,
      "learning_rate": 0.0005,
      "loss": 2.8214962482452393,
      "step": 1411
    },
    {
      "epoch": 0.22781542433042917,
      "grad_norm": 1.2097461223602295,
      "learning_rate": 0.0005,
      "loss": 3.0746099948883057,
      "step": 1412
    },
    {
      "epoch": 0.22797676669893513,
      "grad_norm": 1.4275591373443604,
      "learning_rate": 0.0005,
      "loss": 3.4881272315979004,
      "step": 1413
    },
    {
      "epoch": 0.2281381090674411,
      "grad_norm": 1.3819137811660767,
      "learning_rate": 0.0005,
      "loss": 3.1113152503967285,
      "step": 1414
    },
    {
      "epoch": 0.22829945143594707,
      "grad_norm": 2.078063726425171,
      "learning_rate": 0.0005,
      "loss": 3.6307787895202637,
      "step": 1415
    },
    {
      "epoch": 0.22846079380445306,
      "grad_norm": 1.4773035049438477,
      "learning_rate": 0.0005,
      "loss": 3.1651573181152344,
      "step": 1416
    },
    {
      "epoch": 0.228622136172959,
      "grad_norm": 1.6112697124481201,
      "learning_rate": 0.0005,
      "loss": 3.3705084323883057,
      "step": 1417
    },
    {
      "epoch": 0.228783478541465,
      "grad_norm": 1.8768845796585083,
      "learning_rate": 0.0005,
      "loss": 3.3567442893981934,
      "step": 1418
    },
    {
      "epoch": 0.22894482090997095,
      "grad_norm": 1.2995880842208862,
      "learning_rate": 0.0005,
      "loss": 2.974152088165283,
      "step": 1419
    },
    {
      "epoch": 0.22910616327847694,
      "grad_norm": 2.018064498901367,
      "learning_rate": 0.0005,
      "loss": 3.282355308532715,
      "step": 1420
    },
    {
      "epoch": 0.2292675056469829,
      "grad_norm": 1.653626561164856,
      "learning_rate": 0.0005,
      "loss": 3.8745312690734863,
      "step": 1421
    },
    {
      "epoch": 0.22942884801548888,
      "grad_norm": 1.2884105443954468,
      "learning_rate": 0.0005,
      "loss": 3.4863696098327637,
      "step": 1422
    },
    {
      "epoch": 0.22959019038399484,
      "grad_norm": 1.263260006904602,
      "learning_rate": 0.0005,
      "loss": 3.418208599090576,
      "step": 1423
    },
    {
      "epoch": 0.22975153275250082,
      "grad_norm": 1.264724850654602,
      "learning_rate": 0.0005,
      "loss": 3.2224044799804688,
      "step": 1424
    },
    {
      "epoch": 0.22991287512100678,
      "grad_norm": 1.6598583459854126,
      "learning_rate": 0.0005,
      "loss": 3.3261983394622803,
      "step": 1425
    },
    {
      "epoch": 0.23007421748951273,
      "grad_norm": 1.3534965515136719,
      "learning_rate": 0.0005,
      "loss": 2.981999397277832,
      "step": 1426
    },
    {
      "epoch": 0.23023555985801872,
      "grad_norm": 1.6757491827011108,
      "learning_rate": 0.0005,
      "loss": 4.148753643035889,
      "step": 1427
    },
    {
      "epoch": 0.23039690222652467,
      "grad_norm": 1.1681970357894897,
      "learning_rate": 0.0005,
      "loss": 2.8894052505493164,
      "step": 1428
    },
    {
      "epoch": 0.23055824459503066,
      "grad_norm": 1.4931559562683105,
      "learning_rate": 0.0005,
      "loss": 3.603306293487549,
      "step": 1429
    },
    {
      "epoch": 0.23071958696353662,
      "grad_norm": 1.1680400371551514,
      "learning_rate": 0.0005,
      "loss": 2.838679552078247,
      "step": 1430
    },
    {
      "epoch": 0.2308809293320426,
      "grad_norm": 1.5202385187149048,
      "learning_rate": 0.0005,
      "loss": 3.3172447681427,
      "step": 1431
    },
    {
      "epoch": 0.23104227170054856,
      "grad_norm": 6.322871208190918,
      "learning_rate": 0.0005,
      "loss": 3.249058246612549,
      "step": 1432
    },
    {
      "epoch": 0.23120361406905454,
      "grad_norm": 1.5999687910079956,
      "learning_rate": 0.0005,
      "loss": 3.0203003883361816,
      "step": 1433
    },
    {
      "epoch": 0.2313649564375605,
      "grad_norm": 3.225579023361206,
      "learning_rate": 0.0005,
      "loss": 3.4149858951568604,
      "step": 1434
    },
    {
      "epoch": 0.23152629880606648,
      "grad_norm": 2.2307615280151367,
      "learning_rate": 0.0005,
      "loss": 3.3141374588012695,
      "step": 1435
    },
    {
      "epoch": 0.23168764117457244,
      "grad_norm": 2.0497353076934814,
      "learning_rate": 0.0005,
      "loss": 3.191169261932373,
      "step": 1436
    },
    {
      "epoch": 0.23184898354307842,
      "grad_norm": 1.7739815711975098,
      "learning_rate": 0.0005,
      "loss": 3.4031729698181152,
      "step": 1437
    },
    {
      "epoch": 0.23201032591158438,
      "grad_norm": 1.4860641956329346,
      "learning_rate": 0.0005,
      "loss": 2.777031421661377,
      "step": 1438
    },
    {
      "epoch": 0.23217166828009037,
      "grad_norm": 1.9041558504104614,
      "learning_rate": 0.0005,
      "loss": 3.0277504920959473,
      "step": 1439
    },
    {
      "epoch": 0.23233301064859632,
      "grad_norm": 1.2965319156646729,
      "learning_rate": 0.0005,
      "loss": 3.7270517349243164,
      "step": 1440
    },
    {
      "epoch": 0.23249435301710228,
      "grad_norm": 1.3892747163772583,
      "learning_rate": 0.0005,
      "loss": 3.6696736812591553,
      "step": 1441
    },
    {
      "epoch": 0.23265569538560826,
      "grad_norm": 1.4376471042633057,
      "learning_rate": 0.0005,
      "loss": 2.8701024055480957,
      "step": 1442
    },
    {
      "epoch": 0.23281703775411422,
      "grad_norm": 1.3895000219345093,
      "learning_rate": 0.0005,
      "loss": 3.2553231716156006,
      "step": 1443
    },
    {
      "epoch": 0.2329783801226202,
      "grad_norm": 1.330858588218689,
      "learning_rate": 0.0005,
      "loss": 3.685063362121582,
      "step": 1444
    },
    {
      "epoch": 0.23313972249112616,
      "grad_norm": 2.469485282897949,
      "learning_rate": 0.0005,
      "loss": 2.8219480514526367,
      "step": 1445
    },
    {
      "epoch": 0.23330106485963215,
      "grad_norm": 1.3868998289108276,
      "learning_rate": 0.0005,
      "loss": 3.16524600982666,
      "step": 1446
    },
    {
      "epoch": 0.2334624072281381,
      "grad_norm": 1.205271601676941,
      "learning_rate": 0.0005,
      "loss": 3.3897690773010254,
      "step": 1447
    },
    {
      "epoch": 0.2336237495966441,
      "grad_norm": 1.3813799619674683,
      "learning_rate": 0.0005,
      "loss": 3.3402047157287598,
      "step": 1448
    },
    {
      "epoch": 0.23378509196515004,
      "grad_norm": 1.7988779544830322,
      "learning_rate": 0.0005,
      "loss": 3.3990254402160645,
      "step": 1449
    },
    {
      "epoch": 0.23394643433365603,
      "grad_norm": 1.2889872789382935,
      "learning_rate": 0.0005,
      "loss": 3.4191792011260986,
      "step": 1450
    },
    {
      "epoch": 0.23410777670216198,
      "grad_norm": 1.3659318685531616,
      "learning_rate": 0.0005,
      "loss": 3.1882941722869873,
      "step": 1451
    },
    {
      "epoch": 0.23426911907066797,
      "grad_norm": 1.4459037780761719,
      "learning_rate": 0.0005,
      "loss": 2.7662453651428223,
      "step": 1452
    },
    {
      "epoch": 0.23443046143917393,
      "grad_norm": 1.3971556425094604,
      "learning_rate": 0.0005,
      "loss": 2.923492193222046,
      "step": 1453
    },
    {
      "epoch": 0.2345918038076799,
      "grad_norm": 1.7995027303695679,
      "learning_rate": 0.0005,
      "loss": 3.5248148441314697,
      "step": 1454
    },
    {
      "epoch": 0.23475314617618587,
      "grad_norm": 1.6778404712677002,
      "learning_rate": 0.0005,
      "loss": 3.2832441329956055,
      "step": 1455
    },
    {
      "epoch": 0.23491448854469182,
      "grad_norm": 1.3411849737167358,
      "learning_rate": 0.0005,
      "loss": 3.5022573471069336,
      "step": 1456
    },
    {
      "epoch": 0.2350758309131978,
      "grad_norm": 1.6724203824996948,
      "learning_rate": 0.0005,
      "loss": 3.204286575317383,
      "step": 1457
    },
    {
      "epoch": 0.23523717328170377,
      "grad_norm": 2.5130836963653564,
      "learning_rate": 0.0005,
      "loss": 3.0071003437042236,
      "step": 1458
    },
    {
      "epoch": 0.23539851565020975,
      "grad_norm": 1.5267523527145386,
      "learning_rate": 0.0005,
      "loss": 3.057711362838745,
      "step": 1459
    },
    {
      "epoch": 0.2355598580187157,
      "grad_norm": 1.7381603717803955,
      "learning_rate": 0.0005,
      "loss": 3.2340874671936035,
      "step": 1460
    },
    {
      "epoch": 0.2357212003872217,
      "grad_norm": 1.6917680501937866,
      "learning_rate": 0.0005,
      "loss": 3.1693828105926514,
      "step": 1461
    },
    {
      "epoch": 0.23588254275572765,
      "grad_norm": 1.4833282232284546,
      "learning_rate": 0.0005,
      "loss": 3.272184371948242,
      "step": 1462
    },
    {
      "epoch": 0.23604388512423363,
      "grad_norm": 1.5960791110992432,
      "learning_rate": 0.0005,
      "loss": 3.371854305267334,
      "step": 1463
    },
    {
      "epoch": 0.2362052274927396,
      "grad_norm": 1.4845095872879028,
      "learning_rate": 0.0005,
      "loss": 3.2127461433410645,
      "step": 1464
    },
    {
      "epoch": 0.23636656986124557,
      "grad_norm": 1.4195804595947266,
      "learning_rate": 0.0005,
      "loss": 3.205540657043457,
      "step": 1465
    },
    {
      "epoch": 0.23652791222975153,
      "grad_norm": 1.8350666761398315,
      "learning_rate": 0.0005,
      "loss": 3.7420177459716797,
      "step": 1466
    },
    {
      "epoch": 0.23668925459825751,
      "grad_norm": 1.5919194221496582,
      "learning_rate": 0.0005,
      "loss": 3.036538600921631,
      "step": 1467
    },
    {
      "epoch": 0.23685059696676347,
      "grad_norm": 1.5289723873138428,
      "learning_rate": 0.0005,
      "loss": 3.135253667831421,
      "step": 1468
    },
    {
      "epoch": 0.23701193933526943,
      "grad_norm": 2.672919988632202,
      "learning_rate": 0.0005,
      "loss": 3.6997361183166504,
      "step": 1469
    },
    {
      "epoch": 0.2371732817037754,
      "grad_norm": 1.3796628713607788,
      "learning_rate": 0.0005,
      "loss": 2.9344305992126465,
      "step": 1470
    },
    {
      "epoch": 0.23733462407228137,
      "grad_norm": 1.4271003007888794,
      "learning_rate": 0.0005,
      "loss": 3.4249706268310547,
      "step": 1471
    },
    {
      "epoch": 0.23749596644078735,
      "grad_norm": 1.4324052333831787,
      "learning_rate": 0.0005,
      "loss": 3.0739994049072266,
      "step": 1472
    },
    {
      "epoch": 0.2376573088092933,
      "grad_norm": 1.2858556509017944,
      "learning_rate": 0.0005,
      "loss": 3.450113534927368,
      "step": 1473
    },
    {
      "epoch": 0.2378186511777993,
      "grad_norm": 1.490950107574463,
      "learning_rate": 0.0005,
      "loss": 3.801433801651001,
      "step": 1474
    },
    {
      "epoch": 0.23797999354630525,
      "grad_norm": 1.3472425937652588,
      "learning_rate": 0.0005,
      "loss": 3.2858800888061523,
      "step": 1475
    },
    {
      "epoch": 0.23814133591481124,
      "grad_norm": 1.236762523651123,
      "learning_rate": 0.0005,
      "loss": 3.1905016899108887,
      "step": 1476
    },
    {
      "epoch": 0.2383026782833172,
      "grad_norm": 1.5638941526412964,
      "learning_rate": 0.0005,
      "loss": 3.432058811187744,
      "step": 1477
    },
    {
      "epoch": 0.23846402065182318,
      "grad_norm": 2.287832260131836,
      "learning_rate": 0.0005,
      "loss": 3.326460361480713,
      "step": 1478
    },
    {
      "epoch": 0.23862536302032913,
      "grad_norm": 1.4600815773010254,
      "learning_rate": 0.0005,
      "loss": 3.2761964797973633,
      "step": 1479
    },
    {
      "epoch": 0.23878670538883512,
      "grad_norm": 1.7432634830474854,
      "learning_rate": 0.0005,
      "loss": 3.2703185081481934,
      "step": 1480
    },
    {
      "epoch": 0.23894804775734108,
      "grad_norm": 1.587207555770874,
      "learning_rate": 0.0005,
      "loss": 3.3679351806640625,
      "step": 1481
    },
    {
      "epoch": 0.23910939012584706,
      "grad_norm": 1.5246978998184204,
      "learning_rate": 0.0005,
      "loss": 3.28592848777771,
      "step": 1482
    },
    {
      "epoch": 0.23927073249435302,
      "grad_norm": 1.4240901470184326,
      "learning_rate": 0.0005,
      "loss": 3.194167137145996,
      "step": 1483
    },
    {
      "epoch": 0.23943207486285897,
      "grad_norm": 1.4908435344696045,
      "learning_rate": 0.0005,
      "loss": 3.0718257427215576,
      "step": 1484
    },
    {
      "epoch": 0.23959341723136496,
      "grad_norm": 1.4235860109329224,
      "learning_rate": 0.0005,
      "loss": 3.4330482482910156,
      "step": 1485
    },
    {
      "epoch": 0.23975475959987091,
      "grad_norm": 2.3437936305999756,
      "learning_rate": 0.0005,
      "loss": 3.66405987739563,
      "step": 1486
    },
    {
      "epoch": 0.2399161019683769,
      "grad_norm": 1.860547423362732,
      "learning_rate": 0.0005,
      "loss": 3.1828925609588623,
      "step": 1487
    },
    {
      "epoch": 0.24007744433688286,
      "grad_norm": 1.2706987857818604,
      "learning_rate": 0.0005,
      "loss": 3.3418283462524414,
      "step": 1488
    },
    {
      "epoch": 0.24023878670538884,
      "grad_norm": 1.2686960697174072,
      "learning_rate": 0.0005,
      "loss": 3.088235378265381,
      "step": 1489
    },
    {
      "epoch": 0.2404001290738948,
      "grad_norm": 1.401301383972168,
      "learning_rate": 0.0005,
      "loss": 2.8167757987976074,
      "step": 1490
    },
    {
      "epoch": 0.24056147144240078,
      "grad_norm": 1.1288173198699951,
      "learning_rate": 0.0005,
      "loss": 3.7694218158721924,
      "step": 1491
    },
    {
      "epoch": 0.24072281381090674,
      "grad_norm": 1.4664639234542847,
      "learning_rate": 0.0005,
      "loss": 3.2681713104248047,
      "step": 1492
    },
    {
      "epoch": 0.24088415617941272,
      "grad_norm": 1.4199347496032715,
      "learning_rate": 0.0005,
      "loss": 3.3110861778259277,
      "step": 1493
    },
    {
      "epoch": 0.24104549854791868,
      "grad_norm": 1.3436429500579834,
      "learning_rate": 0.0005,
      "loss": 3.185459613800049,
      "step": 1494
    },
    {
      "epoch": 0.24120684091642466,
      "grad_norm": 1.1691840887069702,
      "learning_rate": 0.0005,
      "loss": 3.068556547164917,
      "step": 1495
    },
    {
      "epoch": 0.24136818328493062,
      "grad_norm": 1.2489036321640015,
      "learning_rate": 0.0005,
      "loss": 2.767047643661499,
      "step": 1496
    },
    {
      "epoch": 0.2415295256534366,
      "grad_norm": 1.3089550733566284,
      "learning_rate": 0.0005,
      "loss": 3.6193490028381348,
      "step": 1497
    },
    {
      "epoch": 0.24169086802194256,
      "grad_norm": 1.2364904880523682,
      "learning_rate": 0.0005,
      "loss": 3.1845247745513916,
      "step": 1498
    },
    {
      "epoch": 0.24185221039044852,
      "grad_norm": 1.2207263708114624,
      "learning_rate": 0.0005,
      "loss": 3.415740728378296,
      "step": 1499
    },
    {
      "epoch": 0.2420135527589545,
      "grad_norm": 9.942374229431152,
      "learning_rate": 0.0005,
      "loss": 2.9146461486816406,
      "step": 1500
    },
    {
      "epoch": 0.24217489512746046,
      "grad_norm": 1.4116495847702026,
      "learning_rate": 0.0005,
      "loss": 3.354069471359253,
      "step": 1501
    },
    {
      "epoch": 0.24233623749596644,
      "grad_norm": 1.8158323764801025,
      "learning_rate": 0.0005,
      "loss": 3.713914394378662,
      "step": 1502
    },
    {
      "epoch": 0.2424975798644724,
      "grad_norm": 1.4824507236480713,
      "learning_rate": 0.0005,
      "loss": 3.21980619430542,
      "step": 1503
    },
    {
      "epoch": 0.24265892223297839,
      "grad_norm": 1.46474027633667,
      "learning_rate": 0.0005,
      "loss": 3.281754493713379,
      "step": 1504
    },
    {
      "epoch": 0.24282026460148434,
      "grad_norm": 1.680704116821289,
      "learning_rate": 0.0005,
      "loss": 3.1523184776306152,
      "step": 1505
    },
    {
      "epoch": 0.24298160696999033,
      "grad_norm": 1.4216419458389282,
      "learning_rate": 0.0005,
      "loss": 3.6047921180725098,
      "step": 1506
    },
    {
      "epoch": 0.24314294933849628,
      "grad_norm": 1.1642154455184937,
      "learning_rate": 0.0005,
      "loss": 3.2775466442108154,
      "step": 1507
    },
    {
      "epoch": 0.24330429170700227,
      "grad_norm": 1.287128210067749,
      "learning_rate": 0.0005,
      "loss": 3.1186931133270264,
      "step": 1508
    },
    {
      "epoch": 0.24346563407550822,
      "grad_norm": 1.1363948583602905,
      "learning_rate": 0.0005,
      "loss": 3.5771024227142334,
      "step": 1509
    },
    {
      "epoch": 0.2436269764440142,
      "grad_norm": 1.265478253364563,
      "learning_rate": 0.0005,
      "loss": 3.336188793182373,
      "step": 1510
    },
    {
      "epoch": 0.24378831881252017,
      "grad_norm": 1.1913422346115112,
      "learning_rate": 0.0005,
      "loss": 3.2020809650421143,
      "step": 1511
    },
    {
      "epoch": 0.24394966118102615,
      "grad_norm": 1.4239387512207031,
      "learning_rate": 0.0005,
      "loss": 2.94389009475708,
      "step": 1512
    },
    {
      "epoch": 0.2441110035495321,
      "grad_norm": 1.4288761615753174,
      "learning_rate": 0.0005,
      "loss": 3.4371337890625,
      "step": 1513
    },
    {
      "epoch": 0.24427234591803806,
      "grad_norm": 1.7302837371826172,
      "learning_rate": 0.0005,
      "loss": 3.5379557609558105,
      "step": 1514
    },
    {
      "epoch": 0.24443368828654405,
      "grad_norm": 1.3204869031906128,
      "learning_rate": 0.0005,
      "loss": 3.172882080078125,
      "step": 1515
    },
    {
      "epoch": 0.24459503065505,
      "grad_norm": 1.7811018228530884,
      "learning_rate": 0.0005,
      "loss": 3.203789710998535,
      "step": 1516
    },
    {
      "epoch": 0.244756373023556,
      "grad_norm": 1.2239799499511719,
      "learning_rate": 0.0005,
      "loss": 2.7676448822021484,
      "step": 1517
    },
    {
      "epoch": 0.24491771539206195,
      "grad_norm": 1.1973847150802612,
      "learning_rate": 0.0005,
      "loss": 2.8673505783081055,
      "step": 1518
    },
    {
      "epoch": 0.24507905776056793,
      "grad_norm": 1.3549443483352661,
      "learning_rate": 0.0005,
      "loss": 3.0614490509033203,
      "step": 1519
    },
    {
      "epoch": 0.2452404001290739,
      "grad_norm": 1.557590126991272,
      "learning_rate": 0.0005,
      "loss": 3.452223777770996,
      "step": 1520
    },
    {
      "epoch": 0.24540174249757987,
      "grad_norm": 1.9828721284866333,
      "learning_rate": 0.0005,
      "loss": 3.8930015563964844,
      "step": 1521
    },
    {
      "epoch": 0.24556308486608583,
      "grad_norm": 1.8474102020263672,
      "learning_rate": 0.0005,
      "loss": 3.730252742767334,
      "step": 1522
    },
    {
      "epoch": 0.2457244272345918,
      "grad_norm": 1.4005237817764282,
      "learning_rate": 0.0005,
      "loss": 3.1540727615356445,
      "step": 1523
    },
    {
      "epoch": 0.24588576960309777,
      "grad_norm": 1.4455907344818115,
      "learning_rate": 0.0005,
      "loss": 3.8215932846069336,
      "step": 1524
    },
    {
      "epoch": 0.24604711197160375,
      "grad_norm": 1.7345037460327148,
      "learning_rate": 0.0005,
      "loss": 3.4764742851257324,
      "step": 1525
    },
    {
      "epoch": 0.2462084543401097,
      "grad_norm": 1.7877613306045532,
      "learning_rate": 0.0005,
      "loss": 2.7554965019226074,
      "step": 1526
    },
    {
      "epoch": 0.2463697967086157,
      "grad_norm": 1.4792683124542236,
      "learning_rate": 0.0005,
      "loss": 2.9199180603027344,
      "step": 1527
    },
    {
      "epoch": 0.24653113907712165,
      "grad_norm": 1.359358310699463,
      "learning_rate": 0.0005,
      "loss": 3.3109030723571777,
      "step": 1528
    },
    {
      "epoch": 0.2466924814456276,
      "grad_norm": 1.3547652959823608,
      "learning_rate": 0.0005,
      "loss": 3.0358633995056152,
      "step": 1529
    },
    {
      "epoch": 0.2468538238141336,
      "grad_norm": 1.938842535018921,
      "learning_rate": 0.0005,
      "loss": 2.5967273712158203,
      "step": 1530
    },
    {
      "epoch": 0.24701516618263955,
      "grad_norm": 1.5488996505737305,
      "learning_rate": 0.0005,
      "loss": 2.575000286102295,
      "step": 1531
    },
    {
      "epoch": 0.24717650855114554,
      "grad_norm": 1.3918983936309814,
      "learning_rate": 0.0005,
      "loss": 3.300502300262451,
      "step": 1532
    },
    {
      "epoch": 0.2473378509196515,
      "grad_norm": 1.2987489700317383,
      "learning_rate": 0.0005,
      "loss": 3.141278028488159,
      "step": 1533
    },
    {
      "epoch": 0.24749919328815748,
      "grad_norm": 1.5647035837173462,
      "learning_rate": 0.0005,
      "loss": 3.3739066123962402,
      "step": 1534
    },
    {
      "epoch": 0.24766053565666343,
      "grad_norm": 1.3672243356704712,
      "learning_rate": 0.0005,
      "loss": 3.5796594619750977,
      "step": 1535
    },
    {
      "epoch": 0.24782187802516942,
      "grad_norm": 1.3067660331726074,
      "learning_rate": 0.0005,
      "loss": 3.1528103351593018,
      "step": 1536
    },
    {
      "epoch": 0.24798322039367537,
      "grad_norm": 1.6905784606933594,
      "learning_rate": 0.0005,
      "loss": 3.541537284851074,
      "step": 1537
    },
    {
      "epoch": 0.24814456276218136,
      "grad_norm": 2.4473066329956055,
      "learning_rate": 0.0005,
      "loss": 3.136711835861206,
      "step": 1538
    },
    {
      "epoch": 0.24830590513068732,
      "grad_norm": 1.600053310394287,
      "learning_rate": 0.0005,
      "loss": 3.584115982055664,
      "step": 1539
    },
    {
      "epoch": 0.2484672474991933,
      "grad_norm": 1.3478516340255737,
      "learning_rate": 0.0005,
      "loss": 2.830345630645752,
      "step": 1540
    },
    {
      "epoch": 0.24862858986769926,
      "grad_norm": 1.9324246644973755,
      "learning_rate": 0.0005,
      "loss": 3.2719218730926514,
      "step": 1541
    },
    {
      "epoch": 0.24878993223620524,
      "grad_norm": 1.377687931060791,
      "learning_rate": 0.0005,
      "loss": 3.435147523880005,
      "step": 1542
    },
    {
      "epoch": 0.2489512746047112,
      "grad_norm": 1.8846125602722168,
      "learning_rate": 0.0005,
      "loss": 3.419313669204712,
      "step": 1543
    },
    {
      "epoch": 0.24911261697321715,
      "grad_norm": 1.1780015230178833,
      "learning_rate": 0.0005,
      "loss": 2.994624614715576,
      "step": 1544
    },
    {
      "epoch": 0.24927395934172314,
      "grad_norm": 1.1325129270553589,
      "learning_rate": 0.0005,
      "loss": 2.973118305206299,
      "step": 1545
    },
    {
      "epoch": 0.2494353017102291,
      "grad_norm": 1.5040174722671509,
      "learning_rate": 0.0005,
      "loss": 3.058436870574951,
      "step": 1546
    },
    {
      "epoch": 0.24959664407873508,
      "grad_norm": 1.4376541376113892,
      "learning_rate": 0.0005,
      "loss": 3.3107948303222656,
      "step": 1547
    },
    {
      "epoch": 0.24975798644724104,
      "grad_norm": 1.491065502166748,
      "learning_rate": 0.0005,
      "loss": 3.4333958625793457,
      "step": 1548
    },
    {
      "epoch": 0.24991932881574702,
      "grad_norm": 2.156470775604248,
      "learning_rate": 0.0005,
      "loss": 3.2426304817199707,
      "step": 1549
    },
    {
      "epoch": 0.250080671184253,
      "grad_norm": 1.938185453414917,
      "learning_rate": 0.0005,
      "loss": 3.0567283630371094,
      "step": 1550
    },
    {
      "epoch": 0.25024201355275894,
      "grad_norm": 1.5242335796356201,
      "learning_rate": 0.0005,
      "loss": 3.4294204711914062,
      "step": 1551
    },
    {
      "epoch": 0.25040335592126495,
      "grad_norm": 1.6660243272781372,
      "learning_rate": 0.0005,
      "loss": 3.411564826965332,
      "step": 1552
    },
    {
      "epoch": 0.2505646982897709,
      "grad_norm": 2.2480275630950928,
      "learning_rate": 0.0005,
      "loss": 3.10317325592041,
      "step": 1553
    },
    {
      "epoch": 0.25072604065827686,
      "grad_norm": 2.0322763919830322,
      "learning_rate": 0.0005,
      "loss": 3.2830724716186523,
      "step": 1554
    },
    {
      "epoch": 0.2508873830267828,
      "grad_norm": 1.5556323528289795,
      "learning_rate": 0.0005,
      "loss": 3.640115261077881,
      "step": 1555
    },
    {
      "epoch": 0.25104872539528883,
      "grad_norm": 1.256443977355957,
      "learning_rate": 0.0005,
      "loss": 3.515200614929199,
      "step": 1556
    },
    {
      "epoch": 0.2512100677637948,
      "grad_norm": 1.5667892694473267,
      "learning_rate": 0.0005,
      "loss": 3.306593894958496,
      "step": 1557
    },
    {
      "epoch": 0.25137141013230074,
      "grad_norm": 1.6073228120803833,
      "learning_rate": 0.0005,
      "loss": 3.2359464168548584,
      "step": 1558
    },
    {
      "epoch": 0.2515327525008067,
      "grad_norm": 1.4856642484664917,
      "learning_rate": 0.0005,
      "loss": 3.119992733001709,
      "step": 1559
    },
    {
      "epoch": 0.25169409486931266,
      "grad_norm": 1.251551628112793,
      "learning_rate": 0.0005,
      "loss": 3.433438301086426,
      "step": 1560
    },
    {
      "epoch": 0.25185543723781867,
      "grad_norm": 1.6150712966918945,
      "learning_rate": 0.0005,
      "loss": 3.1833982467651367,
      "step": 1561
    },
    {
      "epoch": 0.2520167796063246,
      "grad_norm": 3.580983877182007,
      "learning_rate": 0.0005,
      "loss": 2.9001049995422363,
      "step": 1562
    },
    {
      "epoch": 0.2521781219748306,
      "grad_norm": 1.241021752357483,
      "learning_rate": 0.0005,
      "loss": 3.2607336044311523,
      "step": 1563
    },
    {
      "epoch": 0.25233946434333654,
      "grad_norm": 10.68409538269043,
      "learning_rate": 0.0005,
      "loss": 3.6516075134277344,
      "step": 1564
    },
    {
      "epoch": 0.25250080671184255,
      "grad_norm": 1.409196138381958,
      "learning_rate": 0.0005,
      "loss": 3.570652484893799,
      "step": 1565
    },
    {
      "epoch": 0.2526621490803485,
      "grad_norm": 1.2400658130645752,
      "learning_rate": 0.0005,
      "loss": 2.9615869522094727,
      "step": 1566
    },
    {
      "epoch": 0.25282349144885446,
      "grad_norm": 1.4355686902999878,
      "learning_rate": 0.0005,
      "loss": 3.2709035873413086,
      "step": 1567
    },
    {
      "epoch": 0.2529848338173604,
      "grad_norm": 108.94119262695312,
      "learning_rate": 0.0005,
      "loss": 2.8776347637176514,
      "step": 1568
    },
    {
      "epoch": 0.25314617618586643,
      "grad_norm": 1.655382752418518,
      "learning_rate": 0.0005,
      "loss": 3.3165884017944336,
      "step": 1569
    },
    {
      "epoch": 0.2533075185543724,
      "grad_norm": 1.4527326822280884,
      "learning_rate": 0.0005,
      "loss": 2.918814182281494,
      "step": 1570
    },
    {
      "epoch": 0.25346886092287835,
      "grad_norm": 1.7033276557922363,
      "learning_rate": 0.0005,
      "loss": 3.2827260494232178,
      "step": 1571
    },
    {
      "epoch": 0.2536302032913843,
      "grad_norm": 1.6554657220840454,
      "learning_rate": 0.0005,
      "loss": 3.940463066101074,
      "step": 1572
    },
    {
      "epoch": 0.25379154565989026,
      "grad_norm": 1.6690161228179932,
      "learning_rate": 0.0005,
      "loss": 3.4801809787750244,
      "step": 1573
    },
    {
      "epoch": 0.2539528880283963,
      "grad_norm": 1.086930751800537,
      "learning_rate": 0.0005,
      "loss": 3.526937961578369,
      "step": 1574
    },
    {
      "epoch": 0.25411423039690223,
      "grad_norm": 1.3746223449707031,
      "learning_rate": 0.0005,
      "loss": 3.2499337196350098,
      "step": 1575
    },
    {
      "epoch": 0.2542755727654082,
      "grad_norm": 1.6346627473831177,
      "learning_rate": 0.0005,
      "loss": 3.0731282234191895,
      "step": 1576
    },
    {
      "epoch": 0.25443691513391414,
      "grad_norm": 1.5189565420150757,
      "learning_rate": 0.0005,
      "loss": 3.0145621299743652,
      "step": 1577
    },
    {
      "epoch": 0.25459825750242016,
      "grad_norm": 1.5421743392944336,
      "learning_rate": 0.0005,
      "loss": 3.337289333343506,
      "step": 1578
    },
    {
      "epoch": 0.2547595998709261,
      "grad_norm": 1.295277714729309,
      "learning_rate": 0.0005,
      "loss": 3.150665521621704,
      "step": 1579
    },
    {
      "epoch": 0.25492094223943207,
      "grad_norm": 1.5919435024261475,
      "learning_rate": 0.0005,
      "loss": 3.352083444595337,
      "step": 1580
    },
    {
      "epoch": 0.255082284607938,
      "grad_norm": 1.502811074256897,
      "learning_rate": 0.0005,
      "loss": 3.2808189392089844,
      "step": 1581
    },
    {
      "epoch": 0.25524362697644404,
      "grad_norm": 1.4425431489944458,
      "learning_rate": 0.0005,
      "loss": 2.886160135269165,
      "step": 1582
    },
    {
      "epoch": 0.25540496934495,
      "grad_norm": 2.139309883117676,
      "learning_rate": 0.0005,
      "loss": 3.1462087631225586,
      "step": 1583
    },
    {
      "epoch": 0.25556631171345595,
      "grad_norm": 1.4285918474197388,
      "learning_rate": 0.0005,
      "loss": 3.0223731994628906,
      "step": 1584
    },
    {
      "epoch": 0.2557276540819619,
      "grad_norm": 1.302526593208313,
      "learning_rate": 0.0005,
      "loss": 2.9652745723724365,
      "step": 1585
    },
    {
      "epoch": 0.2558889964504679,
      "grad_norm": 1.3082737922668457,
      "learning_rate": 0.0005,
      "loss": 3.6442909240722656,
      "step": 1586
    },
    {
      "epoch": 0.2560503388189739,
      "grad_norm": 1.592332363128662,
      "learning_rate": 0.0005,
      "loss": 3.3252079486846924,
      "step": 1587
    },
    {
      "epoch": 0.25621168118747983,
      "grad_norm": 1.4279838800430298,
      "learning_rate": 0.0005,
      "loss": 2.944575071334839,
      "step": 1588
    },
    {
      "epoch": 0.2563730235559858,
      "grad_norm": 1.5703517198562622,
      "learning_rate": 0.0005,
      "loss": 3.451416015625,
      "step": 1589
    },
    {
      "epoch": 0.25653436592449175,
      "grad_norm": 1.6329541206359863,
      "learning_rate": 0.0005,
      "loss": 3.408999443054199,
      "step": 1590
    },
    {
      "epoch": 0.25669570829299776,
      "grad_norm": 1.7835019826889038,
      "learning_rate": 0.0005,
      "loss": 3.684377431869507,
      "step": 1591
    },
    {
      "epoch": 0.2568570506615037,
      "grad_norm": 1.5051021575927734,
      "learning_rate": 0.0005,
      "loss": 3.7739129066467285,
      "step": 1592
    },
    {
      "epoch": 0.2570183930300097,
      "grad_norm": 1.5561312437057495,
      "learning_rate": 0.0005,
      "loss": 3.292377233505249,
      "step": 1593
    },
    {
      "epoch": 0.25717973539851563,
      "grad_norm": 1.3915938138961792,
      "learning_rate": 0.0005,
      "loss": 3.354741096496582,
      "step": 1594
    },
    {
      "epoch": 0.25734107776702164,
      "grad_norm": 1.2948188781738281,
      "learning_rate": 0.0005,
      "loss": 2.92380952835083,
      "step": 1595
    },
    {
      "epoch": 0.2575024201355276,
      "grad_norm": 1.5722659826278687,
      "learning_rate": 0.0005,
      "loss": 3.4996886253356934,
      "step": 1596
    },
    {
      "epoch": 0.25766376250403356,
      "grad_norm": 1.581465482711792,
      "learning_rate": 0.0005,
      "loss": 3.5547666549682617,
      "step": 1597
    },
    {
      "epoch": 0.2578251048725395,
      "grad_norm": 1.5556238889694214,
      "learning_rate": 0.0005,
      "loss": 3.088705539703369,
      "step": 1598
    },
    {
      "epoch": 0.2579864472410455,
      "grad_norm": 1.7777069807052612,
      "learning_rate": 0.0005,
      "loss": 3.503448963165283,
      "step": 1599
    },
    {
      "epoch": 0.2581477896095515,
      "grad_norm": 1.362998604774475,
      "learning_rate": 0.0005,
      "loss": 2.678828716278076,
      "step": 1600
    },
    {
      "epoch": 0.25830913197805744,
      "grad_norm": 1.3254700899124146,
      "learning_rate": 0.0005,
      "loss": 3.336543083190918,
      "step": 1601
    },
    {
      "epoch": 0.2584704743465634,
      "grad_norm": 1.1662037372589111,
      "learning_rate": 0.0005,
      "loss": 3.1558544635772705,
      "step": 1602
    },
    {
      "epoch": 0.25863181671506935,
      "grad_norm": 1.3930257558822632,
      "learning_rate": 0.0005,
      "loss": 3.3309035301208496,
      "step": 1603
    },
    {
      "epoch": 0.25879315908357536,
      "grad_norm": 1.7576138973236084,
      "learning_rate": 0.0005,
      "loss": 3.2572739124298096,
      "step": 1604
    },
    {
      "epoch": 0.2589545014520813,
      "grad_norm": 18.374719619750977,
      "learning_rate": 0.0005,
      "loss": 3.0090713500976562,
      "step": 1605
    },
    {
      "epoch": 0.2591158438205873,
      "grad_norm": 10.72853946685791,
      "learning_rate": 0.0005,
      "loss": 2.8627240657806396,
      "step": 1606
    },
    {
      "epoch": 0.25927718618909323,
      "grad_norm": 1.6290076971054077,
      "learning_rate": 0.0005,
      "loss": 3.1008801460266113,
      "step": 1607
    },
    {
      "epoch": 0.25943852855759925,
      "grad_norm": 1.6372883319854736,
      "learning_rate": 0.0005,
      "loss": 3.123154878616333,
      "step": 1608
    },
    {
      "epoch": 0.2595998709261052,
      "grad_norm": 1.53804349899292,
      "learning_rate": 0.0005,
      "loss": 3.0029163360595703,
      "step": 1609
    },
    {
      "epoch": 0.25976121329461116,
      "grad_norm": 1.3677220344543457,
      "learning_rate": 0.0005,
      "loss": 3.193943977355957,
      "step": 1610
    },
    {
      "epoch": 0.2599225556631171,
      "grad_norm": 1.3841097354888916,
      "learning_rate": 0.0005,
      "loss": 3.245738983154297,
      "step": 1611
    },
    {
      "epoch": 0.26008389803162313,
      "grad_norm": 1.6780726909637451,
      "learning_rate": 0.0005,
      "loss": 2.926283836364746,
      "step": 1612
    },
    {
      "epoch": 0.2602452404001291,
      "grad_norm": 1.5621551275253296,
      "learning_rate": 0.0005,
      "loss": 3.0798444747924805,
      "step": 1613
    },
    {
      "epoch": 0.26040658276863504,
      "grad_norm": 1.374362826347351,
      "learning_rate": 0.0005,
      "loss": 3.140047311782837,
      "step": 1614
    },
    {
      "epoch": 0.260567925137141,
      "grad_norm": 1.6402031183242798,
      "learning_rate": 0.0005,
      "loss": 2.846940755844116,
      "step": 1615
    },
    {
      "epoch": 0.26072926750564696,
      "grad_norm": 1.4847900867462158,
      "learning_rate": 0.0005,
      "loss": 3.431392192840576,
      "step": 1616
    },
    {
      "epoch": 0.26089060987415297,
      "grad_norm": 1.5563013553619385,
      "learning_rate": 0.0005,
      "loss": 3.400418758392334,
      "step": 1617
    },
    {
      "epoch": 0.2610519522426589,
      "grad_norm": 1.792910099029541,
      "learning_rate": 0.0005,
      "loss": 3.867997884750366,
      "step": 1618
    },
    {
      "epoch": 0.2612132946111649,
      "grad_norm": 1.4163093566894531,
      "learning_rate": 0.0005,
      "loss": 3.071242332458496,
      "step": 1619
    },
    {
      "epoch": 0.26137463697967084,
      "grad_norm": 1.1509305238723755,
      "learning_rate": 0.0005,
      "loss": 3.031999349594116,
      "step": 1620
    },
    {
      "epoch": 0.26153597934817685,
      "grad_norm": 1.5386030673980713,
      "learning_rate": 0.0005,
      "loss": 3.0979514122009277,
      "step": 1621
    },
    {
      "epoch": 0.2616973217166828,
      "grad_norm": 1.2734450101852417,
      "learning_rate": 0.0005,
      "loss": 3.253704786300659,
      "step": 1622
    },
    {
      "epoch": 0.26185866408518876,
      "grad_norm": 1.3574612140655518,
      "learning_rate": 0.0005,
      "loss": 3.2417027950286865,
      "step": 1623
    },
    {
      "epoch": 0.2620200064536947,
      "grad_norm": 2.4867148399353027,
      "learning_rate": 0.0005,
      "loss": 3.849055051803589,
      "step": 1624
    },
    {
      "epoch": 0.26218134882220073,
      "grad_norm": 1.5146878957748413,
      "learning_rate": 0.0005,
      "loss": 3.0760533809661865,
      "step": 1625
    },
    {
      "epoch": 0.2623426911907067,
      "grad_norm": 1.4255038499832153,
      "learning_rate": 0.0005,
      "loss": 3.1591060161590576,
      "step": 1626
    },
    {
      "epoch": 0.26250403355921265,
      "grad_norm": 1.4919010400772095,
      "learning_rate": 0.0005,
      "loss": 3.2336041927337646,
      "step": 1627
    },
    {
      "epoch": 0.2626653759277186,
      "grad_norm": 1.481250524520874,
      "learning_rate": 0.0005,
      "loss": 2.847080945968628,
      "step": 1628
    },
    {
      "epoch": 0.2628267182962246,
      "grad_norm": 1.8229814767837524,
      "learning_rate": 0.0005,
      "loss": 3.403824806213379,
      "step": 1629
    },
    {
      "epoch": 0.26298806066473057,
      "grad_norm": 1.9368354082107544,
      "learning_rate": 0.0005,
      "loss": 2.916506767272949,
      "step": 1630
    },
    {
      "epoch": 0.26314940303323653,
      "grad_norm": 1.49202561378479,
      "learning_rate": 0.0005,
      "loss": 3.6891281604766846,
      "step": 1631
    },
    {
      "epoch": 0.2633107454017425,
      "grad_norm": 1.6999320983886719,
      "learning_rate": 0.0005,
      "loss": 3.2508749961853027,
      "step": 1632
    },
    {
      "epoch": 0.26347208777024844,
      "grad_norm": 1.3576339483261108,
      "learning_rate": 0.0005,
      "loss": 3.229560613632202,
      "step": 1633
    },
    {
      "epoch": 0.26363343013875445,
      "grad_norm": 2.129791259765625,
      "learning_rate": 0.0005,
      "loss": 3.5419342517852783,
      "step": 1634
    },
    {
      "epoch": 0.2637947725072604,
      "grad_norm": 1.5662890672683716,
      "learning_rate": 0.0005,
      "loss": 3.0094780921936035,
      "step": 1635
    },
    {
      "epoch": 0.26395611487576637,
      "grad_norm": 1.4313404560089111,
      "learning_rate": 0.0005,
      "loss": 3.276224136352539,
      "step": 1636
    },
    {
      "epoch": 0.2641174572442723,
      "grad_norm": 1.9298311471939087,
      "learning_rate": 0.0005,
      "loss": 3.686324119567871,
      "step": 1637
    },
    {
      "epoch": 0.26427879961277834,
      "grad_norm": 1.1663532257080078,
      "learning_rate": 0.0005,
      "loss": 2.852919816970825,
      "step": 1638
    },
    {
      "epoch": 0.2644401419812843,
      "grad_norm": 1.353412389755249,
      "learning_rate": 0.0005,
      "loss": 3.509397506713867,
      "step": 1639
    },
    {
      "epoch": 0.26460148434979025,
      "grad_norm": 1.353730320930481,
      "learning_rate": 0.0005,
      "loss": 3.352783203125,
      "step": 1640
    },
    {
      "epoch": 0.2647628267182962,
      "grad_norm": 1.4927400350570679,
      "learning_rate": 0.0005,
      "loss": 3.085134267807007,
      "step": 1641
    },
    {
      "epoch": 0.2649241690868022,
      "grad_norm": 1.8347214460372925,
      "learning_rate": 0.0005,
      "loss": 3.604006052017212,
      "step": 1642
    },
    {
      "epoch": 0.2650855114553082,
      "grad_norm": 1.2460768222808838,
      "learning_rate": 0.0005,
      "loss": 3.2519848346710205,
      "step": 1643
    },
    {
      "epoch": 0.26524685382381413,
      "grad_norm": 1.0898252725601196,
      "learning_rate": 0.0005,
      "loss": 3.338751792907715,
      "step": 1644
    },
    {
      "epoch": 0.2654081961923201,
      "grad_norm": 1.368887186050415,
      "learning_rate": 0.0005,
      "loss": 3.5804409980773926,
      "step": 1645
    },
    {
      "epoch": 0.26556953856082605,
      "grad_norm": 1.4023422002792358,
      "learning_rate": 0.0005,
      "loss": 3.158735513687134,
      "step": 1646
    },
    {
      "epoch": 0.26573088092933206,
      "grad_norm": 1.5394017696380615,
      "learning_rate": 0.0005,
      "loss": 3.2300310134887695,
      "step": 1647
    },
    {
      "epoch": 0.265892223297838,
      "grad_norm": 1.3189154863357544,
      "learning_rate": 0.0005,
      "loss": 3.3391146659851074,
      "step": 1648
    },
    {
      "epoch": 0.26605356566634397,
      "grad_norm": 1.4290072917938232,
      "learning_rate": 0.0005,
      "loss": 2.474639654159546,
      "step": 1649
    },
    {
      "epoch": 0.26621490803484993,
      "grad_norm": 2.164731502532959,
      "learning_rate": 0.0005,
      "loss": 3.2613587379455566,
      "step": 1650
    },
    {
      "epoch": 0.26637625040335594,
      "grad_norm": 7.421266555786133,
      "learning_rate": 0.0005,
      "loss": 3.115845203399658,
      "step": 1651
    },
    {
      "epoch": 0.2665375927718619,
      "grad_norm": 1.6847145557403564,
      "learning_rate": 0.0005,
      "loss": 2.589463233947754,
      "step": 1652
    },
    {
      "epoch": 0.26669893514036785,
      "grad_norm": 1.8459521532058716,
      "learning_rate": 0.0005,
      "loss": 3.309518337249756,
      "step": 1653
    },
    {
      "epoch": 0.2668602775088738,
      "grad_norm": 1.3885271549224854,
      "learning_rate": 0.0005,
      "loss": 3.3515594005584717,
      "step": 1654
    },
    {
      "epoch": 0.2670216198773798,
      "grad_norm": 1.271689534187317,
      "learning_rate": 0.0005,
      "loss": 3.0464117527008057,
      "step": 1655
    },
    {
      "epoch": 0.2671829622458858,
      "grad_norm": 1.7058203220367432,
      "learning_rate": 0.0005,
      "loss": 3.3335866928100586,
      "step": 1656
    },
    {
      "epoch": 0.26734430461439174,
      "grad_norm": 1.322982668876648,
      "learning_rate": 0.0005,
      "loss": 3.198869228363037,
      "step": 1657
    },
    {
      "epoch": 0.2675056469828977,
      "grad_norm": 1.5503326654434204,
      "learning_rate": 0.0005,
      "loss": 3.2790212631225586,
      "step": 1658
    },
    {
      "epoch": 0.2676669893514037,
      "grad_norm": 1.2720026969909668,
      "learning_rate": 0.0005,
      "loss": 3.463895082473755,
      "step": 1659
    },
    {
      "epoch": 0.26782833171990966,
      "grad_norm": 1.1736624240875244,
      "learning_rate": 0.0005,
      "loss": 3.0526344776153564,
      "step": 1660
    },
    {
      "epoch": 0.2679896740884156,
      "grad_norm": 1.596054196357727,
      "learning_rate": 0.0005,
      "loss": 3.1971704959869385,
      "step": 1661
    },
    {
      "epoch": 0.2681510164569216,
      "grad_norm": 1.5887982845306396,
      "learning_rate": 0.0005,
      "loss": 3.559108018875122,
      "step": 1662
    },
    {
      "epoch": 0.26831235882542753,
      "grad_norm": 1.640757441520691,
      "learning_rate": 0.0005,
      "loss": 3.4891085624694824,
      "step": 1663
    },
    {
      "epoch": 0.26847370119393354,
      "grad_norm": 1.6390581130981445,
      "learning_rate": 0.0005,
      "loss": 3.23006010055542,
      "step": 1664
    },
    {
      "epoch": 0.2686350435624395,
      "grad_norm": 1.4377750158309937,
      "learning_rate": 0.0005,
      "loss": 3.725656509399414,
      "step": 1665
    },
    {
      "epoch": 0.26879638593094546,
      "grad_norm": 1.940714955329895,
      "learning_rate": 0.0005,
      "loss": 3.3210065364837646,
      "step": 1666
    },
    {
      "epoch": 0.2689577282994514,
      "grad_norm": 1.4086887836456299,
      "learning_rate": 0.0005,
      "loss": 3.1768062114715576,
      "step": 1667
    },
    {
      "epoch": 0.2691190706679574,
      "grad_norm": 1.7567188739776611,
      "learning_rate": 0.0005,
      "loss": 3.884979248046875,
      "step": 1668
    },
    {
      "epoch": 0.2692804130364634,
      "grad_norm": 1.562415599822998,
      "learning_rate": 0.0005,
      "loss": 3.2139482498168945,
      "step": 1669
    },
    {
      "epoch": 0.26944175540496934,
      "grad_norm": 1.5614094734191895,
      "learning_rate": 0.0005,
      "loss": 3.464414119720459,
      "step": 1670
    },
    {
      "epoch": 0.2696030977734753,
      "grad_norm": 1.2561694383621216,
      "learning_rate": 0.0005,
      "loss": 3.3530635833740234,
      "step": 1671
    },
    {
      "epoch": 0.2697644401419813,
      "grad_norm": 1.4079277515411377,
      "learning_rate": 0.0005,
      "loss": 2.9364023208618164,
      "step": 1672
    },
    {
      "epoch": 0.26992578251048727,
      "grad_norm": 1.4413665533065796,
      "learning_rate": 0.0005,
      "loss": 3.7071712017059326,
      "step": 1673
    },
    {
      "epoch": 0.2700871248789932,
      "grad_norm": 1.1974130868911743,
      "learning_rate": 0.0005,
      "loss": 3.220889091491699,
      "step": 1674
    },
    {
      "epoch": 0.2702484672474992,
      "grad_norm": 1.2587807178497314,
      "learning_rate": 0.0005,
      "loss": 3.341981887817383,
      "step": 1675
    },
    {
      "epoch": 0.27040980961600514,
      "grad_norm": 1.559591293334961,
      "learning_rate": 0.0005,
      "loss": 2.7492592334747314,
      "step": 1676
    },
    {
      "epoch": 0.27057115198451115,
      "grad_norm": 1.4169957637786865,
      "learning_rate": 0.0005,
      "loss": 2.902708053588867,
      "step": 1677
    },
    {
      "epoch": 0.2707324943530171,
      "grad_norm": 1.4257659912109375,
      "learning_rate": 0.0005,
      "loss": 3.3808438777923584,
      "step": 1678
    },
    {
      "epoch": 0.27089383672152306,
      "grad_norm": 1.352614402770996,
      "learning_rate": 0.0005,
      "loss": 2.744605541229248,
      "step": 1679
    },
    {
      "epoch": 0.271055179090029,
      "grad_norm": 1.2593601942062378,
      "learning_rate": 0.0005,
      "loss": 2.966757297515869,
      "step": 1680
    },
    {
      "epoch": 0.27121652145853503,
      "grad_norm": 1.7770379781723022,
      "learning_rate": 0.0005,
      "loss": 3.4899559020996094,
      "step": 1681
    },
    {
      "epoch": 0.271377863827041,
      "grad_norm": 1.3414604663848877,
      "learning_rate": 0.0005,
      "loss": 3.311180830001831,
      "step": 1682
    },
    {
      "epoch": 0.27153920619554694,
      "grad_norm": 1.3099298477172852,
      "learning_rate": 0.0005,
      "loss": 3.2256479263305664,
      "step": 1683
    },
    {
      "epoch": 0.2717005485640529,
      "grad_norm": 1.7013559341430664,
      "learning_rate": 0.0005,
      "loss": 3.574258804321289,
      "step": 1684
    },
    {
      "epoch": 0.2718618909325589,
      "grad_norm": 1.189624309539795,
      "learning_rate": 0.0005,
      "loss": 3.1745786666870117,
      "step": 1685
    },
    {
      "epoch": 0.27202323330106487,
      "grad_norm": 1.7267290353775024,
      "learning_rate": 0.0005,
      "loss": 3.043466567993164,
      "step": 1686
    },
    {
      "epoch": 0.2721845756695708,
      "grad_norm": 1.0535928010940552,
      "learning_rate": 0.0005,
      "loss": 2.5558199882507324,
      "step": 1687
    },
    {
      "epoch": 0.2723459180380768,
      "grad_norm": 1.6098930835723877,
      "learning_rate": 0.0005,
      "loss": 3.2773079872131348,
      "step": 1688
    },
    {
      "epoch": 0.2725072604065828,
      "grad_norm": 1.4510284662246704,
      "learning_rate": 0.0005,
      "loss": 3.0027055740356445,
      "step": 1689
    },
    {
      "epoch": 0.27266860277508875,
      "grad_norm": 1.4205520153045654,
      "learning_rate": 0.0005,
      "loss": 3.1629581451416016,
      "step": 1690
    },
    {
      "epoch": 0.2728299451435947,
      "grad_norm": 1.4980623722076416,
      "learning_rate": 0.0005,
      "loss": 3.4229817390441895,
      "step": 1691
    },
    {
      "epoch": 0.27299128751210067,
      "grad_norm": 2.0293538570404053,
      "learning_rate": 0.0005,
      "loss": 3.0423665046691895,
      "step": 1692
    },
    {
      "epoch": 0.2731526298806066,
      "grad_norm": 1.887667179107666,
      "learning_rate": 0.0005,
      "loss": 3.202197313308716,
      "step": 1693
    },
    {
      "epoch": 0.27331397224911264,
      "grad_norm": 1.3265752792358398,
      "learning_rate": 0.0005,
      "loss": 3.2447187900543213,
      "step": 1694
    },
    {
      "epoch": 0.2734753146176186,
      "grad_norm": 1.2193373441696167,
      "learning_rate": 0.0005,
      "loss": 3.1061298847198486,
      "step": 1695
    },
    {
      "epoch": 0.27363665698612455,
      "grad_norm": 1.2150245904922485,
      "learning_rate": 0.0005,
      "loss": 3.3396410942077637,
      "step": 1696
    },
    {
      "epoch": 0.2737979993546305,
      "grad_norm": 1.5260615348815918,
      "learning_rate": 0.0005,
      "loss": 3.5924832820892334,
      "step": 1697
    },
    {
      "epoch": 0.2739593417231365,
      "grad_norm": 1.6300634145736694,
      "learning_rate": 0.0005,
      "loss": 3.015157699584961,
      "step": 1698
    },
    {
      "epoch": 0.2741206840916425,
      "grad_norm": 1.3623263835906982,
      "learning_rate": 0.0005,
      "loss": 2.89652156829834,
      "step": 1699
    },
    {
      "epoch": 0.27428202646014843,
      "grad_norm": 1.088668942451477,
      "learning_rate": 0.0005,
      "loss": 3.2758169174194336,
      "step": 1700
    },
    {
      "epoch": 0.2744433688286544,
      "grad_norm": 2.1121771335601807,
      "learning_rate": 0.0005,
      "loss": 3.3446900844573975,
      "step": 1701
    },
    {
      "epoch": 0.2746047111971604,
      "grad_norm": 1.2956770658493042,
      "learning_rate": 0.0005,
      "loss": 3.2470476627349854,
      "step": 1702
    },
    {
      "epoch": 0.27476605356566636,
      "grad_norm": 1.3161108493804932,
      "learning_rate": 0.0005,
      "loss": 2.8312907218933105,
      "step": 1703
    },
    {
      "epoch": 0.2749273959341723,
      "grad_norm": 1.5188132524490356,
      "learning_rate": 0.0005,
      "loss": 3.502849817276001,
      "step": 1704
    },
    {
      "epoch": 0.27508873830267827,
      "grad_norm": 1.4393161535263062,
      "learning_rate": 0.0005,
      "loss": 3.4333395957946777,
      "step": 1705
    },
    {
      "epoch": 0.2752500806711842,
      "grad_norm": 1.9982534646987915,
      "learning_rate": 0.0005,
      "loss": 3.396336555480957,
      "step": 1706
    },
    {
      "epoch": 0.27541142303969024,
      "grad_norm": 1.6625549793243408,
      "learning_rate": 0.0005,
      "loss": 3.5694780349731445,
      "step": 1707
    },
    {
      "epoch": 0.2755727654081962,
      "grad_norm": 1.5257140398025513,
      "learning_rate": 0.0005,
      "loss": 3.3930468559265137,
      "step": 1708
    },
    {
      "epoch": 0.27573410777670215,
      "grad_norm": 1.374059796333313,
      "learning_rate": 0.0005,
      "loss": 3.5069756507873535,
      "step": 1709
    },
    {
      "epoch": 0.2758954501452081,
      "grad_norm": 1.4254953861236572,
      "learning_rate": 0.0005,
      "loss": 3.7138609886169434,
      "step": 1710
    },
    {
      "epoch": 0.2760567925137141,
      "grad_norm": 2.036724805831909,
      "learning_rate": 0.0005,
      "loss": 3.2971532344818115,
      "step": 1711
    },
    {
      "epoch": 0.2762181348822201,
      "grad_norm": 1.2795811891555786,
      "learning_rate": 0.0005,
      "loss": 2.876094341278076,
      "step": 1712
    },
    {
      "epoch": 0.27637947725072604,
      "grad_norm": 1.3051139116287231,
      "learning_rate": 0.0005,
      "loss": 2.911823034286499,
      "step": 1713
    },
    {
      "epoch": 0.276540819619232,
      "grad_norm": 1.2088426351547241,
      "learning_rate": 0.0005,
      "loss": 3.306715965270996,
      "step": 1714
    },
    {
      "epoch": 0.276702161987738,
      "grad_norm": 1.5069187879562378,
      "learning_rate": 0.0005,
      "loss": 3.4377493858337402,
      "step": 1715
    },
    {
      "epoch": 0.27686350435624396,
      "grad_norm": 1.343248724937439,
      "learning_rate": 0.0005,
      "loss": 3.275068759918213,
      "step": 1716
    },
    {
      "epoch": 0.2770248467247499,
      "grad_norm": 1.9208714962005615,
      "learning_rate": 0.0005,
      "loss": 3.1912012100219727,
      "step": 1717
    },
    {
      "epoch": 0.2771861890932559,
      "grad_norm": 1.1678391695022583,
      "learning_rate": 0.0005,
      "loss": 3.2024967670440674,
      "step": 1718
    },
    {
      "epoch": 0.27734753146176183,
      "grad_norm": 1.368882179260254,
      "learning_rate": 0.0005,
      "loss": 3.3956689834594727,
      "step": 1719
    },
    {
      "epoch": 0.27750887383026784,
      "grad_norm": 1.1993846893310547,
      "learning_rate": 0.0005,
      "loss": 4.090764999389648,
      "step": 1720
    },
    {
      "epoch": 0.2776702161987738,
      "grad_norm": 1.222820520401001,
      "learning_rate": 0.0005,
      "loss": 3.0065882205963135,
      "step": 1721
    },
    {
      "epoch": 0.27783155856727976,
      "grad_norm": 1.2772321701049805,
      "learning_rate": 0.0005,
      "loss": 3.4004945755004883,
      "step": 1722
    },
    {
      "epoch": 0.2779929009357857,
      "grad_norm": 1.3987234830856323,
      "learning_rate": 0.0005,
      "loss": 3.467820644378662,
      "step": 1723
    },
    {
      "epoch": 0.2781542433042917,
      "grad_norm": 1.462973952293396,
      "learning_rate": 0.0005,
      "loss": 3.431020736694336,
      "step": 1724
    },
    {
      "epoch": 0.2783155856727977,
      "grad_norm": 1.3487762212753296,
      "learning_rate": 0.0005,
      "loss": 3.289076805114746,
      "step": 1725
    },
    {
      "epoch": 0.27847692804130364,
      "grad_norm": 2.194991111755371,
      "learning_rate": 0.0005,
      "loss": 3.082163095474243,
      "step": 1726
    },
    {
      "epoch": 0.2786382704098096,
      "grad_norm": 2.1145477294921875,
      "learning_rate": 0.0005,
      "loss": 3.3321175575256348,
      "step": 1727
    },
    {
      "epoch": 0.2787996127783156,
      "grad_norm": 1.3719003200531006,
      "learning_rate": 0.0005,
      "loss": 3.2726502418518066,
      "step": 1728
    },
    {
      "epoch": 0.27896095514682157,
      "grad_norm": 1.4718132019042969,
      "learning_rate": 0.0005,
      "loss": 3.0055580139160156,
      "step": 1729
    },
    {
      "epoch": 0.2791222975153275,
      "grad_norm": 1.5826911926269531,
      "learning_rate": 0.0005,
      "loss": 3.0378570556640625,
      "step": 1730
    },
    {
      "epoch": 0.2792836398838335,
      "grad_norm": 1.3293660879135132,
      "learning_rate": 0.0005,
      "loss": 3.1459927558898926,
      "step": 1731
    },
    {
      "epoch": 0.2794449822523395,
      "grad_norm": 2.0264768600463867,
      "learning_rate": 0.0005,
      "loss": 3.213899612426758,
      "step": 1732
    },
    {
      "epoch": 0.27960632462084545,
      "grad_norm": 9.970885276794434,
      "learning_rate": 0.0005,
      "loss": 3.390965461730957,
      "step": 1733
    },
    {
      "epoch": 0.2797676669893514,
      "grad_norm": 1.8843863010406494,
      "learning_rate": 0.0005,
      "loss": 3.2614541053771973,
      "step": 1734
    },
    {
      "epoch": 0.27992900935785736,
      "grad_norm": 1.2183090448379517,
      "learning_rate": 0.0005,
      "loss": 2.7073307037353516,
      "step": 1735
    },
    {
      "epoch": 0.2800903517263633,
      "grad_norm": 1.3186222314834595,
      "learning_rate": 0.0005,
      "loss": 3.5593385696411133,
      "step": 1736
    },
    {
      "epoch": 0.28025169409486933,
      "grad_norm": 1.4297603368759155,
      "learning_rate": 0.0005,
      "loss": 3.3982577323913574,
      "step": 1737
    },
    {
      "epoch": 0.2804130364633753,
      "grad_norm": 1.1130813360214233,
      "learning_rate": 0.0005,
      "loss": 3.0729598999023438,
      "step": 1738
    },
    {
      "epoch": 0.28057437883188124,
      "grad_norm": 1.356188178062439,
      "learning_rate": 0.0005,
      "loss": 2.8837637901306152,
      "step": 1739
    },
    {
      "epoch": 0.2807357212003872,
      "grad_norm": 1.3115952014923096,
      "learning_rate": 0.0005,
      "loss": 3.117859125137329,
      "step": 1740
    },
    {
      "epoch": 0.2808970635688932,
      "grad_norm": 1.7377715110778809,
      "learning_rate": 0.0005,
      "loss": 3.2218496799468994,
      "step": 1741
    },
    {
      "epoch": 0.28105840593739917,
      "grad_norm": 1.6839492321014404,
      "learning_rate": 0.0005,
      "loss": 2.9061779975891113,
      "step": 1742
    },
    {
      "epoch": 0.2812197483059051,
      "grad_norm": 1.4828730821609497,
      "learning_rate": 0.0005,
      "loss": 3.1294476985931396,
      "step": 1743
    },
    {
      "epoch": 0.2813810906744111,
      "grad_norm": 1.6068451404571533,
      "learning_rate": 0.0005,
      "loss": 3.3092641830444336,
      "step": 1744
    },
    {
      "epoch": 0.2815424330429171,
      "grad_norm": 1.3454712629318237,
      "learning_rate": 0.0005,
      "loss": 3.390259265899658,
      "step": 1745
    },
    {
      "epoch": 0.28170377541142305,
      "grad_norm": 1.4476759433746338,
      "learning_rate": 0.0005,
      "loss": 3.2686686515808105,
      "step": 1746
    },
    {
      "epoch": 0.281865117779929,
      "grad_norm": 1.1562622785568237,
      "learning_rate": 0.0005,
      "loss": 3.363090991973877,
      "step": 1747
    },
    {
      "epoch": 0.28202646014843497,
      "grad_norm": 1.11978280544281,
      "learning_rate": 0.0005,
      "loss": 3.01322603225708,
      "step": 1748
    },
    {
      "epoch": 0.2821878025169409,
      "grad_norm": 1.4904158115386963,
      "learning_rate": 0.0005,
      "loss": 3.2363572120666504,
      "step": 1749
    },
    {
      "epoch": 0.28234914488544693,
      "grad_norm": 1.5849292278289795,
      "learning_rate": 0.0005,
      "loss": 3.0206336975097656,
      "step": 1750
    },
    {
      "epoch": 0.2825104872539529,
      "grad_norm": 1.2088490724563599,
      "learning_rate": 0.0005,
      "loss": 3.0136682987213135,
      "step": 1751
    },
    {
      "epoch": 0.28267182962245885,
      "grad_norm": 1.1113977432250977,
      "learning_rate": 0.0005,
      "loss": 3.463459014892578,
      "step": 1752
    },
    {
      "epoch": 0.2828331719909648,
      "grad_norm": 1.2932733297348022,
      "learning_rate": 0.0005,
      "loss": 3.3628745079040527,
      "step": 1753
    },
    {
      "epoch": 0.2829945143594708,
      "grad_norm": 1.3741222620010376,
      "learning_rate": 0.0005,
      "loss": 2.644103765487671,
      "step": 1754
    },
    {
      "epoch": 0.2831558567279768,
      "grad_norm": 1.2295154333114624,
      "learning_rate": 0.0005,
      "loss": 3.2363457679748535,
      "step": 1755
    },
    {
      "epoch": 0.28331719909648273,
      "grad_norm": 1.7252329587936401,
      "learning_rate": 0.0005,
      "loss": 3.574312448501587,
      "step": 1756
    },
    {
      "epoch": 0.2834785414649887,
      "grad_norm": 1.450014352798462,
      "learning_rate": 0.0005,
      "loss": 3.545048475265503,
      "step": 1757
    },
    {
      "epoch": 0.2836398838334947,
      "grad_norm": 1.466988444328308,
      "learning_rate": 0.0005,
      "loss": 3.464639663696289,
      "step": 1758
    },
    {
      "epoch": 0.28380122620200066,
      "grad_norm": 1.401928186416626,
      "learning_rate": 0.0005,
      "loss": 3.301589250564575,
      "step": 1759
    },
    {
      "epoch": 0.2839625685705066,
      "grad_norm": 1.2490131855010986,
      "learning_rate": 0.0005,
      "loss": 3.487262725830078,
      "step": 1760
    },
    {
      "epoch": 0.28412391093901257,
      "grad_norm": 1.2690715789794922,
      "learning_rate": 0.0005,
      "loss": 3.10622239112854,
      "step": 1761
    },
    {
      "epoch": 0.2842852533075186,
      "grad_norm": 1.3636445999145508,
      "learning_rate": 0.0005,
      "loss": 3.344472885131836,
      "step": 1762
    },
    {
      "epoch": 0.28444659567602454,
      "grad_norm": 2.367931842803955,
      "learning_rate": 0.0005,
      "loss": 3.1270813941955566,
      "step": 1763
    },
    {
      "epoch": 0.2846079380445305,
      "grad_norm": 1.3686606884002686,
      "learning_rate": 0.0005,
      "loss": 2.8615832328796387,
      "step": 1764
    },
    {
      "epoch": 0.28476928041303645,
      "grad_norm": 1.2716727256774902,
      "learning_rate": 0.0005,
      "loss": 2.593221664428711,
      "step": 1765
    },
    {
      "epoch": 0.2849306227815424,
      "grad_norm": 1.7852351665496826,
      "learning_rate": 0.0005,
      "loss": 3.063011884689331,
      "step": 1766
    },
    {
      "epoch": 0.2850919651500484,
      "grad_norm": 1.5325676202774048,
      "learning_rate": 0.0005,
      "loss": 3.4365644454956055,
      "step": 1767
    },
    {
      "epoch": 0.2852533075185544,
      "grad_norm": 1.2021340131759644,
      "learning_rate": 0.0005,
      "loss": 3.0895848274230957,
      "step": 1768
    },
    {
      "epoch": 0.28541464988706033,
      "grad_norm": 1.2623122930526733,
      "learning_rate": 0.0005,
      "loss": 3.0689797401428223,
      "step": 1769
    },
    {
      "epoch": 0.2855759922555663,
      "grad_norm": 1.1923588514328003,
      "learning_rate": 0.0005,
      "loss": 3.653045654296875,
      "step": 1770
    },
    {
      "epoch": 0.2857373346240723,
      "grad_norm": 1.2695127725601196,
      "learning_rate": 0.0005,
      "loss": 2.931250810623169,
      "step": 1771
    },
    {
      "epoch": 0.28589867699257826,
      "grad_norm": 1.2952057123184204,
      "learning_rate": 0.0005,
      "loss": 3.2761783599853516,
      "step": 1772
    },
    {
      "epoch": 0.2860600193610842,
      "grad_norm": 1.0938596725463867,
      "learning_rate": 0.0005,
      "loss": 2.788327693939209,
      "step": 1773
    },
    {
      "epoch": 0.2862213617295902,
      "grad_norm": 1.3174192905426025,
      "learning_rate": 0.0005,
      "loss": 3.594677448272705,
      "step": 1774
    },
    {
      "epoch": 0.2863827040980962,
      "grad_norm": 1.4964107275009155,
      "learning_rate": 0.0005,
      "loss": 3.1004788875579834,
      "step": 1775
    },
    {
      "epoch": 0.28654404646660214,
      "grad_norm": 1.2542916536331177,
      "learning_rate": 0.0005,
      "loss": 3.368093967437744,
      "step": 1776
    },
    {
      "epoch": 0.2867053888351081,
      "grad_norm": 1.4645711183547974,
      "learning_rate": 0.0005,
      "loss": 3.290027379989624,
      "step": 1777
    },
    {
      "epoch": 0.28686673120361406,
      "grad_norm": 1.6416008472442627,
      "learning_rate": 0.0005,
      "loss": 3.0757193565368652,
      "step": 1778
    },
    {
      "epoch": 0.28702807357212,
      "grad_norm": 1.7537628412246704,
      "learning_rate": 0.0005,
      "loss": 2.8896913528442383,
      "step": 1779
    },
    {
      "epoch": 0.287189415940626,
      "grad_norm": 1.2194457054138184,
      "learning_rate": 0.0005,
      "loss": 2.840693950653076,
      "step": 1780
    },
    {
      "epoch": 0.287350758309132,
      "grad_norm": 1.540880560874939,
      "learning_rate": 0.0005,
      "loss": 2.803785562515259,
      "step": 1781
    },
    {
      "epoch": 0.28751210067763794,
      "grad_norm": 1.458186149597168,
      "learning_rate": 0.0005,
      "loss": 2.843583822250366,
      "step": 1782
    },
    {
      "epoch": 0.2876734430461439,
      "grad_norm": 1.2660086154937744,
      "learning_rate": 0.0005,
      "loss": 3.467200756072998,
      "step": 1783
    },
    {
      "epoch": 0.2878347854146499,
      "grad_norm": 1.51308012008667,
      "learning_rate": 0.0005,
      "loss": 3.093240976333618,
      "step": 1784
    },
    {
      "epoch": 0.28799612778315586,
      "grad_norm": 1.6285613775253296,
      "learning_rate": 0.0005,
      "loss": 3.2404017448425293,
      "step": 1785
    },
    {
      "epoch": 0.2881574701516618,
      "grad_norm": 1.3013235330581665,
      "learning_rate": 0.0005,
      "loss": 3.3891892433166504,
      "step": 1786
    },
    {
      "epoch": 0.2883188125201678,
      "grad_norm": 1.348549723625183,
      "learning_rate": 0.0005,
      "loss": 3.5159192085266113,
      "step": 1787
    },
    {
      "epoch": 0.2884801548886738,
      "grad_norm": 1.2189679145812988,
      "learning_rate": 0.0005,
      "loss": 3.263371706008911,
      "step": 1788
    },
    {
      "epoch": 0.28864149725717975,
      "grad_norm": 1.3662558794021606,
      "learning_rate": 0.0005,
      "loss": 3.7745137214660645,
      "step": 1789
    },
    {
      "epoch": 0.2888028396256857,
      "grad_norm": 1.4164711236953735,
      "learning_rate": 0.0005,
      "loss": 3.0800669193267822,
      "step": 1790
    },
    {
      "epoch": 0.28896418199419166,
      "grad_norm": 1.4388400316238403,
      "learning_rate": 0.0005,
      "loss": 3.1792054176330566,
      "step": 1791
    },
    {
      "epoch": 0.2891255243626976,
      "grad_norm": 3.589033842086792,
      "learning_rate": 0.0005,
      "loss": 2.9884328842163086,
      "step": 1792
    },
    {
      "epoch": 0.28928686673120363,
      "grad_norm": 1.1580810546875,
      "learning_rate": 0.0005,
      "loss": 2.963210105895996,
      "step": 1793
    },
    {
      "epoch": 0.2894482090997096,
      "grad_norm": 1.349265217781067,
      "learning_rate": 0.0005,
      "loss": 3.2295548915863037,
      "step": 1794
    },
    {
      "epoch": 0.28960955146821554,
      "grad_norm": 1.8494012355804443,
      "learning_rate": 0.0005,
      "loss": 3.340151786804199,
      "step": 1795
    },
    {
      "epoch": 0.2897708938367215,
      "grad_norm": 1.2651612758636475,
      "learning_rate": 0.0005,
      "loss": 2.8547821044921875,
      "step": 1796
    },
    {
      "epoch": 0.2899322362052275,
      "grad_norm": 1.2597126960754395,
      "learning_rate": 0.0005,
      "loss": 3.499425172805786,
      "step": 1797
    },
    {
      "epoch": 0.29009357857373347,
      "grad_norm": 1.2966203689575195,
      "learning_rate": 0.0005,
      "loss": 3.243290424346924,
      "step": 1798
    },
    {
      "epoch": 0.2902549209422394,
      "grad_norm": 1.5663657188415527,
      "learning_rate": 0.0005,
      "loss": 3.2387797832489014,
      "step": 1799
    },
    {
      "epoch": 0.2904162633107454,
      "grad_norm": 1.8739047050476074,
      "learning_rate": 0.0005,
      "loss": 3.1831893920898438,
      "step": 1800
    },
    {
      "epoch": 0.2905776056792514,
      "grad_norm": 1.255643367767334,
      "learning_rate": 0.0005,
      "loss": 3.5833306312561035,
      "step": 1801
    },
    {
      "epoch": 0.29073894804775735,
      "grad_norm": 1.160396933555603,
      "learning_rate": 0.0005,
      "loss": 3.185368537902832,
      "step": 1802
    },
    {
      "epoch": 0.2909002904162633,
      "grad_norm": 1.395114779472351,
      "learning_rate": 0.0005,
      "loss": 2.878486156463623,
      "step": 1803
    },
    {
      "epoch": 0.29106163278476926,
      "grad_norm": 1.5755478143692017,
      "learning_rate": 0.0005,
      "loss": 3.0755367279052734,
      "step": 1804
    },
    {
      "epoch": 0.2912229751532753,
      "grad_norm": 1.3133001327514648,
      "learning_rate": 0.0005,
      "loss": 3.381883382797241,
      "step": 1805
    },
    {
      "epoch": 0.29138431752178123,
      "grad_norm": 1.3535622358322144,
      "learning_rate": 0.0005,
      "loss": 2.9614782333374023,
      "step": 1806
    },
    {
      "epoch": 0.2915456598902872,
      "grad_norm": 1.2834088802337646,
      "learning_rate": 0.0005,
      "loss": 3.296539545059204,
      "step": 1807
    },
    {
      "epoch": 0.29170700225879315,
      "grad_norm": 1.4855986833572388,
      "learning_rate": 0.0005,
      "loss": 3.025813579559326,
      "step": 1808
    },
    {
      "epoch": 0.2918683446272991,
      "grad_norm": 1.0959458351135254,
      "learning_rate": 0.0005,
      "loss": 2.95988130569458,
      "step": 1809
    },
    {
      "epoch": 0.2920296869958051,
      "grad_norm": 3.9662160873413086,
      "learning_rate": 0.0005,
      "loss": 3.602097988128662,
      "step": 1810
    },
    {
      "epoch": 0.2921910293643111,
      "grad_norm": 1.966443419456482,
      "learning_rate": 0.0005,
      "loss": 4.184871196746826,
      "step": 1811
    },
    {
      "epoch": 0.29235237173281703,
      "grad_norm": 1.1878368854522705,
      "learning_rate": 0.0005,
      "loss": 2.861825942993164,
      "step": 1812
    },
    {
      "epoch": 0.292513714101323,
      "grad_norm": 1.639655590057373,
      "learning_rate": 0.0005,
      "loss": 3.2271833419799805,
      "step": 1813
    },
    {
      "epoch": 0.292675056469829,
      "grad_norm": 1.268957257270813,
      "learning_rate": 0.0005,
      "loss": 2.8615169525146484,
      "step": 1814
    },
    {
      "epoch": 0.29283639883833495,
      "grad_norm": 1.2611298561096191,
      "learning_rate": 0.0005,
      "loss": 3.1263418197631836,
      "step": 1815
    },
    {
      "epoch": 0.2929977412068409,
      "grad_norm": 1.4700592756271362,
      "learning_rate": 0.0005,
      "loss": 3.086972236633301,
      "step": 1816
    },
    {
      "epoch": 0.29315908357534687,
      "grad_norm": 1.5894207954406738,
      "learning_rate": 0.0005,
      "loss": 3.016618490219116,
      "step": 1817
    },
    {
      "epoch": 0.2933204259438529,
      "grad_norm": 1.5683356523513794,
      "learning_rate": 0.0005,
      "loss": 3.043355703353882,
      "step": 1818
    },
    {
      "epoch": 0.29348176831235884,
      "grad_norm": 1.7983478307724,
      "learning_rate": 0.0005,
      "loss": 2.7352659702301025,
      "step": 1819
    },
    {
      "epoch": 0.2936431106808648,
      "grad_norm": 1.5878549814224243,
      "learning_rate": 0.0005,
      "loss": 3.0712966918945312,
      "step": 1820
    },
    {
      "epoch": 0.29380445304937075,
      "grad_norm": 1.7268106937408447,
      "learning_rate": 0.0005,
      "loss": 3.137742280960083,
      "step": 1821
    },
    {
      "epoch": 0.2939657954178767,
      "grad_norm": 1.557915210723877,
      "learning_rate": 0.0005,
      "loss": 3.202962875366211,
      "step": 1822
    },
    {
      "epoch": 0.2941271377863827,
      "grad_norm": 1.3534806966781616,
      "learning_rate": 0.0005,
      "loss": 2.9966378211975098,
      "step": 1823
    },
    {
      "epoch": 0.2942884801548887,
      "grad_norm": 1.6992299556732178,
      "learning_rate": 0.0005,
      "loss": 3.0792388916015625,
      "step": 1824
    },
    {
      "epoch": 0.29444982252339463,
      "grad_norm": 1.391797661781311,
      "learning_rate": 0.0005,
      "loss": 3.338019371032715,
      "step": 1825
    },
    {
      "epoch": 0.2946111648919006,
      "grad_norm": 1.2975831031799316,
      "learning_rate": 0.0005,
      "loss": 2.8376035690307617,
      "step": 1826
    },
    {
      "epoch": 0.2947725072604066,
      "grad_norm": 1.3454097509384155,
      "learning_rate": 0.0005,
      "loss": 3.1389050483703613,
      "step": 1827
    },
    {
      "epoch": 0.29493384962891256,
      "grad_norm": 1.1338152885437012,
      "learning_rate": 0.0005,
      "loss": 3.247074604034424,
      "step": 1828
    },
    {
      "epoch": 0.2950951919974185,
      "grad_norm": 1.655468225479126,
      "learning_rate": 0.0005,
      "loss": 2.941814422607422,
      "step": 1829
    },
    {
      "epoch": 0.2952565343659245,
      "grad_norm": 1.6804240942001343,
      "learning_rate": 0.0005,
      "loss": 3.0049004554748535,
      "step": 1830
    },
    {
      "epoch": 0.2954178767344305,
      "grad_norm": 1.2365602254867554,
      "learning_rate": 0.0005,
      "loss": 3.2372965812683105,
      "step": 1831
    },
    {
      "epoch": 0.29557921910293644,
      "grad_norm": 1.967453122138977,
      "learning_rate": 0.0005,
      "loss": 3.118410587310791,
      "step": 1832
    },
    {
      "epoch": 0.2957405614714424,
      "grad_norm": 1.4192774295806885,
      "learning_rate": 0.0005,
      "loss": 2.716758966445923,
      "step": 1833
    },
    {
      "epoch": 0.29590190383994835,
      "grad_norm": 1.4390705823898315,
      "learning_rate": 0.0005,
      "loss": 3.1400108337402344,
      "step": 1834
    },
    {
      "epoch": 0.29606324620845437,
      "grad_norm": 1.3827846050262451,
      "learning_rate": 0.0005,
      "loss": 3.226874351501465,
      "step": 1835
    },
    {
      "epoch": 0.2962245885769603,
      "grad_norm": 1.3506726026535034,
      "learning_rate": 0.0005,
      "loss": 3.2083539962768555,
      "step": 1836
    },
    {
      "epoch": 0.2963859309454663,
      "grad_norm": 2.804365396499634,
      "learning_rate": 0.0005,
      "loss": 3.1404056549072266,
      "step": 1837
    },
    {
      "epoch": 0.29654727331397224,
      "grad_norm": 1.5117645263671875,
      "learning_rate": 0.0005,
      "loss": 2.9717469215393066,
      "step": 1838
    },
    {
      "epoch": 0.2967086156824782,
      "grad_norm": 1.947782039642334,
      "learning_rate": 0.0005,
      "loss": 3.0270743370056152,
      "step": 1839
    },
    {
      "epoch": 0.2968699580509842,
      "grad_norm": 1.4561156034469604,
      "learning_rate": 0.0005,
      "loss": 3.051774501800537,
      "step": 1840
    },
    {
      "epoch": 0.29703130041949016,
      "grad_norm": 1.531156301498413,
      "learning_rate": 0.0005,
      "loss": 3.066838264465332,
      "step": 1841
    },
    {
      "epoch": 0.2971926427879961,
      "grad_norm": 1.373503565788269,
      "learning_rate": 0.0005,
      "loss": 3.2309558391571045,
      "step": 1842
    },
    {
      "epoch": 0.2973539851565021,
      "grad_norm": 1.3158272504806519,
      "learning_rate": 0.0005,
      "loss": 2.9983253479003906,
      "step": 1843
    },
    {
      "epoch": 0.2975153275250081,
      "grad_norm": 1.4942419528961182,
      "learning_rate": 0.0005,
      "loss": 3.49928617477417,
      "step": 1844
    },
    {
      "epoch": 0.29767666989351405,
      "grad_norm": 1.7960286140441895,
      "learning_rate": 0.0005,
      "loss": 3.4126296043395996,
      "step": 1845
    },
    {
      "epoch": 0.29783801226202,
      "grad_norm": 1.4493337869644165,
      "learning_rate": 0.0005,
      "loss": 3.280790328979492,
      "step": 1846
    },
    {
      "epoch": 0.29799935463052596,
      "grad_norm": 1.619758129119873,
      "learning_rate": 0.0005,
      "loss": 3.243330240249634,
      "step": 1847
    },
    {
      "epoch": 0.29816069699903197,
      "grad_norm": 1.576988697052002,
      "learning_rate": 0.0005,
      "loss": 2.942803144454956,
      "step": 1848
    },
    {
      "epoch": 0.29832203936753793,
      "grad_norm": 1.3208377361297607,
      "learning_rate": 0.0005,
      "loss": 3.402475595474243,
      "step": 1849
    },
    {
      "epoch": 0.2984833817360439,
      "grad_norm": 1.2072128057479858,
      "learning_rate": 0.0005,
      "loss": 2.8395042419433594,
      "step": 1850
    },
    {
      "epoch": 0.29864472410454984,
      "grad_norm": 1.2782554626464844,
      "learning_rate": 0.0005,
      "loss": 3.5336227416992188,
      "step": 1851
    },
    {
      "epoch": 0.2988060664730558,
      "grad_norm": 1.4492113590240479,
      "learning_rate": 0.0005,
      "loss": 3.2681400775909424,
      "step": 1852
    },
    {
      "epoch": 0.2989674088415618,
      "grad_norm": 1.3384236097335815,
      "learning_rate": 0.0005,
      "loss": 2.9598326683044434,
      "step": 1853
    },
    {
      "epoch": 0.29912875121006777,
      "grad_norm": 1.4182696342468262,
      "learning_rate": 0.0005,
      "loss": 3.136622428894043,
      "step": 1854
    },
    {
      "epoch": 0.2992900935785737,
      "grad_norm": 1.584389090538025,
      "learning_rate": 0.0005,
      "loss": 3.971221923828125,
      "step": 1855
    },
    {
      "epoch": 0.2994514359470797,
      "grad_norm": 1.4585137367248535,
      "learning_rate": 0.0005,
      "loss": 3.3038690090179443,
      "step": 1856
    },
    {
      "epoch": 0.2996127783155857,
      "grad_norm": 1.066211223602295,
      "learning_rate": 0.0005,
      "loss": 3.7339296340942383,
      "step": 1857
    },
    {
      "epoch": 0.29977412068409165,
      "grad_norm": 1.2061095237731934,
      "learning_rate": 0.0005,
      "loss": 3.031564235687256,
      "step": 1858
    },
    {
      "epoch": 0.2999354630525976,
      "grad_norm": 2.07033371925354,
      "learning_rate": 0.0005,
      "loss": 2.672278881072998,
      "step": 1859
    },
    {
      "epoch": 0.30009680542110356,
      "grad_norm": 1.2894423007965088,
      "learning_rate": 0.0005,
      "loss": 3.0173587799072266,
      "step": 1860
    },
    {
      "epoch": 0.3002581477896096,
      "grad_norm": 1.2389951944351196,
      "learning_rate": 0.0005,
      "loss": 3.353351593017578,
      "step": 1861
    },
    {
      "epoch": 0.30041949015811553,
      "grad_norm": 1.3025031089782715,
      "learning_rate": 0.0005,
      "loss": 2.662031650543213,
      "step": 1862
    },
    {
      "epoch": 0.3005808325266215,
      "grad_norm": 1.1370468139648438,
      "learning_rate": 0.0005,
      "loss": 3.0649142265319824,
      "step": 1863
    },
    {
      "epoch": 0.30074217489512745,
      "grad_norm": 1.164680004119873,
      "learning_rate": 0.0005,
      "loss": 3.293778657913208,
      "step": 1864
    },
    {
      "epoch": 0.30090351726363346,
      "grad_norm": 1.3350653648376465,
      "learning_rate": 0.0005,
      "loss": 3.6213865280151367,
      "step": 1865
    },
    {
      "epoch": 0.3010648596321394,
      "grad_norm": 1.4028363227844238,
      "learning_rate": 0.0005,
      "loss": 3.0802125930786133,
      "step": 1866
    },
    {
      "epoch": 0.30122620200064537,
      "grad_norm": 1.7853282690048218,
      "learning_rate": 0.0005,
      "loss": 3.071390390396118,
      "step": 1867
    },
    {
      "epoch": 0.30138754436915133,
      "grad_norm": 1.3089382648468018,
      "learning_rate": 0.0005,
      "loss": 2.596281051635742,
      "step": 1868
    },
    {
      "epoch": 0.3015488867376573,
      "grad_norm": 1.165877103805542,
      "learning_rate": 0.0005,
      "loss": 2.9312233924865723,
      "step": 1869
    },
    {
      "epoch": 0.3017102291061633,
      "grad_norm": 1.4610114097595215,
      "learning_rate": 0.0005,
      "loss": 3.0262069702148438,
      "step": 1870
    },
    {
      "epoch": 0.30187157147466925,
      "grad_norm": 1.2449337244033813,
      "learning_rate": 0.0005,
      "loss": 3.164407730102539,
      "step": 1871
    },
    {
      "epoch": 0.3020329138431752,
      "grad_norm": 1.1693706512451172,
      "learning_rate": 0.0005,
      "loss": 2.515852212905884,
      "step": 1872
    },
    {
      "epoch": 0.30219425621168117,
      "grad_norm": 1.7240347862243652,
      "learning_rate": 0.0005,
      "loss": 3.211610794067383,
      "step": 1873
    },
    {
      "epoch": 0.3023555985801872,
      "grad_norm": 1.7415540218353271,
      "learning_rate": 0.0005,
      "loss": 3.0020859241485596,
      "step": 1874
    },
    {
      "epoch": 0.30251694094869314,
      "grad_norm": 1.9668593406677246,
      "learning_rate": 0.0005,
      "loss": 3.279207229614258,
      "step": 1875
    },
    {
      "epoch": 0.3026782833171991,
      "grad_norm": 1.8079360723495483,
      "learning_rate": 0.0005,
      "loss": 2.7628235816955566,
      "step": 1876
    },
    {
      "epoch": 0.30283962568570505,
      "grad_norm": 1.458528995513916,
      "learning_rate": 0.0005,
      "loss": 2.9875216484069824,
      "step": 1877
    },
    {
      "epoch": 0.30300096805421106,
      "grad_norm": 1.58708918094635,
      "learning_rate": 0.0005,
      "loss": 3.1313071250915527,
      "step": 1878
    },
    {
      "epoch": 0.303162310422717,
      "grad_norm": 1.6227216720581055,
      "learning_rate": 0.0005,
      "loss": 2.8089985847473145,
      "step": 1879
    },
    {
      "epoch": 0.303323652791223,
      "grad_norm": 1.170641303062439,
      "learning_rate": 0.0005,
      "loss": 3.452130079269409,
      "step": 1880
    },
    {
      "epoch": 0.30348499515972893,
      "grad_norm": 1.6768302917480469,
      "learning_rate": 0.0005,
      "loss": 3.2834620475769043,
      "step": 1881
    },
    {
      "epoch": 0.3036463375282349,
      "grad_norm": 1.3482022285461426,
      "learning_rate": 0.0005,
      "loss": 3.590498685836792,
      "step": 1882
    },
    {
      "epoch": 0.3038076798967409,
      "grad_norm": 1.1605921983718872,
      "learning_rate": 0.0005,
      "loss": 3.348991870880127,
      "step": 1883
    },
    {
      "epoch": 0.30396902226524686,
      "grad_norm": 1.6856955289840698,
      "learning_rate": 0.0005,
      "loss": 3.273533344268799,
      "step": 1884
    },
    {
      "epoch": 0.3041303646337528,
      "grad_norm": 1.1910202503204346,
      "learning_rate": 0.0005,
      "loss": 3.144501209259033,
      "step": 1885
    },
    {
      "epoch": 0.30429170700225877,
      "grad_norm": 1.5814967155456543,
      "learning_rate": 0.0005,
      "loss": 3.011356830596924,
      "step": 1886
    },
    {
      "epoch": 0.3044530493707648,
      "grad_norm": 1.511114478111267,
      "learning_rate": 0.0005,
      "loss": 3.3366594314575195,
      "step": 1887
    },
    {
      "epoch": 0.30461439173927074,
      "grad_norm": 1.188210368156433,
      "learning_rate": 0.0005,
      "loss": 3.03910756111145,
      "step": 1888
    },
    {
      "epoch": 0.3047757341077767,
      "grad_norm": 1.087332010269165,
      "learning_rate": 0.0005,
      "loss": 3.4382033348083496,
      "step": 1889
    },
    {
      "epoch": 0.30493707647628265,
      "grad_norm": 1.7330272197723389,
      "learning_rate": 0.0005,
      "loss": 3.214137554168701,
      "step": 1890
    },
    {
      "epoch": 0.30509841884478867,
      "grad_norm": 1.5223848819732666,
      "learning_rate": 0.0005,
      "loss": 3.412118911743164,
      "step": 1891
    },
    {
      "epoch": 0.3052597612132946,
      "grad_norm": 1.3137800693511963,
      "learning_rate": 0.0005,
      "loss": 3.6168742179870605,
      "step": 1892
    },
    {
      "epoch": 0.3054211035818006,
      "grad_norm": 1.26141357421875,
      "learning_rate": 0.0005,
      "loss": 2.850353479385376,
      "step": 1893
    },
    {
      "epoch": 0.30558244595030654,
      "grad_norm": 1.4844084978103638,
      "learning_rate": 0.0005,
      "loss": 3.242281913757324,
      "step": 1894
    },
    {
      "epoch": 0.3057437883188125,
      "grad_norm": 1.2682050466537476,
      "learning_rate": 0.0005,
      "loss": 2.6868691444396973,
      "step": 1895
    },
    {
      "epoch": 0.3059051306873185,
      "grad_norm": 1.5640400648117065,
      "learning_rate": 0.0005,
      "loss": 3.328319787979126,
      "step": 1896
    },
    {
      "epoch": 0.30606647305582446,
      "grad_norm": 1.3414385318756104,
      "learning_rate": 0.0005,
      "loss": 3.0416982173919678,
      "step": 1897
    },
    {
      "epoch": 0.3062278154243304,
      "grad_norm": 1.4473884105682373,
      "learning_rate": 0.0005,
      "loss": 3.283522605895996,
      "step": 1898
    },
    {
      "epoch": 0.3063891577928364,
      "grad_norm": 1.2215648889541626,
      "learning_rate": 0.0005,
      "loss": 2.8699076175689697,
      "step": 1899
    },
    {
      "epoch": 0.3065505001613424,
      "grad_norm": 1.5470738410949707,
      "learning_rate": 0.0005,
      "loss": 2.9712281227111816,
      "step": 1900
    },
    {
      "epoch": 0.30671184252984834,
      "grad_norm": 1.3633294105529785,
      "learning_rate": 0.0005,
      "loss": 3.11820912361145,
      "step": 1901
    },
    {
      "epoch": 0.3068731848983543,
      "grad_norm": 1.6859815120697021,
      "learning_rate": 0.0005,
      "loss": 3.4378905296325684,
      "step": 1902
    },
    {
      "epoch": 0.30703452726686026,
      "grad_norm": 1.7894104719161987,
      "learning_rate": 0.0005,
      "loss": 3.258516550064087,
      "step": 1903
    },
    {
      "epoch": 0.30719586963536627,
      "grad_norm": 1.8376177549362183,
      "learning_rate": 0.0005,
      "loss": 2.73738431930542,
      "step": 1904
    },
    {
      "epoch": 0.3073572120038722,
      "grad_norm": 1.1083998680114746,
      "learning_rate": 0.0005,
      "loss": 3.188051700592041,
      "step": 1905
    },
    {
      "epoch": 0.3075185543723782,
      "grad_norm": 1.8945673704147339,
      "learning_rate": 0.0005,
      "loss": 2.875523567199707,
      "step": 1906
    },
    {
      "epoch": 0.30767989674088414,
      "grad_norm": 2.6326189041137695,
      "learning_rate": 0.0005,
      "loss": 3.0354251861572266,
      "step": 1907
    },
    {
      "epoch": 0.30784123910939015,
      "grad_norm": 1.5075864791870117,
      "learning_rate": 0.0005,
      "loss": 2.931920051574707,
      "step": 1908
    },
    {
      "epoch": 0.3080025814778961,
      "grad_norm": 1.2565399408340454,
      "learning_rate": 0.0005,
      "loss": 2.6094980239868164,
      "step": 1909
    },
    {
      "epoch": 0.30816392384640207,
      "grad_norm": 2.1851930618286133,
      "learning_rate": 0.0005,
      "loss": 3.13352632522583,
      "step": 1910
    },
    {
      "epoch": 0.308325266214908,
      "grad_norm": 1.4060555696487427,
      "learning_rate": 0.0005,
      "loss": 2.948291778564453,
      "step": 1911
    },
    {
      "epoch": 0.308486608583414,
      "grad_norm": 2.0316431522369385,
      "learning_rate": 0.0005,
      "loss": 3.139497756958008,
      "step": 1912
    },
    {
      "epoch": 0.30864795095192,
      "grad_norm": 2.571460247039795,
      "learning_rate": 0.0005,
      "loss": 2.7321460247039795,
      "step": 1913
    },
    {
      "epoch": 0.30880929332042595,
      "grad_norm": 1.7576658725738525,
      "learning_rate": 0.0005,
      "loss": 3.3942995071411133,
      "step": 1914
    },
    {
      "epoch": 0.3089706356889319,
      "grad_norm": 1.3204593658447266,
      "learning_rate": 0.0005,
      "loss": 3.2182602882385254,
      "step": 1915
    },
    {
      "epoch": 0.30913197805743786,
      "grad_norm": 1.5416325330734253,
      "learning_rate": 0.0005,
      "loss": 3.2497458457946777,
      "step": 1916
    },
    {
      "epoch": 0.3092933204259439,
      "grad_norm": 1.4940810203552246,
      "learning_rate": 0.0005,
      "loss": 3.2618675231933594,
      "step": 1917
    },
    {
      "epoch": 0.30945466279444983,
      "grad_norm": 1.4516431093215942,
      "learning_rate": 0.0005,
      "loss": 2.8403127193450928,
      "step": 1918
    },
    {
      "epoch": 0.3096160051629558,
      "grad_norm": 1.5391850471496582,
      "learning_rate": 0.0005,
      "loss": 2.9383647441864014,
      "step": 1919
    },
    {
      "epoch": 0.30977734753146174,
      "grad_norm": 1.4468859434127808,
      "learning_rate": 0.0005,
      "loss": 2.9436347484588623,
      "step": 1920
    },
    {
      "epoch": 0.30993868989996776,
      "grad_norm": 1.5109468698501587,
      "learning_rate": 0.0005,
      "loss": 2.957404136657715,
      "step": 1921
    },
    {
      "epoch": 0.3101000322684737,
      "grad_norm": 1.349501132965088,
      "learning_rate": 0.0005,
      "loss": 2.7066707611083984,
      "step": 1922
    },
    {
      "epoch": 0.31026137463697967,
      "grad_norm": 1.1101231575012207,
      "learning_rate": 0.0005,
      "loss": 3.2107882499694824,
      "step": 1923
    },
    {
      "epoch": 0.3104227170054856,
      "grad_norm": 1.4343684911727905,
      "learning_rate": 0.0005,
      "loss": 2.792816162109375,
      "step": 1924
    },
    {
      "epoch": 0.3105840593739916,
      "grad_norm": 1.3048944473266602,
      "learning_rate": 0.0005,
      "loss": 2.664513349533081,
      "step": 1925
    },
    {
      "epoch": 0.3107454017424976,
      "grad_norm": 1.9038093090057373,
      "learning_rate": 0.0005,
      "loss": 3.135131597518921,
      "step": 1926
    },
    {
      "epoch": 0.31090674411100355,
      "grad_norm": 1.4402709007263184,
      "learning_rate": 0.0005,
      "loss": 3.1860785484313965,
      "step": 1927
    },
    {
      "epoch": 0.3110680864795095,
      "grad_norm": 1.3068803548812866,
      "learning_rate": 0.0005,
      "loss": 2.9238381385803223,
      "step": 1928
    },
    {
      "epoch": 0.31122942884801547,
      "grad_norm": 1.2939971685409546,
      "learning_rate": 0.0005,
      "loss": 3.036252498626709,
      "step": 1929
    },
    {
      "epoch": 0.3113907712165215,
      "grad_norm": 1.4327774047851562,
      "learning_rate": 0.0005,
      "loss": 2.9869699478149414,
      "step": 1930
    },
    {
      "epoch": 0.31155211358502743,
      "grad_norm": 1.6590627431869507,
      "learning_rate": 0.0005,
      "loss": 3.0778238773345947,
      "step": 1931
    },
    {
      "epoch": 0.3117134559535334,
      "grad_norm": 1.7450278997421265,
      "learning_rate": 0.0005,
      "loss": 3.361980438232422,
      "step": 1932
    },
    {
      "epoch": 0.31187479832203935,
      "grad_norm": 1.4185757637023926,
      "learning_rate": 0.0005,
      "loss": 3.118821859359741,
      "step": 1933
    },
    {
      "epoch": 0.31203614069054536,
      "grad_norm": 1.4359557628631592,
      "learning_rate": 0.0005,
      "loss": 3.191035747528076,
      "step": 1934
    },
    {
      "epoch": 0.3121974830590513,
      "grad_norm": 1.5248194932937622,
      "learning_rate": 0.0005,
      "loss": 3.5399484634399414,
      "step": 1935
    },
    {
      "epoch": 0.3123588254275573,
      "grad_norm": 1.22724449634552,
      "learning_rate": 0.0005,
      "loss": 3.2615647315979004,
      "step": 1936
    },
    {
      "epoch": 0.31252016779606323,
      "grad_norm": 2.8197007179260254,
      "learning_rate": 0.0005,
      "loss": 3.073975086212158,
      "step": 1937
    },
    {
      "epoch": 0.31268151016456924,
      "grad_norm": 1.8084739446640015,
      "learning_rate": 0.0005,
      "loss": 2.8980932235717773,
      "step": 1938
    },
    {
      "epoch": 0.3128428525330752,
      "grad_norm": 1.3955751657485962,
      "learning_rate": 0.0005,
      "loss": 2.8350841999053955,
      "step": 1939
    },
    {
      "epoch": 0.31300419490158116,
      "grad_norm": 1.7852344512939453,
      "learning_rate": 0.0005,
      "loss": 3.34432053565979,
      "step": 1940
    },
    {
      "epoch": 0.3131655372700871,
      "grad_norm": 1.165114402770996,
      "learning_rate": 0.0005,
      "loss": 3.1980783939361572,
      "step": 1941
    },
    {
      "epoch": 0.31332687963859307,
      "grad_norm": 1.5971204042434692,
      "learning_rate": 0.0005,
      "loss": 3.7342801094055176,
      "step": 1942
    },
    {
      "epoch": 0.3134882220070991,
      "grad_norm": 1.8147262334823608,
      "learning_rate": 0.0005,
      "loss": 3.713585376739502,
      "step": 1943
    },
    {
      "epoch": 0.31364956437560504,
      "grad_norm": 1.3533284664154053,
      "learning_rate": 0.0005,
      "loss": 2.988541841506958,
      "step": 1944
    },
    {
      "epoch": 0.313810906744111,
      "grad_norm": 1.4972258806228638,
      "learning_rate": 0.0005,
      "loss": 3.4763169288635254,
      "step": 1945
    },
    {
      "epoch": 0.31397224911261695,
      "grad_norm": 1.4121699333190918,
      "learning_rate": 0.0005,
      "loss": 3.466329574584961,
      "step": 1946
    },
    {
      "epoch": 0.31413359148112296,
      "grad_norm": 1.3232735395431519,
      "learning_rate": 0.0005,
      "loss": 3.2338790893554688,
      "step": 1947
    },
    {
      "epoch": 0.3142949338496289,
      "grad_norm": 1.3468751907348633,
      "learning_rate": 0.0005,
      "loss": 3.174001693725586,
      "step": 1948
    },
    {
      "epoch": 0.3144562762181349,
      "grad_norm": 1.0624512434005737,
      "learning_rate": 0.0005,
      "loss": 3.6509037017822266,
      "step": 1949
    },
    {
      "epoch": 0.31461761858664083,
      "grad_norm": 5.514957904815674,
      "learning_rate": 0.0005,
      "loss": 3.7338175773620605,
      "step": 1950
    },
    {
      "epoch": 0.31477896095514685,
      "grad_norm": 1.2486640214920044,
      "learning_rate": 0.0005,
      "loss": 2.8417835235595703,
      "step": 1951
    },
    {
      "epoch": 0.3149403033236528,
      "grad_norm": 1.2283551692962646,
      "learning_rate": 0.0005,
      "loss": 3.935474395751953,
      "step": 1952
    },
    {
      "epoch": 0.31510164569215876,
      "grad_norm": 1.3404017686843872,
      "learning_rate": 0.0005,
      "loss": 2.8967678546905518,
      "step": 1953
    },
    {
      "epoch": 0.3152629880606647,
      "grad_norm": 2.4460651874542236,
      "learning_rate": 0.0005,
      "loss": 2.90714693069458,
      "step": 1954
    },
    {
      "epoch": 0.3154243304291707,
      "grad_norm": 1.689305067062378,
      "learning_rate": 0.0005,
      "loss": 2.935318946838379,
      "step": 1955
    },
    {
      "epoch": 0.3155856727976767,
      "grad_norm": 1.7905874252319336,
      "learning_rate": 0.0005,
      "loss": 2.809889554977417,
      "step": 1956
    },
    {
      "epoch": 0.31574701516618264,
      "grad_norm": 1.4084447622299194,
      "learning_rate": 0.0005,
      "loss": 3.0347297191619873,
      "step": 1957
    },
    {
      "epoch": 0.3159083575346886,
      "grad_norm": 1.3045508861541748,
      "learning_rate": 0.0005,
      "loss": 3.0375075340270996,
      "step": 1958
    },
    {
      "epoch": 0.31606969990319456,
      "grad_norm": 1.8142588138580322,
      "learning_rate": 0.0005,
      "loss": 3.0664215087890625,
      "step": 1959
    },
    {
      "epoch": 0.31623104227170057,
      "grad_norm": 1.6433422565460205,
      "learning_rate": 0.0005,
      "loss": 3.1838512420654297,
      "step": 1960
    },
    {
      "epoch": 0.3163923846402065,
      "grad_norm": 1.2892982959747314,
      "learning_rate": 0.0005,
      "loss": 3.334146499633789,
      "step": 1961
    },
    {
      "epoch": 0.3165537270087125,
      "grad_norm": 1.382015585899353,
      "learning_rate": 0.0005,
      "loss": 3.106032133102417,
      "step": 1962
    },
    {
      "epoch": 0.31671506937721844,
      "grad_norm": 1.5268372297286987,
      "learning_rate": 0.0005,
      "loss": 3.466386556625366,
      "step": 1963
    },
    {
      "epoch": 0.31687641174572445,
      "grad_norm": 1.6101863384246826,
      "learning_rate": 0.0005,
      "loss": 3.221073627471924,
      "step": 1964
    },
    {
      "epoch": 0.3170377541142304,
      "grad_norm": 1.3996964693069458,
      "learning_rate": 0.0005,
      "loss": 3.3331212997436523,
      "step": 1965
    },
    {
      "epoch": 0.31719909648273636,
      "grad_norm": 1.3811490535736084,
      "learning_rate": 0.0005,
      "loss": 2.7654056549072266,
      "step": 1966
    },
    {
      "epoch": 0.3173604388512423,
      "grad_norm": 1.4512428045272827,
      "learning_rate": 0.0005,
      "loss": 3.5706992149353027,
      "step": 1967
    },
    {
      "epoch": 0.31752178121974833,
      "grad_norm": 1.196157693862915,
      "learning_rate": 0.0005,
      "loss": 2.906266212463379,
      "step": 1968
    },
    {
      "epoch": 0.3176831235882543,
      "grad_norm": 1.4007840156555176,
      "learning_rate": 0.0005,
      "loss": 3.214184045791626,
      "step": 1969
    },
    {
      "epoch": 0.31784446595676025,
      "grad_norm": 1.5570459365844727,
      "learning_rate": 0.0005,
      "loss": 3.1889219284057617,
      "step": 1970
    },
    {
      "epoch": 0.3180058083252662,
      "grad_norm": 1.3277584314346313,
      "learning_rate": 0.0005,
      "loss": 3.124925374984741,
      "step": 1971
    },
    {
      "epoch": 0.31816715069377216,
      "grad_norm": 1.4049128293991089,
      "learning_rate": 0.0005,
      "loss": 3.292827606201172,
      "step": 1972
    },
    {
      "epoch": 0.3183284930622782,
      "grad_norm": 1.2046384811401367,
      "learning_rate": 0.0005,
      "loss": 2.917365550994873,
      "step": 1973
    },
    {
      "epoch": 0.31848983543078413,
      "grad_norm": 1.1728390455245972,
      "learning_rate": 0.0005,
      "loss": 3.0709495544433594,
      "step": 1974
    },
    {
      "epoch": 0.3186511777992901,
      "grad_norm": 1.2084660530090332,
      "learning_rate": 0.0005,
      "loss": 2.789485454559326,
      "step": 1975
    },
    {
      "epoch": 0.31881252016779604,
      "grad_norm": 2.7704455852508545,
      "learning_rate": 0.0005,
      "loss": 3.3997721672058105,
      "step": 1976
    },
    {
      "epoch": 0.31897386253630206,
      "grad_norm": 4.422848224639893,
      "learning_rate": 0.0005,
      "loss": 3.0791735649108887,
      "step": 1977
    },
    {
      "epoch": 0.319135204904808,
      "grad_norm": 1.2645221948623657,
      "learning_rate": 0.0005,
      "loss": 3.201876163482666,
      "step": 1978
    },
    {
      "epoch": 0.31929654727331397,
      "grad_norm": 1.7006614208221436,
      "learning_rate": 0.0005,
      "loss": 3.0372676849365234,
      "step": 1979
    },
    {
      "epoch": 0.3194578896418199,
      "grad_norm": 1.3508658409118652,
      "learning_rate": 0.0005,
      "loss": 3.3720507621765137,
      "step": 1980
    },
    {
      "epoch": 0.31961923201032594,
      "grad_norm": 1.208852767944336,
      "learning_rate": 0.0005,
      "loss": 3.6694796085357666,
      "step": 1981
    },
    {
      "epoch": 0.3197805743788319,
      "grad_norm": 1.3147742748260498,
      "learning_rate": 0.0005,
      "loss": 3.0383312702178955,
      "step": 1982
    },
    {
      "epoch": 0.31994191674733785,
      "grad_norm": 1.2731373310089111,
      "learning_rate": 0.0005,
      "loss": 3.0021286010742188,
      "step": 1983
    },
    {
      "epoch": 0.3201032591158438,
      "grad_norm": 1.4158340692520142,
      "learning_rate": 0.0005,
      "loss": 2.959148406982422,
      "step": 1984
    },
    {
      "epoch": 0.32026460148434976,
      "grad_norm": 1.3502899408340454,
      "learning_rate": 0.0005,
      "loss": 2.8398308753967285,
      "step": 1985
    },
    {
      "epoch": 0.3204259438528558,
      "grad_norm": 1.2940521240234375,
      "learning_rate": 0.0005,
      "loss": 2.959476947784424,
      "step": 1986
    },
    {
      "epoch": 0.32058728622136173,
      "grad_norm": 1.2097216844558716,
      "learning_rate": 0.0005,
      "loss": 3.277937412261963,
      "step": 1987
    },
    {
      "epoch": 0.3207486285898677,
      "grad_norm": 2.528620958328247,
      "learning_rate": 0.0005,
      "loss": 3.1449971199035645,
      "step": 1988
    },
    {
      "epoch": 0.32090997095837365,
      "grad_norm": 1.3521101474761963,
      "learning_rate": 0.0005,
      "loss": 3.073399543762207,
      "step": 1989
    },
    {
      "epoch": 0.32107131332687966,
      "grad_norm": 1.3070241212844849,
      "learning_rate": 0.0005,
      "loss": 3.003326892852783,
      "step": 1990
    },
    {
      "epoch": 0.3212326556953856,
      "grad_norm": 1.2562352418899536,
      "learning_rate": 0.0005,
      "loss": 2.7665326595306396,
      "step": 1991
    },
    {
      "epoch": 0.3213939980638916,
      "grad_norm": 1.2715684175491333,
      "learning_rate": 0.0005,
      "loss": 3.436950445175171,
      "step": 1992
    },
    {
      "epoch": 0.32155534043239753,
      "grad_norm": 1.307817816734314,
      "learning_rate": 0.0005,
      "loss": 3.061286449432373,
      "step": 1993
    },
    {
      "epoch": 0.32171668280090354,
      "grad_norm": 1.2140058279037476,
      "learning_rate": 0.0005,
      "loss": 2.901111125946045,
      "step": 1994
    },
    {
      "epoch": 0.3218780251694095,
      "grad_norm": 1.8013083934783936,
      "learning_rate": 0.0005,
      "loss": 3.230869770050049,
      "step": 1995
    },
    {
      "epoch": 0.32203936753791546,
      "grad_norm": 1.4997811317443848,
      "learning_rate": 0.0005,
      "loss": 3.436253547668457,
      "step": 1996
    },
    {
      "epoch": 0.3222007099064214,
      "grad_norm": 1.9375580549240112,
      "learning_rate": 0.0005,
      "loss": 3.3466548919677734,
      "step": 1997
    },
    {
      "epoch": 0.32236205227492737,
      "grad_norm": 1.675916314125061,
      "learning_rate": 0.0005,
      "loss": 3.2587523460388184,
      "step": 1998
    },
    {
      "epoch": 0.3225233946434334,
      "grad_norm": 2.265385150909424,
      "learning_rate": 0.0005,
      "loss": 2.867432117462158,
      "step": 1999
    },
    {
      "epoch": 0.32268473701193934,
      "grad_norm": 1.7530548572540283,
      "learning_rate": 0.0005,
      "loss": 3.4197559356689453,
      "step": 2000
    },
    {
      "epoch": 0.3228460793804453,
      "grad_norm": 1.1260559558868408,
      "learning_rate": 0.0005,
      "loss": 2.8480138778686523,
      "step": 2001
    },
    {
      "epoch": 0.32300742174895125,
      "grad_norm": 1.039448857307434,
      "learning_rate": 0.0005,
      "loss": 3.6189470291137695,
      "step": 2002
    },
    {
      "epoch": 0.32316876411745726,
      "grad_norm": 1.2235502004623413,
      "learning_rate": 0.0005,
      "loss": 3.58258056640625,
      "step": 2003
    },
    {
      "epoch": 0.3233301064859632,
      "grad_norm": 1.5367838144302368,
      "learning_rate": 0.0005,
      "loss": 2.992924213409424,
      "step": 2004
    },
    {
      "epoch": 0.3234914488544692,
      "grad_norm": 1.7916791439056396,
      "learning_rate": 0.0005,
      "loss": 3.113186836242676,
      "step": 2005
    },
    {
      "epoch": 0.32365279122297513,
      "grad_norm": 5.673649787902832,
      "learning_rate": 0.0005,
      "loss": 3.5078539848327637,
      "step": 2006
    },
    {
      "epoch": 0.32381413359148115,
      "grad_norm": 1.6280205249786377,
      "learning_rate": 0.0005,
      "loss": 3.4184656143188477,
      "step": 2007
    },
    {
      "epoch": 0.3239754759599871,
      "grad_norm": 2.277369260787964,
      "learning_rate": 0.0005,
      "loss": 3.2724056243896484,
      "step": 2008
    },
    {
      "epoch": 0.32413681832849306,
      "grad_norm": 1.348899245262146,
      "learning_rate": 0.0005,
      "loss": 2.784467935562134,
      "step": 2009
    },
    {
      "epoch": 0.324298160696999,
      "grad_norm": 1.7089422941207886,
      "learning_rate": 0.0005,
      "loss": 3.1606459617614746,
      "step": 2010
    },
    {
      "epoch": 0.32445950306550503,
      "grad_norm": 1.5309805870056152,
      "learning_rate": 0.0005,
      "loss": 3.1794471740722656,
      "step": 2011
    },
    {
      "epoch": 0.324620845434011,
      "grad_norm": 1.3711954355239868,
      "learning_rate": 0.0005,
      "loss": 3.4904346466064453,
      "step": 2012
    },
    {
      "epoch": 0.32478218780251694,
      "grad_norm": 1.5870118141174316,
      "learning_rate": 0.0005,
      "loss": 3.3905394077301025,
      "step": 2013
    },
    {
      "epoch": 0.3249435301710229,
      "grad_norm": 1.3907575607299805,
      "learning_rate": 0.0005,
      "loss": 3.2951347827911377,
      "step": 2014
    },
    {
      "epoch": 0.32510487253952886,
      "grad_norm": 1.371112585067749,
      "learning_rate": 0.0005,
      "loss": 3.6161108016967773,
      "step": 2015
    },
    {
      "epoch": 0.32526621490803487,
      "grad_norm": 1.5197018384933472,
      "learning_rate": 0.0005,
      "loss": 3.6353888511657715,
      "step": 2016
    },
    {
      "epoch": 0.3254275572765408,
      "grad_norm": 1.2595016956329346,
      "learning_rate": 0.0005,
      "loss": 2.643709659576416,
      "step": 2017
    },
    {
      "epoch": 0.3255888996450468,
      "grad_norm": 1.2224706411361694,
      "learning_rate": 0.0005,
      "loss": 3.4705779552459717,
      "step": 2018
    },
    {
      "epoch": 0.32575024201355274,
      "grad_norm": 2.9741833209991455,
      "learning_rate": 0.0005,
      "loss": 2.9582109451293945,
      "step": 2019
    },
    {
      "epoch": 0.32591158438205875,
      "grad_norm": 1.2330583333969116,
      "learning_rate": 0.0005,
      "loss": 2.6827588081359863,
      "step": 2020
    },
    {
      "epoch": 0.3260729267505647,
      "grad_norm": 1.7604284286499023,
      "learning_rate": 0.0005,
      "loss": 3.258882999420166,
      "step": 2021
    },
    {
      "epoch": 0.32623426911907066,
      "grad_norm": 1.509171962738037,
      "learning_rate": 0.0005,
      "loss": 2.8814334869384766,
      "step": 2022
    },
    {
      "epoch": 0.3263956114875766,
      "grad_norm": 1.3942821025848389,
      "learning_rate": 0.0005,
      "loss": 2.951266288757324,
      "step": 2023
    },
    {
      "epoch": 0.32655695385608263,
      "grad_norm": 1.5618027448654175,
      "learning_rate": 0.0005,
      "loss": 2.944425106048584,
      "step": 2024
    },
    {
      "epoch": 0.3267182962245886,
      "grad_norm": 1.2718790769577026,
      "learning_rate": 0.0005,
      "loss": 3.423374891281128,
      "step": 2025
    },
    {
      "epoch": 0.32687963859309455,
      "grad_norm": 1.3668544292449951,
      "learning_rate": 0.0005,
      "loss": 3.2254080772399902,
      "step": 2026
    },
    {
      "epoch": 0.3270409809616005,
      "grad_norm": 1.3907686471939087,
      "learning_rate": 0.0005,
      "loss": 3.4014041423797607,
      "step": 2027
    },
    {
      "epoch": 0.32720232333010646,
      "grad_norm": 1.2957872152328491,
      "learning_rate": 0.0005,
      "loss": 3.4251620769500732,
      "step": 2028
    },
    {
      "epoch": 0.32736366569861247,
      "grad_norm": 1.2109438180923462,
      "learning_rate": 0.0005,
      "loss": 3.3840317726135254,
      "step": 2029
    },
    {
      "epoch": 0.32752500806711843,
      "grad_norm": 1.833294153213501,
      "learning_rate": 0.0005,
      "loss": 2.953890085220337,
      "step": 2030
    },
    {
      "epoch": 0.3276863504356244,
      "grad_norm": 1.2988686561584473,
      "learning_rate": 0.0005,
      "loss": 3.3539364337921143,
      "step": 2031
    },
    {
      "epoch": 0.32784769280413034,
      "grad_norm": 1.4661720991134644,
      "learning_rate": 0.0005,
      "loss": 3.135601043701172,
      "step": 2032
    },
    {
      "epoch": 0.32800903517263635,
      "grad_norm": 1.4674383401870728,
      "learning_rate": 0.0005,
      "loss": 2.7645649909973145,
      "step": 2033
    },
    {
      "epoch": 0.3281703775411423,
      "grad_norm": 1.457397699356079,
      "learning_rate": 0.0005,
      "loss": 3.39500093460083,
      "step": 2034
    },
    {
      "epoch": 0.32833171990964827,
      "grad_norm": 1.380354642868042,
      "learning_rate": 0.0005,
      "loss": 3.735745429992676,
      "step": 2035
    },
    {
      "epoch": 0.3284930622781542,
      "grad_norm": 1.5454480648040771,
      "learning_rate": 0.0005,
      "loss": 3.251685380935669,
      "step": 2036
    },
    {
      "epoch": 0.32865440464666024,
      "grad_norm": 1.2306166887283325,
      "learning_rate": 0.0005,
      "loss": 2.7387232780456543,
      "step": 2037
    },
    {
      "epoch": 0.3288157470151662,
      "grad_norm": 3.103501558303833,
      "learning_rate": 0.0005,
      "loss": 3.1451210975646973,
      "step": 2038
    },
    {
      "epoch": 0.32897708938367215,
      "grad_norm": 1.3435142040252686,
      "learning_rate": 0.0005,
      "loss": 3.22816801071167,
      "step": 2039
    },
    {
      "epoch": 0.3291384317521781,
      "grad_norm": 1.3248130083084106,
      "learning_rate": 0.0005,
      "loss": 3.1157546043395996,
      "step": 2040
    },
    {
      "epoch": 0.3292997741206841,
      "grad_norm": 1.5398778915405273,
      "learning_rate": 0.0005,
      "loss": 2.9582183361053467,
      "step": 2041
    },
    {
      "epoch": 0.3294611164891901,
      "grad_norm": 1.3527787923812866,
      "learning_rate": 0.0005,
      "loss": 3.0857596397399902,
      "step": 2042
    },
    {
      "epoch": 0.32962245885769603,
      "grad_norm": 1.1812008619308472,
      "learning_rate": 0.0005,
      "loss": 2.6872291564941406,
      "step": 2043
    },
    {
      "epoch": 0.329783801226202,
      "grad_norm": 1.9432705640792847,
      "learning_rate": 0.0005,
      "loss": 3.0395328998565674,
      "step": 2044
    },
    {
      "epoch": 0.32994514359470795,
      "grad_norm": 1.566942811012268,
      "learning_rate": 0.0005,
      "loss": 3.263211727142334,
      "step": 2045
    },
    {
      "epoch": 0.33010648596321396,
      "grad_norm": 1.279402256011963,
      "learning_rate": 0.0005,
      "loss": 3.128943920135498,
      "step": 2046
    },
    {
      "epoch": 0.3302678283317199,
      "grad_norm": 1.4232909679412842,
      "learning_rate": 0.0005,
      "loss": 3.125744104385376,
      "step": 2047
    },
    {
      "epoch": 0.33042917070022587,
      "grad_norm": 1.0952845811843872,
      "learning_rate": 0.0005,
      "loss": 2.921097755432129,
      "step": 2048
    },
    {
      "epoch": 0.33059051306873183,
      "grad_norm": 1.8895585536956787,
      "learning_rate": 0.0005,
      "loss": 4.525646209716797,
      "step": 2049
    },
    {
      "epoch": 0.33075185543723784,
      "grad_norm": 1.5280158519744873,
      "learning_rate": 0.0005,
      "loss": 3.1069297790527344,
      "step": 2050
    },
    {
      "epoch": 0.3309131978057438,
      "grad_norm": 2.118910789489746,
      "learning_rate": 0.0005,
      "loss": 3.0872278213500977,
      "step": 2051
    },
    {
      "epoch": 0.33107454017424975,
      "grad_norm": 1.3580071926116943,
      "learning_rate": 0.0005,
      "loss": 2.818737506866455,
      "step": 2052
    },
    {
      "epoch": 0.3312358825427557,
      "grad_norm": 1.3956161737442017,
      "learning_rate": 0.0005,
      "loss": 2.706299304962158,
      "step": 2053
    },
    {
      "epoch": 0.3313972249112617,
      "grad_norm": 1.8207931518554688,
      "learning_rate": 0.0005,
      "loss": 2.951597213745117,
      "step": 2054
    },
    {
      "epoch": 0.3315585672797677,
      "grad_norm": 1.3227565288543701,
      "learning_rate": 0.0005,
      "loss": 3.043827533721924,
      "step": 2055
    },
    {
      "epoch": 0.33171990964827364,
      "grad_norm": 1.3646342754364014,
      "learning_rate": 0.0005,
      "loss": 3.384718179702759,
      "step": 2056
    },
    {
      "epoch": 0.3318812520167796,
      "grad_norm": 1.3451100587844849,
      "learning_rate": 0.0005,
      "loss": 3.1234757900238037,
      "step": 2057
    },
    {
      "epoch": 0.33204259438528555,
      "grad_norm": 1.2882297039031982,
      "learning_rate": 0.0005,
      "loss": 2.7499358654022217,
      "step": 2058
    },
    {
      "epoch": 0.33220393675379156,
      "grad_norm": 1.5540801286697388,
      "learning_rate": 0.0005,
      "loss": 3.5479869842529297,
      "step": 2059
    },
    {
      "epoch": 0.3323652791222975,
      "grad_norm": 2.336564302444458,
      "learning_rate": 0.0005,
      "loss": 3.026913642883301,
      "step": 2060
    },
    {
      "epoch": 0.3325266214908035,
      "grad_norm": 1.3068974018096924,
      "learning_rate": 0.0005,
      "loss": 3.1047203540802,
      "step": 2061
    },
    {
      "epoch": 0.33268796385930943,
      "grad_norm": 1.1798757314682007,
      "learning_rate": 0.0005,
      "loss": 2.7116611003875732,
      "step": 2062
    },
    {
      "epoch": 0.33284930622781544,
      "grad_norm": 1.2890632152557373,
      "learning_rate": 0.0005,
      "loss": 2.9663963317871094,
      "step": 2063
    },
    {
      "epoch": 0.3330106485963214,
      "grad_norm": 1.209266185760498,
      "learning_rate": 0.0005,
      "loss": 3.3604698181152344,
      "step": 2064
    },
    {
      "epoch": 0.33317199096482736,
      "grad_norm": 1.4800745248794556,
      "learning_rate": 0.0005,
      "loss": 2.8677616119384766,
      "step": 2065
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 1.191434383392334,
      "learning_rate": 0.0005,
      "loss": 3.13617205619812,
      "step": 2066
    },
    {
      "epoch": 0.3334946757018393,
      "grad_norm": 1.367977499961853,
      "learning_rate": 0.0005,
      "loss": 2.7137084007263184,
      "step": 2067
    },
    {
      "epoch": 0.3336560180703453,
      "grad_norm": 1.9741219282150269,
      "learning_rate": 0.0005,
      "loss": 3.1908984184265137,
      "step": 2068
    },
    {
      "epoch": 0.33381736043885124,
      "grad_norm": 1.2872980833053589,
      "learning_rate": 0.0005,
      "loss": 3.1456003189086914,
      "step": 2069
    },
    {
      "epoch": 0.3339787028073572,
      "grad_norm": 1.509007453918457,
      "learning_rate": 0.0005,
      "loss": 2.972870111465454,
      "step": 2070
    },
    {
      "epoch": 0.33414004517586315,
      "grad_norm": 1.823904275894165,
      "learning_rate": 0.0005,
      "loss": 3.024348497390747,
      "step": 2071
    },
    {
      "epoch": 0.33430138754436917,
      "grad_norm": 1.5347527265548706,
      "learning_rate": 0.0005,
      "loss": 3.0572962760925293,
      "step": 2072
    },
    {
      "epoch": 0.3344627299128751,
      "grad_norm": 1.7534971237182617,
      "learning_rate": 0.0005,
      "loss": 3.431342124938965,
      "step": 2073
    },
    {
      "epoch": 0.3346240722813811,
      "grad_norm": 1.361124038696289,
      "learning_rate": 0.0005,
      "loss": 2.8490536212921143,
      "step": 2074
    },
    {
      "epoch": 0.33478541464988704,
      "grad_norm": 1.3361976146697998,
      "learning_rate": 0.0005,
      "loss": 2.821404218673706,
      "step": 2075
    },
    {
      "epoch": 0.33494675701839305,
      "grad_norm": 1.7581300735473633,
      "learning_rate": 0.0005,
      "loss": 3.0964057445526123,
      "step": 2076
    },
    {
      "epoch": 0.335108099386899,
      "grad_norm": 1.3502297401428223,
      "learning_rate": 0.0005,
      "loss": 3.009896993637085,
      "step": 2077
    },
    {
      "epoch": 0.33526944175540496,
      "grad_norm": 1.4094687700271606,
      "learning_rate": 0.0005,
      "loss": 3.0634052753448486,
      "step": 2078
    },
    {
      "epoch": 0.3354307841239109,
      "grad_norm": 3.264713764190674,
      "learning_rate": 0.0005,
      "loss": 2.8517470359802246,
      "step": 2079
    },
    {
      "epoch": 0.33559212649241693,
      "grad_norm": 1.8196606636047363,
      "learning_rate": 0.0005,
      "loss": 2.8879852294921875,
      "step": 2080
    },
    {
      "epoch": 0.3357534688609229,
      "grad_norm": 1.3090156316757202,
      "learning_rate": 0.0005,
      "loss": 3.066570281982422,
      "step": 2081
    },
    {
      "epoch": 0.33591481122942884,
      "grad_norm": 1.3510795831680298,
      "learning_rate": 0.0005,
      "loss": 3.0137083530426025,
      "step": 2082
    },
    {
      "epoch": 0.3360761535979348,
      "grad_norm": 1.6176447868347168,
      "learning_rate": 0.0005,
      "loss": 3.314328908920288,
      "step": 2083
    },
    {
      "epoch": 0.3362374959664408,
      "grad_norm": 4.427891254425049,
      "learning_rate": 0.0005,
      "loss": 3.746419906616211,
      "step": 2084
    },
    {
      "epoch": 0.33639883833494677,
      "grad_norm": 1.9224495887756348,
      "learning_rate": 0.0005,
      "loss": 3.0568902492523193,
      "step": 2085
    },
    {
      "epoch": 0.3365601807034527,
      "grad_norm": 1.464279294013977,
      "learning_rate": 0.0005,
      "loss": 2.6319780349731445,
      "step": 2086
    },
    {
      "epoch": 0.3367215230719587,
      "grad_norm": 1.368402123451233,
      "learning_rate": 0.0005,
      "loss": 2.9366872310638428,
      "step": 2087
    },
    {
      "epoch": 0.33688286544046464,
      "grad_norm": 1.3662068843841553,
      "learning_rate": 0.0005,
      "loss": 3.047480583190918,
      "step": 2088
    },
    {
      "epoch": 0.33704420780897065,
      "grad_norm": 1.67141854763031,
      "learning_rate": 0.0005,
      "loss": 3.032209873199463,
      "step": 2089
    },
    {
      "epoch": 0.3372055501774766,
      "grad_norm": 1.8061909675598145,
      "learning_rate": 0.0005,
      "loss": 2.844557285308838,
      "step": 2090
    },
    {
      "epoch": 0.33736689254598257,
      "grad_norm": 1.8256523609161377,
      "learning_rate": 0.0005,
      "loss": 2.782186985015869,
      "step": 2091
    },
    {
      "epoch": 0.3375282349144885,
      "grad_norm": 1.2577142715454102,
      "learning_rate": 0.0005,
      "loss": 3.5218958854675293,
      "step": 2092
    },
    {
      "epoch": 0.33768957728299454,
      "grad_norm": 1.5286195278167725,
      "learning_rate": 0.0005,
      "loss": 2.8269360065460205,
      "step": 2093
    },
    {
      "epoch": 0.3378509196515005,
      "grad_norm": 1.3240108489990234,
      "learning_rate": 0.0005,
      "loss": 3.1531872749328613,
      "step": 2094
    },
    {
      "epoch": 0.33801226202000645,
      "grad_norm": 1.1808323860168457,
      "learning_rate": 0.0005,
      "loss": 2.975741386413574,
      "step": 2095
    },
    {
      "epoch": 0.3381736043885124,
      "grad_norm": 1.3328614234924316,
      "learning_rate": 0.0005,
      "loss": 2.9791860580444336,
      "step": 2096
    },
    {
      "epoch": 0.3383349467570184,
      "grad_norm": 1.6247847080230713,
      "learning_rate": 0.0005,
      "loss": 3.2858657836914062,
      "step": 2097
    },
    {
      "epoch": 0.3384962891255244,
      "grad_norm": 1.617395043373108,
      "learning_rate": 0.0005,
      "loss": 3.3273916244506836,
      "step": 2098
    },
    {
      "epoch": 0.33865763149403033,
      "grad_norm": 1.1832939386367798,
      "learning_rate": 0.0005,
      "loss": 3.296995162963867,
      "step": 2099
    },
    {
      "epoch": 0.3388189738625363,
      "grad_norm": 1.221198558807373,
      "learning_rate": 0.0005,
      "loss": 3.6755080223083496,
      "step": 2100
    },
    {
      "epoch": 0.33898031623104224,
      "grad_norm": 1.536336064338684,
      "learning_rate": 0.0005,
      "loss": 3.101379871368408,
      "step": 2101
    },
    {
      "epoch": 0.33914165859954826,
      "grad_norm": 1.327133297920227,
      "learning_rate": 0.0005,
      "loss": 3.3318562507629395,
      "step": 2102
    },
    {
      "epoch": 0.3393030009680542,
      "grad_norm": 1.52426278591156,
      "learning_rate": 0.0005,
      "loss": 3.184126853942871,
      "step": 2103
    },
    {
      "epoch": 0.33946434333656017,
      "grad_norm": 1.368839144706726,
      "learning_rate": 0.0005,
      "loss": 2.9617223739624023,
      "step": 2104
    },
    {
      "epoch": 0.3396256857050661,
      "grad_norm": 1.9926995038986206,
      "learning_rate": 0.0005,
      "loss": 2.5914182662963867,
      "step": 2105
    },
    {
      "epoch": 0.33978702807357214,
      "grad_norm": 1.8419170379638672,
      "learning_rate": 0.0005,
      "loss": 2.9919490814208984,
      "step": 2106
    },
    {
      "epoch": 0.3399483704420781,
      "grad_norm": 1.3414889574050903,
      "learning_rate": 0.0005,
      "loss": 2.7934818267822266,
      "step": 2107
    },
    {
      "epoch": 0.34010971281058405,
      "grad_norm": 1.75336492061615,
      "learning_rate": 0.0005,
      "loss": 3.5096583366394043,
      "step": 2108
    },
    {
      "epoch": 0.34027105517909,
      "grad_norm": 1.6347157955169678,
      "learning_rate": 0.0005,
      "loss": 3.5114905834198,
      "step": 2109
    },
    {
      "epoch": 0.340432397547596,
      "grad_norm": 1.2873307466506958,
      "learning_rate": 0.0005,
      "loss": 3.5920915603637695,
      "step": 2110
    },
    {
      "epoch": 0.340593739916102,
      "grad_norm": 1.2271157503128052,
      "learning_rate": 0.0005,
      "loss": 3.0410327911376953,
      "step": 2111
    },
    {
      "epoch": 0.34075508228460794,
      "grad_norm": 1.6400971412658691,
      "learning_rate": 0.0005,
      "loss": 3.0411102771759033,
      "step": 2112
    },
    {
      "epoch": 0.3409164246531139,
      "grad_norm": 1.613625407218933,
      "learning_rate": 0.0005,
      "loss": 3.0137977600097656,
      "step": 2113
    },
    {
      "epoch": 0.3410777670216199,
      "grad_norm": 1.8237780332565308,
      "learning_rate": 0.0005,
      "loss": 3.278846263885498,
      "step": 2114
    },
    {
      "epoch": 0.34123910939012586,
      "grad_norm": 1.6981252431869507,
      "learning_rate": 0.0005,
      "loss": 3.553446054458618,
      "step": 2115
    },
    {
      "epoch": 0.3414004517586318,
      "grad_norm": 1.340404987335205,
      "learning_rate": 0.0005,
      "loss": 3.1249070167541504,
      "step": 2116
    },
    {
      "epoch": 0.3415617941271378,
      "grad_norm": 1.1762827634811401,
      "learning_rate": 0.0005,
      "loss": 3.0447494983673096,
      "step": 2117
    },
    {
      "epoch": 0.34172313649564373,
      "grad_norm": 1.4866487979888916,
      "learning_rate": 0.0005,
      "loss": 2.8896191120147705,
      "step": 2118
    },
    {
      "epoch": 0.34188447886414974,
      "grad_norm": 1.6593772172927856,
      "learning_rate": 0.0005,
      "loss": 3.224189281463623,
      "step": 2119
    },
    {
      "epoch": 0.3420458212326557,
      "grad_norm": 1.175758957862854,
      "learning_rate": 0.0005,
      "loss": 3.273249626159668,
      "step": 2120
    },
    {
      "epoch": 0.34220716360116166,
      "grad_norm": 1.339167594909668,
      "learning_rate": 0.0005,
      "loss": 3.2433621883392334,
      "step": 2121
    },
    {
      "epoch": 0.3423685059696676,
      "grad_norm": 1.4690910577774048,
      "learning_rate": 0.0005,
      "loss": 2.3158295154571533,
      "step": 2122
    },
    {
      "epoch": 0.3425298483381736,
      "grad_norm": 1.5414453744888306,
      "learning_rate": 0.0005,
      "loss": 3.0319952964782715,
      "step": 2123
    },
    {
      "epoch": 0.3426911907066796,
      "grad_norm": 1.2166974544525146,
      "learning_rate": 0.0005,
      "loss": 2.8603618144989014,
      "step": 2124
    },
    {
      "epoch": 0.34285253307518554,
      "grad_norm": 1.3799058198928833,
      "learning_rate": 0.0005,
      "loss": 2.687779426574707,
      "step": 2125
    },
    {
      "epoch": 0.3430138754436915,
      "grad_norm": 2.0784285068511963,
      "learning_rate": 0.0005,
      "loss": 3.268256902694702,
      "step": 2126
    },
    {
      "epoch": 0.3431752178121975,
      "grad_norm": 1.3301317691802979,
      "learning_rate": 0.0005,
      "loss": 3.3043620586395264,
      "step": 2127
    },
    {
      "epoch": 0.34333656018070347,
      "grad_norm": 1.182403326034546,
      "learning_rate": 0.0005,
      "loss": 2.8062350749969482,
      "step": 2128
    },
    {
      "epoch": 0.3434979025492094,
      "grad_norm": 1.346721887588501,
      "learning_rate": 0.0005,
      "loss": 3.0432510375976562,
      "step": 2129
    },
    {
      "epoch": 0.3436592449177154,
      "grad_norm": 1.510867953300476,
      "learning_rate": 0.0005,
      "loss": 3.0989513397216797,
      "step": 2130
    },
    {
      "epoch": 0.34382058728622134,
      "grad_norm": 1.22191321849823,
      "learning_rate": 0.0005,
      "loss": 2.9327518939971924,
      "step": 2131
    },
    {
      "epoch": 0.34398192965472735,
      "grad_norm": 2.159287691116333,
      "learning_rate": 0.0005,
      "loss": 2.952589511871338,
      "step": 2132
    },
    {
      "epoch": 0.3441432720232333,
      "grad_norm": 1.6252052783966064,
      "learning_rate": 0.0005,
      "loss": 3.1994752883911133,
      "step": 2133
    },
    {
      "epoch": 0.34430461439173926,
      "grad_norm": 1.7230693101882935,
      "learning_rate": 0.0005,
      "loss": 2.7239181995391846,
      "step": 2134
    },
    {
      "epoch": 0.3444659567602452,
      "grad_norm": 2.1132562160491943,
      "learning_rate": 0.0005,
      "loss": 3.2325797080993652,
      "step": 2135
    },
    {
      "epoch": 0.34462729912875123,
      "grad_norm": 4.113104343414307,
      "learning_rate": 0.0005,
      "loss": 2.6848268508911133,
      "step": 2136
    },
    {
      "epoch": 0.3447886414972572,
      "grad_norm": 1.1989644765853882,
      "learning_rate": 0.0005,
      "loss": 3.095219850540161,
      "step": 2137
    },
    {
      "epoch": 0.34494998386576314,
      "grad_norm": 1.7584413290023804,
      "learning_rate": 0.0005,
      "loss": 3.10516619682312,
      "step": 2138
    },
    {
      "epoch": 0.3451113262342691,
      "grad_norm": 1.4727236032485962,
      "learning_rate": 0.0005,
      "loss": 3.3203506469726562,
      "step": 2139
    },
    {
      "epoch": 0.3452726686027751,
      "grad_norm": 1.3548260927200317,
      "learning_rate": 0.0005,
      "loss": 3.246915102005005,
      "step": 2140
    },
    {
      "epoch": 0.34543401097128107,
      "grad_norm": 1.5563467741012573,
      "learning_rate": 0.0005,
      "loss": 3.0773754119873047,
      "step": 2141
    },
    {
      "epoch": 0.345595353339787,
      "grad_norm": 1.4918030500411987,
      "learning_rate": 0.0005,
      "loss": 3.13608455657959,
      "step": 2142
    },
    {
      "epoch": 0.345756695708293,
      "grad_norm": 2.3466596603393555,
      "learning_rate": 0.0005,
      "loss": 3.0526835918426514,
      "step": 2143
    },
    {
      "epoch": 0.345918038076799,
      "grad_norm": 1.3927165269851685,
      "learning_rate": 0.0005,
      "loss": 2.9052820205688477,
      "step": 2144
    },
    {
      "epoch": 0.34607938044530495,
      "grad_norm": 1.414836049079895,
      "learning_rate": 0.0005,
      "loss": 2.8586325645446777,
      "step": 2145
    },
    {
      "epoch": 0.3462407228138109,
      "grad_norm": 1.6214247941970825,
      "learning_rate": 0.0005,
      "loss": 2.9336161613464355,
      "step": 2146
    },
    {
      "epoch": 0.34640206518231687,
      "grad_norm": 1.569151520729065,
      "learning_rate": 0.0005,
      "loss": 3.161898136138916,
      "step": 2147
    },
    {
      "epoch": 0.3465634075508228,
      "grad_norm": 1.5350552797317505,
      "learning_rate": 0.0005,
      "loss": 3.116934299468994,
      "step": 2148
    },
    {
      "epoch": 0.34672474991932883,
      "grad_norm": 1.222652554512024,
      "learning_rate": 0.0005,
      "loss": 2.7302699089050293,
      "step": 2149
    },
    {
      "epoch": 0.3468860922878348,
      "grad_norm": 1.3308660984039307,
      "learning_rate": 0.0005,
      "loss": 3.369699478149414,
      "step": 2150
    },
    {
      "epoch": 0.34704743465634075,
      "grad_norm": 1.5281625986099243,
      "learning_rate": 0.0005,
      "loss": 3.0130910873413086,
      "step": 2151
    },
    {
      "epoch": 0.3472087770248467,
      "grad_norm": 1.0577590465545654,
      "learning_rate": 0.0005,
      "loss": 2.7650651931762695,
      "step": 2152
    },
    {
      "epoch": 0.3473701193933527,
      "grad_norm": 1.3988564014434814,
      "learning_rate": 0.0005,
      "loss": 2.943979263305664,
      "step": 2153
    },
    {
      "epoch": 0.3475314617618587,
      "grad_norm": 1.3482080698013306,
      "learning_rate": 0.0005,
      "loss": 3.3093385696411133,
      "step": 2154
    },
    {
      "epoch": 0.34769280413036463,
      "grad_norm": 1.2728111743927002,
      "learning_rate": 0.0005,
      "loss": 3.1996946334838867,
      "step": 2155
    },
    {
      "epoch": 0.3478541464988706,
      "grad_norm": 1.4850075244903564,
      "learning_rate": 0.0005,
      "loss": 2.43776798248291,
      "step": 2156
    },
    {
      "epoch": 0.3480154888673766,
      "grad_norm": 1.152748942375183,
      "learning_rate": 0.0005,
      "loss": 2.8868331909179688,
      "step": 2157
    },
    {
      "epoch": 0.34817683123588256,
      "grad_norm": 1.5729137659072876,
      "learning_rate": 0.0005,
      "loss": 2.959597587585449,
      "step": 2158
    },
    {
      "epoch": 0.3483381736043885,
      "grad_norm": 1.9340627193450928,
      "learning_rate": 0.0005,
      "loss": 3.3417558670043945,
      "step": 2159
    },
    {
      "epoch": 0.34849951597289447,
      "grad_norm": 1.640826940536499,
      "learning_rate": 0.0005,
      "loss": 3.469428300857544,
      "step": 2160
    },
    {
      "epoch": 0.3486608583414004,
      "grad_norm": 1.2278047800064087,
      "learning_rate": 0.0005,
      "loss": 3.244617462158203,
      "step": 2161
    },
    {
      "epoch": 0.34882220070990644,
      "grad_norm": 1.6020050048828125,
      "learning_rate": 0.0005,
      "loss": 3.520781993865967,
      "step": 2162
    },
    {
      "epoch": 0.3489835430784124,
      "grad_norm": 1.5654059648513794,
      "learning_rate": 0.0005,
      "loss": 3.3326756954193115,
      "step": 2163
    },
    {
      "epoch": 0.34914488544691835,
      "grad_norm": 1.2294831275939941,
      "learning_rate": 0.0005,
      "loss": 2.886899948120117,
      "step": 2164
    },
    {
      "epoch": 0.3493062278154243,
      "grad_norm": 1.5952720642089844,
      "learning_rate": 0.0005,
      "loss": 2.615023136138916,
      "step": 2165
    },
    {
      "epoch": 0.3494675701839303,
      "grad_norm": 1.4117718935012817,
      "learning_rate": 0.0005,
      "loss": 2.6515393257141113,
      "step": 2166
    },
    {
      "epoch": 0.3496289125524363,
      "grad_norm": 1.4638864994049072,
      "learning_rate": 0.0005,
      "loss": 2.911512851715088,
      "step": 2167
    },
    {
      "epoch": 0.34979025492094223,
      "grad_norm": 1.7090857028961182,
      "learning_rate": 0.0005,
      "loss": 2.899710178375244,
      "step": 2168
    },
    {
      "epoch": 0.3499515972894482,
      "grad_norm": 1.1327651739120483,
      "learning_rate": 0.0005,
      "loss": 3.1824958324432373,
      "step": 2169
    },
    {
      "epoch": 0.3501129396579542,
      "grad_norm": 1.438546895980835,
      "learning_rate": 0.0005,
      "loss": 3.421957492828369,
      "step": 2170
    },
    {
      "epoch": 0.35027428202646016,
      "grad_norm": 1.3667726516723633,
      "learning_rate": 0.0005,
      "loss": 2.9575276374816895,
      "step": 2171
    },
    {
      "epoch": 0.3504356243949661,
      "grad_norm": 1.5398824214935303,
      "learning_rate": 0.0005,
      "loss": 3.4158949851989746,
      "step": 2172
    },
    {
      "epoch": 0.3505969667634721,
      "grad_norm": 1.4052245616912842,
      "learning_rate": 0.0005,
      "loss": 2.826005697250366,
      "step": 2173
    },
    {
      "epoch": 0.35075830913197803,
      "grad_norm": 2.2859785556793213,
      "learning_rate": 0.0005,
      "loss": 3.507254123687744,
      "step": 2174
    },
    {
      "epoch": 0.35091965150048404,
      "grad_norm": 1.3386602401733398,
      "learning_rate": 0.0005,
      "loss": 3.3636066913604736,
      "step": 2175
    },
    {
      "epoch": 0.35108099386899,
      "grad_norm": 1.492030143737793,
      "learning_rate": 0.0005,
      "loss": 2.6048994064331055,
      "step": 2176
    },
    {
      "epoch": 0.35124233623749596,
      "grad_norm": 1.4649333953857422,
      "learning_rate": 0.0005,
      "loss": 3.12974214553833,
      "step": 2177
    },
    {
      "epoch": 0.3514036786060019,
      "grad_norm": 1.438405156135559,
      "learning_rate": 0.0005,
      "loss": 3.0436081886291504,
      "step": 2178
    },
    {
      "epoch": 0.3515650209745079,
      "grad_norm": 1.3234102725982666,
      "learning_rate": 0.0005,
      "loss": 3.1415863037109375,
      "step": 2179
    },
    {
      "epoch": 0.3517263633430139,
      "grad_norm": 2.2078380584716797,
      "learning_rate": 0.0005,
      "loss": 3.081117630004883,
      "step": 2180
    },
    {
      "epoch": 0.35188770571151984,
      "grad_norm": 1.2641583681106567,
      "learning_rate": 0.0005,
      "loss": 2.669393539428711,
      "step": 2181
    },
    {
      "epoch": 0.3520490480800258,
      "grad_norm": 1.3021785020828247,
      "learning_rate": 0.0005,
      "loss": 2.4248533248901367,
      "step": 2182
    },
    {
      "epoch": 0.3522103904485318,
      "grad_norm": 1.2714712619781494,
      "learning_rate": 0.0005,
      "loss": 3.218661308288574,
      "step": 2183
    },
    {
      "epoch": 0.35237173281703776,
      "grad_norm": 1.5721811056137085,
      "learning_rate": 0.0005,
      "loss": 3.730011463165283,
      "step": 2184
    },
    {
      "epoch": 0.3525330751855437,
      "grad_norm": 1.9032169580459595,
      "learning_rate": 0.0005,
      "loss": 2.8429903984069824,
      "step": 2185
    },
    {
      "epoch": 0.3526944175540497,
      "grad_norm": 1.1965758800506592,
      "learning_rate": 0.0005,
      "loss": 2.8854923248291016,
      "step": 2186
    },
    {
      "epoch": 0.3528557599225557,
      "grad_norm": 2.1911168098449707,
      "learning_rate": 0.0005,
      "loss": 2.8042869567871094,
      "step": 2187
    },
    {
      "epoch": 0.35301710229106165,
      "grad_norm": 1.5604392290115356,
      "learning_rate": 0.0005,
      "loss": 3.125465154647827,
      "step": 2188
    },
    {
      "epoch": 0.3531784446595676,
      "grad_norm": 1.4685527086257935,
      "learning_rate": 0.0005,
      "loss": 3.123434543609619,
      "step": 2189
    },
    {
      "epoch": 0.35333978702807356,
      "grad_norm": 1.7224326133728027,
      "learning_rate": 0.0005,
      "loss": 2.9361209869384766,
      "step": 2190
    },
    {
      "epoch": 0.3535011293965795,
      "grad_norm": 1.2134634256362915,
      "learning_rate": 0.0005,
      "loss": 3.100735664367676,
      "step": 2191
    },
    {
      "epoch": 0.35366247176508553,
      "grad_norm": 1.6212186813354492,
      "learning_rate": 0.0005,
      "loss": 2.773707151412964,
      "step": 2192
    },
    {
      "epoch": 0.3538238141335915,
      "grad_norm": 1.2978798151016235,
      "learning_rate": 0.0005,
      "loss": 2.8747169971466064,
      "step": 2193
    },
    {
      "epoch": 0.35398515650209744,
      "grad_norm": 1.4132006168365479,
      "learning_rate": 0.0005,
      "loss": 3.423555850982666,
      "step": 2194
    },
    {
      "epoch": 0.3541464988706034,
      "grad_norm": 1.5145163536071777,
      "learning_rate": 0.0005,
      "loss": 3.0987374782562256,
      "step": 2195
    },
    {
      "epoch": 0.3543078412391094,
      "grad_norm": 1.131601095199585,
      "learning_rate": 0.0005,
      "loss": 2.694272994995117,
      "step": 2196
    },
    {
      "epoch": 0.35446918360761537,
      "grad_norm": 1.5433906316757202,
      "learning_rate": 0.0005,
      "loss": 3.0625667572021484,
      "step": 2197
    },
    {
      "epoch": 0.3546305259761213,
      "grad_norm": 1.4241923093795776,
      "learning_rate": 0.0005,
      "loss": 3.447925329208374,
      "step": 2198
    },
    {
      "epoch": 0.3547918683446273,
      "grad_norm": 1.683508038520813,
      "learning_rate": 0.0005,
      "loss": 3.034597158432007,
      "step": 2199
    },
    {
      "epoch": 0.3549532107131333,
      "grad_norm": 1.5755938291549683,
      "learning_rate": 0.0005,
      "loss": 3.2614121437072754,
      "step": 2200
    },
    {
      "epoch": 0.35511455308163925,
      "grad_norm": 2.1245272159576416,
      "learning_rate": 0.0005,
      "loss": 2.6958394050598145,
      "step": 2201
    },
    {
      "epoch": 0.3552758954501452,
      "grad_norm": 1.5899138450622559,
      "learning_rate": 0.0005,
      "loss": 3.152679920196533,
      "step": 2202
    },
    {
      "epoch": 0.35543723781865116,
      "grad_norm": 1.4895049333572388,
      "learning_rate": 0.0005,
      "loss": 2.5049514770507812,
      "step": 2203
    },
    {
      "epoch": 0.3555985801871571,
      "grad_norm": 1.3602937459945679,
      "learning_rate": 0.0005,
      "loss": 2.740421772003174,
      "step": 2204
    },
    {
      "epoch": 0.35575992255566313,
      "grad_norm": 1.670569896697998,
      "learning_rate": 0.0005,
      "loss": 3.404820442199707,
      "step": 2205
    },
    {
      "epoch": 0.3559212649241691,
      "grad_norm": 1.6616655588150024,
      "learning_rate": 0.0005,
      "loss": 2.5491607189178467,
      "step": 2206
    },
    {
      "epoch": 0.35608260729267505,
      "grad_norm": 1.5690051317214966,
      "learning_rate": 0.0005,
      "loss": 2.804229736328125,
      "step": 2207
    },
    {
      "epoch": 0.356243949661181,
      "grad_norm": 1.1263483762741089,
      "learning_rate": 0.0005,
      "loss": 3.170243978500366,
      "step": 2208
    },
    {
      "epoch": 0.356405292029687,
      "grad_norm": 1.4916455745697021,
      "learning_rate": 0.0005,
      "loss": 3.118896484375,
      "step": 2209
    },
    {
      "epoch": 0.35656663439819297,
      "grad_norm": 2.2988224029541016,
      "learning_rate": 0.0005,
      "loss": 3.0623912811279297,
      "step": 2210
    },
    {
      "epoch": 0.35672797676669893,
      "grad_norm": 1.1523678302764893,
      "learning_rate": 0.0005,
      "loss": 3.256655693054199,
      "step": 2211
    },
    {
      "epoch": 0.3568893191352049,
      "grad_norm": 1.7319635152816772,
      "learning_rate": 0.0005,
      "loss": 3.2493374347686768,
      "step": 2212
    },
    {
      "epoch": 0.3570506615037109,
      "grad_norm": 1.294468641281128,
      "learning_rate": 0.0005,
      "loss": 3.120767593383789,
      "step": 2213
    },
    {
      "epoch": 0.35721200387221685,
      "grad_norm": 1.5317720174789429,
      "learning_rate": 0.0005,
      "loss": 2.8254380226135254,
      "step": 2214
    },
    {
      "epoch": 0.3573733462407228,
      "grad_norm": 1.9259967803955078,
      "learning_rate": 0.0005,
      "loss": 3.3615217208862305,
      "step": 2215
    },
    {
      "epoch": 0.35753468860922877,
      "grad_norm": 1.2982195615768433,
      "learning_rate": 0.0005,
      "loss": 3.0589425563812256,
      "step": 2216
    },
    {
      "epoch": 0.3576960309777348,
      "grad_norm": 1.2956721782684326,
      "learning_rate": 0.0005,
      "loss": 3.045741558074951,
      "step": 2217
    },
    {
      "epoch": 0.35785737334624074,
      "grad_norm": 1.0670076608657837,
      "learning_rate": 0.0005,
      "loss": 2.7118663787841797,
      "step": 2218
    },
    {
      "epoch": 0.3580187157147467,
      "grad_norm": 1.25813889503479,
      "learning_rate": 0.0005,
      "loss": 3.2559168338775635,
      "step": 2219
    },
    {
      "epoch": 0.35818005808325265,
      "grad_norm": 1.0169339179992676,
      "learning_rate": 0.0005,
      "loss": 2.908395767211914,
      "step": 2220
    },
    {
      "epoch": 0.3583414004517586,
      "grad_norm": 1.2303327322006226,
      "learning_rate": 0.0005,
      "loss": 3.316394329071045,
      "step": 2221
    },
    {
      "epoch": 0.3585027428202646,
      "grad_norm": 1.5813398361206055,
      "learning_rate": 0.0005,
      "loss": 3.251650333404541,
      "step": 2222
    },
    {
      "epoch": 0.3586640851887706,
      "grad_norm": 1.4833067655563354,
      "learning_rate": 0.0005,
      "loss": 3.4047703742980957,
      "step": 2223
    },
    {
      "epoch": 0.35882542755727653,
      "grad_norm": 1.2364394664764404,
      "learning_rate": 0.0005,
      "loss": 3.051485538482666,
      "step": 2224
    },
    {
      "epoch": 0.3589867699257825,
      "grad_norm": 0.9950693249702454,
      "learning_rate": 0.0005,
      "loss": 2.9790921211242676,
      "step": 2225
    },
    {
      "epoch": 0.3591481122942885,
      "grad_norm": 1.3754503726959229,
      "learning_rate": 0.0005,
      "loss": 3.309903144836426,
      "step": 2226
    },
    {
      "epoch": 0.35930945466279446,
      "grad_norm": 1.7462620735168457,
      "learning_rate": 0.0005,
      "loss": 2.6479079723358154,
      "step": 2227
    },
    {
      "epoch": 0.3594707970313004,
      "grad_norm": 1.8926446437835693,
      "learning_rate": 0.0005,
      "loss": 3.154970169067383,
      "step": 2228
    },
    {
      "epoch": 0.35963213939980637,
      "grad_norm": 1.205702781677246,
      "learning_rate": 0.0005,
      "loss": 3.049926519393921,
      "step": 2229
    },
    {
      "epoch": 0.3597934817683124,
      "grad_norm": 1.3954626321792603,
      "learning_rate": 0.0005,
      "loss": 2.7851643562316895,
      "step": 2230
    },
    {
      "epoch": 0.35995482413681834,
      "grad_norm": 1.2009621858596802,
      "learning_rate": 0.0005,
      "loss": 2.8374433517456055,
      "step": 2231
    },
    {
      "epoch": 0.3601161665053243,
      "grad_norm": 1.60430908203125,
      "learning_rate": 0.0005,
      "loss": 2.86600923538208,
      "step": 2232
    },
    {
      "epoch": 0.36027750887383025,
      "grad_norm": 1.6797021627426147,
      "learning_rate": 0.0005,
      "loss": 2.8712706565856934,
      "step": 2233
    },
    {
      "epoch": 0.3604388512423362,
      "grad_norm": 1.1882191896438599,
      "learning_rate": 0.0005,
      "loss": 3.0549590587615967,
      "step": 2234
    },
    {
      "epoch": 0.3606001936108422,
      "grad_norm": 1.397695541381836,
      "learning_rate": 0.0005,
      "loss": 3.201570987701416,
      "step": 2235
    },
    {
      "epoch": 0.3607615359793482,
      "grad_norm": 1.271307349205017,
      "learning_rate": 0.0005,
      "loss": 2.7538790702819824,
      "step": 2236
    },
    {
      "epoch": 0.36092287834785414,
      "grad_norm": 1.4280239343643188,
      "learning_rate": 0.0005,
      "loss": 3.2675604820251465,
      "step": 2237
    },
    {
      "epoch": 0.3610842207163601,
      "grad_norm": 2.5412955284118652,
      "learning_rate": 0.0005,
      "loss": 3.41290283203125,
      "step": 2238
    },
    {
      "epoch": 0.3612455630848661,
      "grad_norm": 1.6139800548553467,
      "learning_rate": 0.0005,
      "loss": 3.125113010406494,
      "step": 2239
    },
    {
      "epoch": 0.36140690545337206,
      "grad_norm": 1.199486494064331,
      "learning_rate": 0.0005,
      "loss": 3.218432903289795,
      "step": 2240
    },
    {
      "epoch": 0.361568247821878,
      "grad_norm": 1.1807116270065308,
      "learning_rate": 0.0005,
      "loss": 2.9589295387268066,
      "step": 2241
    },
    {
      "epoch": 0.361729590190384,
      "grad_norm": 1.7376900911331177,
      "learning_rate": 0.0005,
      "loss": 3.118340015411377,
      "step": 2242
    },
    {
      "epoch": 0.36189093255889,
      "grad_norm": 1.2307806015014648,
      "learning_rate": 0.0005,
      "loss": 3.344449281692505,
      "step": 2243
    },
    {
      "epoch": 0.36205227492739595,
      "grad_norm": 1.2314308881759644,
      "learning_rate": 0.0005,
      "loss": 3.0136005878448486,
      "step": 2244
    },
    {
      "epoch": 0.3622136172959019,
      "grad_norm": 1.4082728624343872,
      "learning_rate": 0.0005,
      "loss": 2.930018186569214,
      "step": 2245
    },
    {
      "epoch": 0.36237495966440786,
      "grad_norm": 1.119013786315918,
      "learning_rate": 0.0005,
      "loss": 3.039184093475342,
      "step": 2246
    },
    {
      "epoch": 0.36253630203291387,
      "grad_norm": 1.8476399183273315,
      "learning_rate": 0.0005,
      "loss": 2.633422374725342,
      "step": 2247
    },
    {
      "epoch": 0.3626976444014198,
      "grad_norm": 1.3205171823501587,
      "learning_rate": 0.0005,
      "loss": 3.0247278213500977,
      "step": 2248
    },
    {
      "epoch": 0.3628589867699258,
      "grad_norm": 1.295910120010376,
      "learning_rate": 0.0005,
      "loss": 3.061253309249878,
      "step": 2249
    },
    {
      "epoch": 0.36302032913843174,
      "grad_norm": 1.2746670246124268,
      "learning_rate": 0.0005,
      "loss": 2.9171793460845947,
      "step": 2250
    },
    {
      "epoch": 0.3631816715069377,
      "grad_norm": 1.4964364767074585,
      "learning_rate": 0.0005,
      "loss": 3.006917715072632,
      "step": 2251
    },
    {
      "epoch": 0.3633430138754437,
      "grad_norm": 1.417090892791748,
      "learning_rate": 0.0005,
      "loss": 3.043351888656616,
      "step": 2252
    },
    {
      "epoch": 0.36350435624394967,
      "grad_norm": 1.1239429712295532,
      "learning_rate": 0.0005,
      "loss": 2.6123178005218506,
      "step": 2253
    },
    {
      "epoch": 0.3636656986124556,
      "grad_norm": 1.1344612836837769,
      "learning_rate": 0.0005,
      "loss": 2.7897095680236816,
      "step": 2254
    },
    {
      "epoch": 0.3638270409809616,
      "grad_norm": 1.2653762102127075,
      "learning_rate": 0.0005,
      "loss": 3.4400267601013184,
      "step": 2255
    },
    {
      "epoch": 0.3639883833494676,
      "grad_norm": 2.0927979946136475,
      "learning_rate": 0.0005,
      "loss": 3.306334972381592,
      "step": 2256
    },
    {
      "epoch": 0.36414972571797355,
      "grad_norm": 1.1739141941070557,
      "learning_rate": 0.0005,
      "loss": 2.894191265106201,
      "step": 2257
    },
    {
      "epoch": 0.3643110680864795,
      "grad_norm": 1.5408791303634644,
      "learning_rate": 0.0005,
      "loss": 2.9746432304382324,
      "step": 2258
    },
    {
      "epoch": 0.36447241045498546,
      "grad_norm": 1.5783709287643433,
      "learning_rate": 0.0005,
      "loss": 2.643451690673828,
      "step": 2259
    },
    {
      "epoch": 0.3646337528234915,
      "grad_norm": 1.2631490230560303,
      "learning_rate": 0.0005,
      "loss": 3.147925615310669,
      "step": 2260
    },
    {
      "epoch": 0.36479509519199743,
      "grad_norm": 1.3294203281402588,
      "learning_rate": 0.0005,
      "loss": 3.0307135581970215,
      "step": 2261
    },
    {
      "epoch": 0.3649564375605034,
      "grad_norm": 1.161340594291687,
      "learning_rate": 0.0005,
      "loss": 3.4989800453186035,
      "step": 2262
    },
    {
      "epoch": 0.36511777992900935,
      "grad_norm": 1.5825899839401245,
      "learning_rate": 0.0005,
      "loss": 3.1841611862182617,
      "step": 2263
    },
    {
      "epoch": 0.3652791222975153,
      "grad_norm": 1.1655113697052002,
      "learning_rate": 0.0005,
      "loss": 3.228269100189209,
      "step": 2264
    },
    {
      "epoch": 0.3654404646660213,
      "grad_norm": 1.6289082765579224,
      "learning_rate": 0.0005,
      "loss": 2.7958600521087646,
      "step": 2265
    },
    {
      "epoch": 0.36560180703452727,
      "grad_norm": 1.9938801527023315,
      "learning_rate": 0.0005,
      "loss": 3.30308198928833,
      "step": 2266
    },
    {
      "epoch": 0.3657631494030332,
      "grad_norm": 7.606457710266113,
      "learning_rate": 0.0005,
      "loss": 3.371570348739624,
      "step": 2267
    },
    {
      "epoch": 0.3659244917715392,
      "grad_norm": 1.738118052482605,
      "learning_rate": 0.0005,
      "loss": 3.3859944343566895,
      "step": 2268
    },
    {
      "epoch": 0.3660858341400452,
      "grad_norm": 1.3712626695632935,
      "learning_rate": 0.0005,
      "loss": 2.8278234004974365,
      "step": 2269
    },
    {
      "epoch": 0.36624717650855115,
      "grad_norm": 1.48121178150177,
      "learning_rate": 0.0005,
      "loss": 2.7739269733428955,
      "step": 2270
    },
    {
      "epoch": 0.3664085188770571,
      "grad_norm": 1.4222643375396729,
      "learning_rate": 0.0005,
      "loss": 3.02424955368042,
      "step": 2271
    },
    {
      "epoch": 0.36656986124556307,
      "grad_norm": 1.7674806118011475,
      "learning_rate": 0.0005,
      "loss": 2.9570579528808594,
      "step": 2272
    },
    {
      "epoch": 0.3667312036140691,
      "grad_norm": 1.2280453443527222,
      "learning_rate": 0.0005,
      "loss": 2.857987403869629,
      "step": 2273
    },
    {
      "epoch": 0.36689254598257504,
      "grad_norm": 1.0983399152755737,
      "learning_rate": 0.0005,
      "loss": 2.403512477874756,
      "step": 2274
    },
    {
      "epoch": 0.367053888351081,
      "grad_norm": 1.5244572162628174,
      "learning_rate": 0.0005,
      "loss": 3.008054256439209,
      "step": 2275
    },
    {
      "epoch": 0.36721523071958695,
      "grad_norm": 1.42417573928833,
      "learning_rate": 0.0005,
      "loss": 3.3164095878601074,
      "step": 2276
    },
    {
      "epoch": 0.3673765730880929,
      "grad_norm": 1.3552411794662476,
      "learning_rate": 0.0005,
      "loss": 2.8655543327331543,
      "step": 2277
    },
    {
      "epoch": 0.3675379154565989,
      "grad_norm": 1.1112732887268066,
      "learning_rate": 0.0005,
      "loss": 3.1152777671813965,
      "step": 2278
    },
    {
      "epoch": 0.3676992578251049,
      "grad_norm": 1.2966786623001099,
      "learning_rate": 0.0005,
      "loss": 2.637979745864868,
      "step": 2279
    },
    {
      "epoch": 0.36786060019361083,
      "grad_norm": 1.3927093744277954,
      "learning_rate": 0.0005,
      "loss": 2.916048526763916,
      "step": 2280
    },
    {
      "epoch": 0.3680219425621168,
      "grad_norm": 1.2422486543655396,
      "learning_rate": 0.0005,
      "loss": 2.9490935802459717,
      "step": 2281
    },
    {
      "epoch": 0.3681832849306228,
      "grad_norm": 1.4834634065628052,
      "learning_rate": 0.0005,
      "loss": 2.958853006362915,
      "step": 2282
    },
    {
      "epoch": 0.36834462729912876,
      "grad_norm": 1.0431556701660156,
      "learning_rate": 0.0005,
      "loss": 2.515617847442627,
      "step": 2283
    },
    {
      "epoch": 0.3685059696676347,
      "grad_norm": 1.7318110466003418,
      "learning_rate": 0.0005,
      "loss": 2.8462464809417725,
      "step": 2284
    },
    {
      "epoch": 0.36866731203614067,
      "grad_norm": 1.3465640544891357,
      "learning_rate": 0.0005,
      "loss": 3.0775413513183594,
      "step": 2285
    },
    {
      "epoch": 0.3688286544046467,
      "grad_norm": 1.8013454675674438,
      "learning_rate": 0.0005,
      "loss": 3.0699877738952637,
      "step": 2286
    },
    {
      "epoch": 0.36898999677315264,
      "grad_norm": 1.7401052713394165,
      "learning_rate": 0.0005,
      "loss": 2.768024206161499,
      "step": 2287
    },
    {
      "epoch": 0.3691513391416586,
      "grad_norm": 1.5464249849319458,
      "learning_rate": 0.0005,
      "loss": 3.246393918991089,
      "step": 2288
    },
    {
      "epoch": 0.36931268151016455,
      "grad_norm": 1.2286059856414795,
      "learning_rate": 0.0005,
      "loss": 2.9605631828308105,
      "step": 2289
    },
    {
      "epoch": 0.36947402387867057,
      "grad_norm": 1.3577649593353271,
      "learning_rate": 0.0005,
      "loss": 3.2507028579711914,
      "step": 2290
    },
    {
      "epoch": 0.3696353662471765,
      "grad_norm": 1.634746789932251,
      "learning_rate": 0.0005,
      "loss": 2.992825984954834,
      "step": 2291
    },
    {
      "epoch": 0.3697967086156825,
      "grad_norm": 1.484081506729126,
      "learning_rate": 0.0005,
      "loss": 2.585841178894043,
      "step": 2292
    },
    {
      "epoch": 0.36995805098418844,
      "grad_norm": 1.1395013332366943,
      "learning_rate": 0.0005,
      "loss": 2.742894172668457,
      "step": 2293
    },
    {
      "epoch": 0.3701193933526944,
      "grad_norm": 1.2654519081115723,
      "learning_rate": 0.0005,
      "loss": 3.1248018741607666,
      "step": 2294
    },
    {
      "epoch": 0.3702807357212004,
      "grad_norm": 1.6053203344345093,
      "learning_rate": 0.0005,
      "loss": 3.1846237182617188,
      "step": 2295
    },
    {
      "epoch": 0.37044207808970636,
      "grad_norm": 1.5814381837844849,
      "learning_rate": 0.0005,
      "loss": 3.166642665863037,
      "step": 2296
    },
    {
      "epoch": 0.3706034204582123,
      "grad_norm": 1.3792370557785034,
      "learning_rate": 0.0005,
      "loss": 2.905216932296753,
      "step": 2297
    },
    {
      "epoch": 0.3707647628267183,
      "grad_norm": 1.2726789712905884,
      "learning_rate": 0.0005,
      "loss": 3.026484966278076,
      "step": 2298
    },
    {
      "epoch": 0.3709261051952243,
      "grad_norm": 1.4913959503173828,
      "learning_rate": 0.0005,
      "loss": 3.3529365062713623,
      "step": 2299
    },
    {
      "epoch": 0.37108744756373024,
      "grad_norm": 2.4028995037078857,
      "learning_rate": 0.0005,
      "loss": 3.217332601547241,
      "step": 2300
    },
    {
      "epoch": 0.3712487899322362,
      "grad_norm": 1.2039735317230225,
      "learning_rate": 0.0005,
      "loss": 3.0037078857421875,
      "step": 2301
    },
    {
      "epoch": 0.37141013230074216,
      "grad_norm": 1.6589208841323853,
      "learning_rate": 0.0005,
      "loss": 3.0887067317962646,
      "step": 2302
    },
    {
      "epoch": 0.37157147466924817,
      "grad_norm": 1.723199486732483,
      "learning_rate": 0.0005,
      "loss": 3.7908506393432617,
      "step": 2303
    },
    {
      "epoch": 0.3717328170377541,
      "grad_norm": 1.3093804121017456,
      "learning_rate": 0.0005,
      "loss": 3.185906410217285,
      "step": 2304
    },
    {
      "epoch": 0.3718941594062601,
      "grad_norm": 1.4381707906723022,
      "learning_rate": 0.0005,
      "loss": 2.947322368621826,
      "step": 2305
    },
    {
      "epoch": 0.37205550177476604,
      "grad_norm": 1.1480861902236938,
      "learning_rate": 0.0005,
      "loss": 2.898860454559326,
      "step": 2306
    },
    {
      "epoch": 0.372216844143272,
      "grad_norm": 1.590072751045227,
      "learning_rate": 0.0005,
      "loss": 2.582606077194214,
      "step": 2307
    },
    {
      "epoch": 0.372378186511778,
      "grad_norm": 1.3347681760787964,
      "learning_rate": 0.0005,
      "loss": 3.701225519180298,
      "step": 2308
    },
    {
      "epoch": 0.37253952888028397,
      "grad_norm": 1.6235517263412476,
      "learning_rate": 0.0005,
      "loss": 3.3005757331848145,
      "step": 2309
    },
    {
      "epoch": 0.3727008712487899,
      "grad_norm": 1.132575273513794,
      "learning_rate": 0.0005,
      "loss": 2.9002904891967773,
      "step": 2310
    },
    {
      "epoch": 0.3728622136172959,
      "grad_norm": 1.2867075204849243,
      "learning_rate": 0.0005,
      "loss": 2.6725354194641113,
      "step": 2311
    },
    {
      "epoch": 0.3730235559858019,
      "grad_norm": 1.2804759740829468,
      "learning_rate": 0.0005,
      "loss": 3.279761791229248,
      "step": 2312
    },
    {
      "epoch": 0.37318489835430785,
      "grad_norm": 1.424343228340149,
      "learning_rate": 0.0005,
      "loss": 2.7033610343933105,
      "step": 2313
    },
    {
      "epoch": 0.3733462407228138,
      "grad_norm": 7.295117378234863,
      "learning_rate": 0.0005,
      "loss": 3.0397226810455322,
      "step": 2314
    },
    {
      "epoch": 0.37350758309131976,
      "grad_norm": 1.2160285711288452,
      "learning_rate": 0.0005,
      "loss": 2.2113070487976074,
      "step": 2315
    },
    {
      "epoch": 0.3736689254598258,
      "grad_norm": 1.4836899042129517,
      "learning_rate": 0.0005,
      "loss": 2.9048500061035156,
      "step": 2316
    },
    {
      "epoch": 0.37383026782833173,
      "grad_norm": 1.5354084968566895,
      "learning_rate": 0.0005,
      "loss": 3.2823944091796875,
      "step": 2317
    },
    {
      "epoch": 0.3739916101968377,
      "grad_norm": 1.5483566522598267,
      "learning_rate": 0.0005,
      "loss": 2.5826637744903564,
      "step": 2318
    },
    {
      "epoch": 0.37415295256534364,
      "grad_norm": 1.1821955442428589,
      "learning_rate": 0.0005,
      "loss": 2.839775800704956,
      "step": 2319
    },
    {
      "epoch": 0.37431429493384966,
      "grad_norm": 1.5035206079483032,
      "learning_rate": 0.0005,
      "loss": 3.291576862335205,
      "step": 2320
    },
    {
      "epoch": 0.3744756373023556,
      "grad_norm": 1.654815435409546,
      "learning_rate": 0.0005,
      "loss": 3.202033519744873,
      "step": 2321
    },
    {
      "epoch": 0.37463697967086157,
      "grad_norm": 1.2104450464248657,
      "learning_rate": 0.0005,
      "loss": 2.6428451538085938,
      "step": 2322
    },
    {
      "epoch": 0.3747983220393675,
      "grad_norm": 1.3731898069381714,
      "learning_rate": 0.0005,
      "loss": 3.119393825531006,
      "step": 2323
    },
    {
      "epoch": 0.3749596644078735,
      "grad_norm": 1.2462214231491089,
      "learning_rate": 0.0005,
      "loss": 2.6466832160949707,
      "step": 2324
    },
    {
      "epoch": 0.3751210067763795,
      "grad_norm": 1.3426986932754517,
      "learning_rate": 0.0005,
      "loss": 2.6942214965820312,
      "step": 2325
    },
    {
      "epoch": 0.37528234914488545,
      "grad_norm": 1.5382184982299805,
      "learning_rate": 0.0005,
      "loss": 2.566718101501465,
      "step": 2326
    },
    {
      "epoch": 0.3754436915133914,
      "grad_norm": 1.4921326637268066,
      "learning_rate": 0.0005,
      "loss": 2.761384963989258,
      "step": 2327
    },
    {
      "epoch": 0.37560503388189737,
      "grad_norm": 1.224314570426941,
      "learning_rate": 0.0005,
      "loss": 2.798448085784912,
      "step": 2328
    },
    {
      "epoch": 0.3757663762504034,
      "grad_norm": 1.5599744319915771,
      "learning_rate": 0.0005,
      "loss": 2.677549123764038,
      "step": 2329
    },
    {
      "epoch": 0.37592771861890933,
      "grad_norm": 1.2308344841003418,
      "learning_rate": 0.0005,
      "loss": 3.0533061027526855,
      "step": 2330
    },
    {
      "epoch": 0.3760890609874153,
      "grad_norm": 1.2618073225021362,
      "learning_rate": 0.0005,
      "loss": 3.099810838699341,
      "step": 2331
    },
    {
      "epoch": 0.37625040335592125,
      "grad_norm": 1.4164823293685913,
      "learning_rate": 0.0005,
      "loss": 2.9642577171325684,
      "step": 2332
    },
    {
      "epoch": 0.37641174572442726,
      "grad_norm": 1.523036003112793,
      "learning_rate": 0.0005,
      "loss": 2.889230251312256,
      "step": 2333
    },
    {
      "epoch": 0.3765730880929332,
      "grad_norm": 1.5923563241958618,
      "learning_rate": 0.0005,
      "loss": 3.3585610389709473,
      "step": 2334
    },
    {
      "epoch": 0.3767344304614392,
      "grad_norm": 1.4439986944198608,
      "learning_rate": 0.0005,
      "loss": 3.209301233291626,
      "step": 2335
    },
    {
      "epoch": 0.37689577282994513,
      "grad_norm": 1.2016545534133911,
      "learning_rate": 0.0005,
      "loss": 3.16835355758667,
      "step": 2336
    },
    {
      "epoch": 0.3770571151984511,
      "grad_norm": 1.1102337837219238,
      "learning_rate": 0.0005,
      "loss": 2.800976514816284,
      "step": 2337
    },
    {
      "epoch": 0.3772184575669571,
      "grad_norm": 1.7330609560012817,
      "learning_rate": 0.0005,
      "loss": 2.6686954498291016,
      "step": 2338
    },
    {
      "epoch": 0.37737979993546306,
      "grad_norm": 2.1583428382873535,
      "learning_rate": 0.0005,
      "loss": 3.4928977489471436,
      "step": 2339
    },
    {
      "epoch": 0.377541142303969,
      "grad_norm": 1.6332522630691528,
      "learning_rate": 0.0005,
      "loss": 3.0014829635620117,
      "step": 2340
    },
    {
      "epoch": 0.37770248467247497,
      "grad_norm": 1.2596768140792847,
      "learning_rate": 0.0005,
      "loss": 2.647367477416992,
      "step": 2341
    },
    {
      "epoch": 0.377863827040981,
      "grad_norm": 1.2306594848632812,
      "learning_rate": 0.0005,
      "loss": 3.152993679046631,
      "step": 2342
    },
    {
      "epoch": 0.37802516940948694,
      "grad_norm": 1.163774847984314,
      "learning_rate": 0.0005,
      "loss": 3.015571117401123,
      "step": 2343
    },
    {
      "epoch": 0.3781865117779929,
      "grad_norm": 1.1588550806045532,
      "learning_rate": 0.0005,
      "loss": 3.4705448150634766,
      "step": 2344
    },
    {
      "epoch": 0.37834785414649885,
      "grad_norm": 1.4511159658432007,
      "learning_rate": 0.0005,
      "loss": 3.206667423248291,
      "step": 2345
    },
    {
      "epoch": 0.37850919651500486,
      "grad_norm": 1.8259974718093872,
      "learning_rate": 0.0005,
      "loss": 3.358523368835449,
      "step": 2346
    },
    {
      "epoch": 0.3786705388835108,
      "grad_norm": 2.084789752960205,
      "learning_rate": 0.0005,
      "loss": 3.428833246231079,
      "step": 2347
    },
    {
      "epoch": 0.3788318812520168,
      "grad_norm": 1.3757678270339966,
      "learning_rate": 0.0005,
      "loss": 2.7838292121887207,
      "step": 2348
    },
    {
      "epoch": 0.37899322362052273,
      "grad_norm": 1.6521090269088745,
      "learning_rate": 0.0005,
      "loss": 3.2236075401306152,
      "step": 2349
    },
    {
      "epoch": 0.3791545659890287,
      "grad_norm": 1.4908417463302612,
      "learning_rate": 0.0005,
      "loss": 2.7397067546844482,
      "step": 2350
    },
    {
      "epoch": 0.3793159083575347,
      "grad_norm": 1.6317311525344849,
      "learning_rate": 0.0005,
      "loss": 2.745898723602295,
      "step": 2351
    },
    {
      "epoch": 0.37947725072604066,
      "grad_norm": 1.4340921640396118,
      "learning_rate": 0.0005,
      "loss": 3.079211473464966,
      "step": 2352
    },
    {
      "epoch": 0.3796385930945466,
      "grad_norm": 1.2402206659317017,
      "learning_rate": 0.0005,
      "loss": 3.1490297317504883,
      "step": 2353
    },
    {
      "epoch": 0.3797999354630526,
      "grad_norm": 1.3310068845748901,
      "learning_rate": 0.0005,
      "loss": 2.8411433696746826,
      "step": 2354
    },
    {
      "epoch": 0.3799612778315586,
      "grad_norm": 1.5445175170898438,
      "learning_rate": 0.0005,
      "loss": 3.129993200302124,
      "step": 2355
    },
    {
      "epoch": 0.38012262020006454,
      "grad_norm": 1.237877607345581,
      "learning_rate": 0.0005,
      "loss": 2.455043315887451,
      "step": 2356
    },
    {
      "epoch": 0.3802839625685705,
      "grad_norm": 1.9080767631530762,
      "learning_rate": 0.0005,
      "loss": 2.885422706604004,
      "step": 2357
    },
    {
      "epoch": 0.38044530493707646,
      "grad_norm": 1.341570258140564,
      "learning_rate": 0.0005,
      "loss": 2.949462413787842,
      "step": 2358
    },
    {
      "epoch": 0.38060664730558247,
      "grad_norm": 1.1503149271011353,
      "learning_rate": 0.0005,
      "loss": 3.0346508026123047,
      "step": 2359
    },
    {
      "epoch": 0.3807679896740884,
      "grad_norm": 1.5233818292617798,
      "learning_rate": 0.0005,
      "loss": 2.9731664657592773,
      "step": 2360
    },
    {
      "epoch": 0.3809293320425944,
      "grad_norm": 1.3783862590789795,
      "learning_rate": 0.0005,
      "loss": 2.7415883541107178,
      "step": 2361
    },
    {
      "epoch": 0.38109067441110034,
      "grad_norm": 1.674548864364624,
      "learning_rate": 0.0005,
      "loss": 2.5943849086761475,
      "step": 2362
    },
    {
      "epoch": 0.38125201677960635,
      "grad_norm": 1.3595960140228271,
      "learning_rate": 0.0005,
      "loss": 2.746948719024658,
      "step": 2363
    },
    {
      "epoch": 0.3814133591481123,
      "grad_norm": 1.0226373672485352,
      "learning_rate": 0.0005,
      "loss": 3.053422451019287,
      "step": 2364
    },
    {
      "epoch": 0.38157470151661826,
      "grad_norm": 1.282696008682251,
      "learning_rate": 0.0005,
      "loss": 2.7400364875793457,
      "step": 2365
    },
    {
      "epoch": 0.3817360438851242,
      "grad_norm": 1.3992120027542114,
      "learning_rate": 0.0005,
      "loss": 3.0154004096984863,
      "step": 2366
    },
    {
      "epoch": 0.3818973862536302,
      "grad_norm": 1.4180316925048828,
      "learning_rate": 0.0005,
      "loss": 2.846127986907959,
      "step": 2367
    },
    {
      "epoch": 0.3820587286221362,
      "grad_norm": 1.303254246711731,
      "learning_rate": 0.0005,
      "loss": 3.6326539516448975,
      "step": 2368
    },
    {
      "epoch": 0.38222007099064215,
      "grad_norm": 1.3863078355789185,
      "learning_rate": 0.0005,
      "loss": 3.0090904235839844,
      "step": 2369
    },
    {
      "epoch": 0.3823814133591481,
      "grad_norm": 1.6442469358444214,
      "learning_rate": 0.0005,
      "loss": 2.997666358947754,
      "step": 2370
    },
    {
      "epoch": 0.38254275572765406,
      "grad_norm": 1.6341097354888916,
      "learning_rate": 0.0005,
      "loss": 3.282844066619873,
      "step": 2371
    },
    {
      "epoch": 0.3827040980961601,
      "grad_norm": 2.7314329147338867,
      "learning_rate": 0.0005,
      "loss": 2.998886823654175,
      "step": 2372
    },
    {
      "epoch": 0.38286544046466603,
      "grad_norm": 1.4016375541687012,
      "learning_rate": 0.0005,
      "loss": 2.8389804363250732,
      "step": 2373
    },
    {
      "epoch": 0.383026782833172,
      "grad_norm": 1.3254790306091309,
      "learning_rate": 0.0005,
      "loss": 2.7991838455200195,
      "step": 2374
    },
    {
      "epoch": 0.38318812520167794,
      "grad_norm": 1.1416655778884888,
      "learning_rate": 0.0005,
      "loss": 2.8035221099853516,
      "step": 2375
    },
    {
      "epoch": 0.38334946757018395,
      "grad_norm": 1.3316158056259155,
      "learning_rate": 0.0005,
      "loss": 2.9891343116760254,
      "step": 2376
    },
    {
      "epoch": 0.3835108099386899,
      "grad_norm": 1.1217249631881714,
      "learning_rate": 0.0005,
      "loss": 2.635202169418335,
      "step": 2377
    },
    {
      "epoch": 0.38367215230719587,
      "grad_norm": 1.7964153289794922,
      "learning_rate": 0.0005,
      "loss": 2.968066692352295,
      "step": 2378
    },
    {
      "epoch": 0.3838334946757018,
      "grad_norm": 1.2280535697937012,
      "learning_rate": 0.0005,
      "loss": 3.054957866668701,
      "step": 2379
    },
    {
      "epoch": 0.3839948370442078,
      "grad_norm": 1.6579967737197876,
      "learning_rate": 0.0005,
      "loss": 3.0479679107666016,
      "step": 2380
    },
    {
      "epoch": 0.3841561794127138,
      "grad_norm": 1.2592905759811401,
      "learning_rate": 0.0005,
      "loss": 2.9281723499298096,
      "step": 2381
    },
    {
      "epoch": 0.38431752178121975,
      "grad_norm": 1.1923391819000244,
      "learning_rate": 0.0005,
      "loss": 3.3317513465881348,
      "step": 2382
    },
    {
      "epoch": 0.3844788641497257,
      "grad_norm": 1.0345195531845093,
      "learning_rate": 0.0005,
      "loss": 3.122891664505005,
      "step": 2383
    },
    {
      "epoch": 0.38464020651823166,
      "grad_norm": 1.2486344575881958,
      "learning_rate": 0.0005,
      "loss": 3.3983712196350098,
      "step": 2384
    },
    {
      "epoch": 0.3848015488867377,
      "grad_norm": 1.6397688388824463,
      "learning_rate": 0.0005,
      "loss": 2.7932589054107666,
      "step": 2385
    },
    {
      "epoch": 0.38496289125524363,
      "grad_norm": 1.1477116346359253,
      "learning_rate": 0.0005,
      "loss": 2.775426149368286,
      "step": 2386
    },
    {
      "epoch": 0.3851242336237496,
      "grad_norm": 1.2821009159088135,
      "learning_rate": 0.0005,
      "loss": 3.1918585300445557,
      "step": 2387
    },
    {
      "epoch": 0.38528557599225555,
      "grad_norm": 1.9862316846847534,
      "learning_rate": 0.0005,
      "loss": 3.3753342628479004,
      "step": 2388
    },
    {
      "epoch": 0.38544691836076156,
      "grad_norm": 1.5657343864440918,
      "learning_rate": 0.0005,
      "loss": 3.5741987228393555,
      "step": 2389
    },
    {
      "epoch": 0.3856082607292675,
      "grad_norm": 1.4240647554397583,
      "learning_rate": 0.0005,
      "loss": 2.6589579582214355,
      "step": 2390
    },
    {
      "epoch": 0.3857696030977735,
      "grad_norm": 1.1673940420150757,
      "learning_rate": 0.0005,
      "loss": 2.562991142272949,
      "step": 2391
    },
    {
      "epoch": 0.38593094546627943,
      "grad_norm": 1.7182424068450928,
      "learning_rate": 0.0005,
      "loss": 3.325618267059326,
      "step": 2392
    },
    {
      "epoch": 0.38609228783478544,
      "grad_norm": 1.5294800996780396,
      "learning_rate": 0.0005,
      "loss": 2.496769428253174,
      "step": 2393
    },
    {
      "epoch": 0.3862536302032914,
      "grad_norm": 1.3983122110366821,
      "learning_rate": 0.0005,
      "loss": 2.6572787761688232,
      "step": 2394
    },
    {
      "epoch": 0.38641497257179735,
      "grad_norm": 1.4650810956954956,
      "learning_rate": 0.0005,
      "loss": 2.944765567779541,
      "step": 2395
    },
    {
      "epoch": 0.3865763149403033,
      "grad_norm": 1.4726814031600952,
      "learning_rate": 0.0005,
      "loss": 3.247182846069336,
      "step": 2396
    },
    {
      "epoch": 0.38673765730880927,
      "grad_norm": 1.664339542388916,
      "learning_rate": 0.0005,
      "loss": 3.1831483840942383,
      "step": 2397
    },
    {
      "epoch": 0.3868989996773153,
      "grad_norm": 1.1812540292739868,
      "learning_rate": 0.0005,
      "loss": 2.960813045501709,
      "step": 2398
    },
    {
      "epoch": 0.38706034204582124,
      "grad_norm": 2.5044515132904053,
      "learning_rate": 0.0005,
      "loss": 3.64617657661438,
      "step": 2399
    },
    {
      "epoch": 0.3872216844143272,
      "grad_norm": 1.8404368162155151,
      "learning_rate": 0.0005,
      "loss": 2.8477530479431152,
      "step": 2400
    },
    {
      "epoch": 0.38738302678283315,
      "grad_norm": 1.8227758407592773,
      "learning_rate": 0.0005,
      "loss": 3.1758933067321777,
      "step": 2401
    },
    {
      "epoch": 0.38754436915133916,
      "grad_norm": 1.874415397644043,
      "learning_rate": 0.0005,
      "loss": 2.8228862285614014,
      "step": 2402
    },
    {
      "epoch": 0.3877057115198451,
      "grad_norm": 1.0673490762710571,
      "learning_rate": 0.0005,
      "loss": 3.014608860015869,
      "step": 2403
    },
    {
      "epoch": 0.3878670538883511,
      "grad_norm": 1.6721121072769165,
      "learning_rate": 0.0005,
      "loss": 2.8585691452026367,
      "step": 2404
    },
    {
      "epoch": 0.38802839625685703,
      "grad_norm": 1.5894643068313599,
      "learning_rate": 0.0005,
      "loss": 2.848966121673584,
      "step": 2405
    },
    {
      "epoch": 0.38818973862536305,
      "grad_norm": 1.3868311643600464,
      "learning_rate": 0.0005,
      "loss": 3.008805990219116,
      "step": 2406
    },
    {
      "epoch": 0.388351080993869,
      "grad_norm": 1.5439802408218384,
      "learning_rate": 0.0005,
      "loss": 2.899595022201538,
      "step": 2407
    },
    {
      "epoch": 0.38851242336237496,
      "grad_norm": 1.215914011001587,
      "learning_rate": 0.0005,
      "loss": 3.2654764652252197,
      "step": 2408
    },
    {
      "epoch": 0.3886737657308809,
      "grad_norm": 1.3675037622451782,
      "learning_rate": 0.0005,
      "loss": 2.952238082885742,
      "step": 2409
    },
    {
      "epoch": 0.3888351080993869,
      "grad_norm": 1.8645704984664917,
      "learning_rate": 0.0005,
      "loss": 2.814516067504883,
      "step": 2410
    },
    {
      "epoch": 0.3889964504678929,
      "grad_norm": 1.1812095642089844,
      "learning_rate": 0.0005,
      "loss": 2.7602977752685547,
      "step": 2411
    },
    {
      "epoch": 0.38915779283639884,
      "grad_norm": 1.3361730575561523,
      "learning_rate": 0.0005,
      "loss": 2.7341372966766357,
      "step": 2412
    },
    {
      "epoch": 0.3893191352049048,
      "grad_norm": 1.5316376686096191,
      "learning_rate": 0.0005,
      "loss": 3.0909905433654785,
      "step": 2413
    },
    {
      "epoch": 0.38948047757341075,
      "grad_norm": 1.1481550931930542,
      "learning_rate": 0.0005,
      "loss": 2.4419684410095215,
      "step": 2414
    },
    {
      "epoch": 0.38964181994191677,
      "grad_norm": 1.4080910682678223,
      "learning_rate": 0.0005,
      "loss": 2.6668944358825684,
      "step": 2415
    },
    {
      "epoch": 0.3898031623104227,
      "grad_norm": 1.1403201818466187,
      "learning_rate": 0.0005,
      "loss": 2.752507209777832,
      "step": 2416
    },
    {
      "epoch": 0.3899645046789287,
      "grad_norm": 1.1374284029006958,
      "learning_rate": 0.0005,
      "loss": 3.0922470092773438,
      "step": 2417
    },
    {
      "epoch": 0.39012584704743464,
      "grad_norm": 1.5042648315429688,
      "learning_rate": 0.0005,
      "loss": 3.421855926513672,
      "step": 2418
    },
    {
      "epoch": 0.39028718941594065,
      "grad_norm": 1.2053076028823853,
      "learning_rate": 0.0005,
      "loss": 2.676870822906494,
      "step": 2419
    },
    {
      "epoch": 0.3904485317844466,
      "grad_norm": 1.324386477470398,
      "learning_rate": 0.0005,
      "loss": 2.8924787044525146,
      "step": 2420
    },
    {
      "epoch": 0.39060987415295256,
      "grad_norm": 1.2898750305175781,
      "learning_rate": 0.0005,
      "loss": 3.099216938018799,
      "step": 2421
    },
    {
      "epoch": 0.3907712165214585,
      "grad_norm": 1.0656102895736694,
      "learning_rate": 0.0005,
      "loss": 2.449265718460083,
      "step": 2422
    },
    {
      "epoch": 0.39093255888996453,
      "grad_norm": 1.4464772939682007,
      "learning_rate": 0.0005,
      "loss": 3.266974925994873,
      "step": 2423
    },
    {
      "epoch": 0.3910939012584705,
      "grad_norm": 1.6097067594528198,
      "learning_rate": 0.0005,
      "loss": 2.9396190643310547,
      "step": 2424
    },
    {
      "epoch": 0.39125524362697645,
      "grad_norm": 1.3843581676483154,
      "learning_rate": 0.0005,
      "loss": 2.987117052078247,
      "step": 2425
    },
    {
      "epoch": 0.3914165859954824,
      "grad_norm": 1.5651525259017944,
      "learning_rate": 0.0005,
      "loss": 2.560276508331299,
      "step": 2426
    },
    {
      "epoch": 0.39157792836398836,
      "grad_norm": 1.348798155784607,
      "learning_rate": 0.0005,
      "loss": 3.256152391433716,
      "step": 2427
    },
    {
      "epoch": 0.39173927073249437,
      "grad_norm": 1.2963117361068726,
      "learning_rate": 0.0005,
      "loss": 2.9890859127044678,
      "step": 2428
    },
    {
      "epoch": 0.39190061310100033,
      "grad_norm": 1.0551725625991821,
      "learning_rate": 0.0005,
      "loss": 3.108368396759033,
      "step": 2429
    },
    {
      "epoch": 0.3920619554695063,
      "grad_norm": 1.3680119514465332,
      "learning_rate": 0.0005,
      "loss": 3.112501621246338,
      "step": 2430
    },
    {
      "epoch": 0.39222329783801224,
      "grad_norm": 1.3585588932037354,
      "learning_rate": 0.0005,
      "loss": 2.4850478172302246,
      "step": 2431
    },
    {
      "epoch": 0.39238464020651825,
      "grad_norm": 1.4802685976028442,
      "learning_rate": 0.0005,
      "loss": 3.141385793685913,
      "step": 2432
    },
    {
      "epoch": 0.3925459825750242,
      "grad_norm": 1.2746233940124512,
      "learning_rate": 0.0005,
      "loss": 3.008632183074951,
      "step": 2433
    },
    {
      "epoch": 0.39270732494353017,
      "grad_norm": 1.2349770069122314,
      "learning_rate": 0.0005,
      "loss": 2.766712188720703,
      "step": 2434
    },
    {
      "epoch": 0.3928686673120361,
      "grad_norm": 1.2950996160507202,
      "learning_rate": 0.0005,
      "loss": 2.8439395427703857,
      "step": 2435
    },
    {
      "epoch": 0.39303000968054214,
      "grad_norm": 1.376971960067749,
      "learning_rate": 0.0005,
      "loss": 2.6732025146484375,
      "step": 2436
    },
    {
      "epoch": 0.3931913520490481,
      "grad_norm": 1.631972074508667,
      "learning_rate": 0.0005,
      "loss": 2.9731249809265137,
      "step": 2437
    },
    {
      "epoch": 0.39335269441755405,
      "grad_norm": 1.4482210874557495,
      "learning_rate": 0.0005,
      "loss": 2.7079684734344482,
      "step": 2438
    },
    {
      "epoch": 0.39351403678606,
      "grad_norm": 1.9619927406311035,
      "learning_rate": 0.0005,
      "loss": 2.7298154830932617,
      "step": 2439
    },
    {
      "epoch": 0.39367537915456596,
      "grad_norm": 1.4102368354797363,
      "learning_rate": 0.0005,
      "loss": 2.8488337993621826,
      "step": 2440
    },
    {
      "epoch": 0.393836721523072,
      "grad_norm": 1.1877497434616089,
      "learning_rate": 0.0005,
      "loss": 3.1878488063812256,
      "step": 2441
    },
    {
      "epoch": 0.39399806389157793,
      "grad_norm": 1.2746381759643555,
      "learning_rate": 0.0005,
      "loss": 2.704288959503174,
      "step": 2442
    },
    {
      "epoch": 0.3941594062600839,
      "grad_norm": 1.1792875528335571,
      "learning_rate": 0.0005,
      "loss": 3.094059944152832,
      "step": 2443
    },
    {
      "epoch": 0.39432074862858985,
      "grad_norm": 1.500197410583496,
      "learning_rate": 0.0005,
      "loss": 2.6804394721984863,
      "step": 2444
    },
    {
      "epoch": 0.39448209099709586,
      "grad_norm": 1.51137113571167,
      "learning_rate": 0.0005,
      "loss": 3.271555185317993,
      "step": 2445
    },
    {
      "epoch": 0.3946434333656018,
      "grad_norm": 1.5609164237976074,
      "learning_rate": 0.0005,
      "loss": 2.7335615158081055,
      "step": 2446
    },
    {
      "epoch": 0.39480477573410777,
      "grad_norm": 1.3144521713256836,
      "learning_rate": 0.0005,
      "loss": 2.7856087684631348,
      "step": 2447
    },
    {
      "epoch": 0.39496611810261373,
      "grad_norm": 1.1093569993972778,
      "learning_rate": 0.0005,
      "loss": 3.1135926246643066,
      "step": 2448
    },
    {
      "epoch": 0.39512746047111974,
      "grad_norm": 1.228766918182373,
      "learning_rate": 0.0005,
      "loss": 3.072765827178955,
      "step": 2449
    },
    {
      "epoch": 0.3952888028396257,
      "grad_norm": 1.2631428241729736,
      "learning_rate": 0.0005,
      "loss": 3.4610533714294434,
      "step": 2450
    },
    {
      "epoch": 0.39545014520813165,
      "grad_norm": 1.1287217140197754,
      "learning_rate": 0.0005,
      "loss": 3.2027459144592285,
      "step": 2451
    },
    {
      "epoch": 0.3956114875766376,
      "grad_norm": 1.2035012245178223,
      "learning_rate": 0.0005,
      "loss": 3.4232921600341797,
      "step": 2452
    },
    {
      "epoch": 0.39577282994514357,
      "grad_norm": 1.874842643737793,
      "learning_rate": 0.0005,
      "loss": 2.5548911094665527,
      "step": 2453
    },
    {
      "epoch": 0.3959341723136496,
      "grad_norm": 1.552288293838501,
      "learning_rate": 0.0005,
      "loss": 2.9718804359436035,
      "step": 2454
    },
    {
      "epoch": 0.39609551468215554,
      "grad_norm": 1.3012324571609497,
      "learning_rate": 0.0005,
      "loss": 3.130331039428711,
      "step": 2455
    },
    {
      "epoch": 0.3962568570506615,
      "grad_norm": 1.2905710935592651,
      "learning_rate": 0.0005,
      "loss": 3.185720443725586,
      "step": 2456
    },
    {
      "epoch": 0.39641819941916745,
      "grad_norm": 1.9569363594055176,
      "learning_rate": 0.0005,
      "loss": 2.915168285369873,
      "step": 2457
    },
    {
      "epoch": 0.39657954178767346,
      "grad_norm": 1.428580403327942,
      "learning_rate": 0.0005,
      "loss": 2.769963502883911,
      "step": 2458
    },
    {
      "epoch": 0.3967408841561794,
      "grad_norm": 1.2226543426513672,
      "learning_rate": 0.0005,
      "loss": 2.952481746673584,
      "step": 2459
    },
    {
      "epoch": 0.3969022265246854,
      "grad_norm": 1.3449971675872803,
      "learning_rate": 0.0005,
      "loss": 2.916381359100342,
      "step": 2460
    },
    {
      "epoch": 0.39706356889319133,
      "grad_norm": 1.7948923110961914,
      "learning_rate": 0.0005,
      "loss": 3.0637524127960205,
      "step": 2461
    },
    {
      "epoch": 0.39722491126169734,
      "grad_norm": 1.36795973777771,
      "learning_rate": 0.0005,
      "loss": 2.798933506011963,
      "step": 2462
    },
    {
      "epoch": 0.3973862536302033,
      "grad_norm": 1.528183937072754,
      "learning_rate": 0.0005,
      "loss": 3.287273406982422,
      "step": 2463
    },
    {
      "epoch": 0.39754759599870926,
      "grad_norm": 1.6693341732025146,
      "learning_rate": 0.0005,
      "loss": 2.744450092315674,
      "step": 2464
    },
    {
      "epoch": 0.3977089383672152,
      "grad_norm": 1.8159689903259277,
      "learning_rate": 0.0005,
      "loss": 2.9285671710968018,
      "step": 2465
    },
    {
      "epoch": 0.3978702807357212,
      "grad_norm": 1.134066104888916,
      "learning_rate": 0.0005,
      "loss": 2.906147003173828,
      "step": 2466
    },
    {
      "epoch": 0.3980316231042272,
      "grad_norm": 1.3095189332962036,
      "learning_rate": 0.0005,
      "loss": 3.3757076263427734,
      "step": 2467
    },
    {
      "epoch": 0.39819296547273314,
      "grad_norm": 1.543700098991394,
      "learning_rate": 0.0005,
      "loss": 2.862722873687744,
      "step": 2468
    },
    {
      "epoch": 0.3983543078412391,
      "grad_norm": 1.6241737604141235,
      "learning_rate": 0.0005,
      "loss": 3.095130443572998,
      "step": 2469
    },
    {
      "epoch": 0.39851565020974505,
      "grad_norm": 1.2112420797348022,
      "learning_rate": 0.0005,
      "loss": 2.9739065170288086,
      "step": 2470
    },
    {
      "epoch": 0.39867699257825107,
      "grad_norm": 1.3977773189544678,
      "learning_rate": 0.0005,
      "loss": 2.713947057723999,
      "step": 2471
    },
    {
      "epoch": 0.398838334946757,
      "grad_norm": 1.6396886110305786,
      "learning_rate": 0.0005,
      "loss": 2.6332550048828125,
      "step": 2472
    },
    {
      "epoch": 0.398999677315263,
      "grad_norm": 1.3444058895111084,
      "learning_rate": 0.0005,
      "loss": 2.7013893127441406,
      "step": 2473
    },
    {
      "epoch": 0.39916101968376894,
      "grad_norm": 1.101944923400879,
      "learning_rate": 0.0005,
      "loss": 3.117598295211792,
      "step": 2474
    },
    {
      "epoch": 0.39932236205227495,
      "grad_norm": 1.318228840827942,
      "learning_rate": 0.0005,
      "loss": 2.6091604232788086,
      "step": 2475
    },
    {
      "epoch": 0.3994837044207809,
      "grad_norm": 1.4078996181488037,
      "learning_rate": 0.0005,
      "loss": 3.335723638534546,
      "step": 2476
    },
    {
      "epoch": 0.39964504678928686,
      "grad_norm": 1.1797640323638916,
      "learning_rate": 0.0005,
      "loss": 3.0970232486724854,
      "step": 2477
    },
    {
      "epoch": 0.3998063891577928,
      "grad_norm": 1.4266529083251953,
      "learning_rate": 0.0005,
      "loss": 3.0435194969177246,
      "step": 2478
    },
    {
      "epoch": 0.39996773152629883,
      "grad_norm": 1.0778720378875732,
      "learning_rate": 0.0005,
      "loss": 3.2506442070007324,
      "step": 2479
    },
    {
      "epoch": 0.4001290738948048,
      "grad_norm": 1.560927391052246,
      "learning_rate": 0.0005,
      "loss": 3.1231043338775635,
      "step": 2480
    },
    {
      "epoch": 0.40029041626331074,
      "grad_norm": 1.1569629907608032,
      "learning_rate": 0.0005,
      "loss": 2.570134162902832,
      "step": 2481
    },
    {
      "epoch": 0.4004517586318167,
      "grad_norm": 1.3870198726654053,
      "learning_rate": 0.0005,
      "loss": 2.897186040878296,
      "step": 2482
    },
    {
      "epoch": 0.40061310100032266,
      "grad_norm": 1.433720588684082,
      "learning_rate": 0.0005,
      "loss": 2.7204484939575195,
      "step": 2483
    },
    {
      "epoch": 0.40077444336882867,
      "grad_norm": 1.3064008951187134,
      "learning_rate": 0.0005,
      "loss": 2.9775161743164062,
      "step": 2484
    },
    {
      "epoch": 0.4009357857373346,
      "grad_norm": 1.1144750118255615,
      "learning_rate": 0.0005,
      "loss": 2.761359214782715,
      "step": 2485
    },
    {
      "epoch": 0.4010971281058406,
      "grad_norm": 1.438258409500122,
      "learning_rate": 0.0005,
      "loss": 2.981081485748291,
      "step": 2486
    },
    {
      "epoch": 0.40125847047434654,
      "grad_norm": 1.0333242416381836,
      "learning_rate": 0.0005,
      "loss": 2.9429738521575928,
      "step": 2487
    },
    {
      "epoch": 0.40141981284285255,
      "grad_norm": 1.1683201789855957,
      "learning_rate": 0.0005,
      "loss": 2.5253078937530518,
      "step": 2488
    },
    {
      "epoch": 0.4015811552113585,
      "grad_norm": 1.212982177734375,
      "learning_rate": 0.0005,
      "loss": 2.743008613586426,
      "step": 2489
    },
    {
      "epoch": 0.40174249757986447,
      "grad_norm": 1.38473379611969,
      "learning_rate": 0.0005,
      "loss": 2.6387534141540527,
      "step": 2490
    },
    {
      "epoch": 0.4019038399483704,
      "grad_norm": 1.2552344799041748,
      "learning_rate": 0.0005,
      "loss": 2.9795193672180176,
      "step": 2491
    },
    {
      "epoch": 0.40206518231687643,
      "grad_norm": 1.362167477607727,
      "learning_rate": 0.0005,
      "loss": 2.8583216667175293,
      "step": 2492
    },
    {
      "epoch": 0.4022265246853824,
      "grad_norm": 1.175372838973999,
      "learning_rate": 0.0005,
      "loss": 3.1023874282836914,
      "step": 2493
    },
    {
      "epoch": 0.40238786705388835,
      "grad_norm": 1.7068288326263428,
      "learning_rate": 0.0005,
      "loss": 3.2010865211486816,
      "step": 2494
    },
    {
      "epoch": 0.4025492094223943,
      "grad_norm": 1.3504208326339722,
      "learning_rate": 0.0005,
      "loss": 2.923048496246338,
      "step": 2495
    },
    {
      "epoch": 0.4027105517909003,
      "grad_norm": 1.1165050268173218,
      "learning_rate": 0.0005,
      "loss": 2.7882516384124756,
      "step": 2496
    },
    {
      "epoch": 0.4028718941594063,
      "grad_norm": 1.5855369567871094,
      "learning_rate": 0.0005,
      "loss": 3.2392783164978027,
      "step": 2497
    },
    {
      "epoch": 0.40303323652791223,
      "grad_norm": 1.2416490316390991,
      "learning_rate": 0.0005,
      "loss": 3.4248623847961426,
      "step": 2498
    },
    {
      "epoch": 0.4031945788964182,
      "grad_norm": 1.6473735570907593,
      "learning_rate": 0.0005,
      "loss": 2.674679756164551,
      "step": 2499
    },
    {
      "epoch": 0.40335592126492414,
      "grad_norm": 1.1304430961608887,
      "learning_rate": 0.0005,
      "loss": 3.3401741981506348,
      "step": 2500
    },
    {
      "epoch": 0.40351726363343016,
      "grad_norm": 1.4141291379928589,
      "learning_rate": 0.0005,
      "loss": 3.240896701812744,
      "step": 2501
    },
    {
      "epoch": 0.4036786060019361,
      "grad_norm": 1.4598019123077393,
      "learning_rate": 0.0005,
      "loss": 2.5769522190093994,
      "step": 2502
    },
    {
      "epoch": 0.40383994837044207,
      "grad_norm": 1.4307620525360107,
      "learning_rate": 0.0005,
      "loss": 2.8417930603027344,
      "step": 2503
    },
    {
      "epoch": 0.404001290738948,
      "grad_norm": 1.4062737226486206,
      "learning_rate": 0.0005,
      "loss": 2.9319143295288086,
      "step": 2504
    },
    {
      "epoch": 0.40416263310745404,
      "grad_norm": 1.4456532001495361,
      "learning_rate": 0.0005,
      "loss": 2.8304948806762695,
      "step": 2505
    },
    {
      "epoch": 0.40432397547596,
      "grad_norm": 1.2919384241104126,
      "learning_rate": 0.0005,
      "loss": 3.833876132965088,
      "step": 2506
    },
    {
      "epoch": 0.40448531784446595,
      "grad_norm": 1.3952196836471558,
      "learning_rate": 0.0005,
      "loss": 2.8701796531677246,
      "step": 2507
    },
    {
      "epoch": 0.4046466602129719,
      "grad_norm": 1.2710025310516357,
      "learning_rate": 0.0005,
      "loss": 3.0294418334960938,
      "step": 2508
    },
    {
      "epoch": 0.4048080025814779,
      "grad_norm": 2.4026265144348145,
      "learning_rate": 0.0005,
      "loss": 2.4154868125915527,
      "step": 2509
    },
    {
      "epoch": 0.4049693449499839,
      "grad_norm": 1.4555548429489136,
      "learning_rate": 0.0005,
      "loss": 2.757251501083374,
      "step": 2510
    },
    {
      "epoch": 0.40513068731848983,
      "grad_norm": 1.796012282371521,
      "learning_rate": 0.0005,
      "loss": 2.7814602851867676,
      "step": 2511
    },
    {
      "epoch": 0.4052920296869958,
      "grad_norm": 2.0439748764038086,
      "learning_rate": 0.0005,
      "loss": 3.077850818634033,
      "step": 2512
    },
    {
      "epoch": 0.40545337205550175,
      "grad_norm": 1.8792775869369507,
      "learning_rate": 0.0005,
      "loss": 3.2465720176696777,
      "step": 2513
    },
    {
      "epoch": 0.40561471442400776,
      "grad_norm": 1.3332122564315796,
      "learning_rate": 0.0005,
      "loss": 3.1497979164123535,
      "step": 2514
    },
    {
      "epoch": 0.4057760567925137,
      "grad_norm": 1.069595456123352,
      "learning_rate": 0.0005,
      "loss": 3.332162857055664,
      "step": 2515
    },
    {
      "epoch": 0.4059373991610197,
      "grad_norm": 1.3450075387954712,
      "learning_rate": 0.0005,
      "loss": 2.787522554397583,
      "step": 2516
    },
    {
      "epoch": 0.40609874152952563,
      "grad_norm": 1.0609593391418457,
      "learning_rate": 0.0005,
      "loss": 3.512662410736084,
      "step": 2517
    },
    {
      "epoch": 0.40626008389803164,
      "grad_norm": 1.1970438957214355,
      "learning_rate": 0.0005,
      "loss": 2.8583977222442627,
      "step": 2518
    },
    {
      "epoch": 0.4064214262665376,
      "grad_norm": 1.2826348543167114,
      "learning_rate": 0.0005,
      "loss": 2.6196229457855225,
      "step": 2519
    },
    {
      "epoch": 0.40658276863504356,
      "grad_norm": 1.2961790561676025,
      "learning_rate": 0.0005,
      "loss": 2.782578945159912,
      "step": 2520
    },
    {
      "epoch": 0.4067441110035495,
      "grad_norm": 1.8047305345535278,
      "learning_rate": 0.0005,
      "loss": 2.7299911975860596,
      "step": 2521
    },
    {
      "epoch": 0.4069054533720555,
      "grad_norm": 1.4534136056900024,
      "learning_rate": 0.0005,
      "loss": 2.613769054412842,
      "step": 2522
    },
    {
      "epoch": 0.4070667957405615,
      "grad_norm": 1.5342578887939453,
      "learning_rate": 0.0005,
      "loss": 3.037308692932129,
      "step": 2523
    },
    {
      "epoch": 0.40722813810906744,
      "grad_norm": 1.6881499290466309,
      "learning_rate": 0.0005,
      "loss": 3.080131769180298,
      "step": 2524
    },
    {
      "epoch": 0.4073894804775734,
      "grad_norm": 1.5176666975021362,
      "learning_rate": 0.0005,
      "loss": 2.753831386566162,
      "step": 2525
    },
    {
      "epoch": 0.4075508228460794,
      "grad_norm": 1.463903546333313,
      "learning_rate": 0.0005,
      "loss": 3.14211368560791,
      "step": 2526
    },
    {
      "epoch": 0.40771216521458536,
      "grad_norm": 1.4203904867172241,
      "learning_rate": 0.0005,
      "loss": 3.334423542022705,
      "step": 2527
    },
    {
      "epoch": 0.4078735075830913,
      "grad_norm": 3.36789870262146,
      "learning_rate": 0.0005,
      "loss": 2.56803560256958,
      "step": 2528
    },
    {
      "epoch": 0.4080348499515973,
      "grad_norm": 1.2993457317352295,
      "learning_rate": 0.0005,
      "loss": 2.5666885375976562,
      "step": 2529
    },
    {
      "epoch": 0.40819619232010323,
      "grad_norm": 3.0131685733795166,
      "learning_rate": 0.0005,
      "loss": 2.815042018890381,
      "step": 2530
    },
    {
      "epoch": 0.40835753468860925,
      "grad_norm": 1.1532882452011108,
      "learning_rate": 0.0005,
      "loss": 2.5333645343780518,
      "step": 2531
    },
    {
      "epoch": 0.4085188770571152,
      "grad_norm": 1.4353752136230469,
      "learning_rate": 0.0005,
      "loss": 3.175201416015625,
      "step": 2532
    },
    {
      "epoch": 0.40868021942562116,
      "grad_norm": 1.4592012166976929,
      "learning_rate": 0.0005,
      "loss": 2.862675189971924,
      "step": 2533
    },
    {
      "epoch": 0.4088415617941271,
      "grad_norm": 1.2806659936904907,
      "learning_rate": 0.0005,
      "loss": 3.2672078609466553,
      "step": 2534
    },
    {
      "epoch": 0.40900290416263313,
      "grad_norm": 1.1189414262771606,
      "learning_rate": 0.0005,
      "loss": 3.3836607933044434,
      "step": 2535
    },
    {
      "epoch": 0.4091642465311391,
      "grad_norm": 1.155364751815796,
      "learning_rate": 0.0005,
      "loss": 2.769116163253784,
      "step": 2536
    },
    {
      "epoch": 0.40932558889964504,
      "grad_norm": 1.6858727931976318,
      "learning_rate": 0.0005,
      "loss": 2.451749324798584,
      "step": 2537
    },
    {
      "epoch": 0.409486931268151,
      "grad_norm": 1.2592300176620483,
      "learning_rate": 0.0005,
      "loss": 2.495986223220825,
      "step": 2538
    },
    {
      "epoch": 0.409648273636657,
      "grad_norm": 1.5473958253860474,
      "learning_rate": 0.0005,
      "loss": 2.853348731994629,
      "step": 2539
    },
    {
      "epoch": 0.40980961600516297,
      "grad_norm": 1.6307175159454346,
      "learning_rate": 0.0005,
      "loss": 3.010786533355713,
      "step": 2540
    },
    {
      "epoch": 0.4099709583736689,
      "grad_norm": 1.6660493612289429,
      "learning_rate": 0.0005,
      "loss": 2.92332124710083,
      "step": 2541
    },
    {
      "epoch": 0.4101323007421749,
      "grad_norm": 1.3648492097854614,
      "learning_rate": 0.0005,
      "loss": 3.10105562210083,
      "step": 2542
    },
    {
      "epoch": 0.41029364311068084,
      "grad_norm": 1.2020097970962524,
      "learning_rate": 0.0005,
      "loss": 2.870964527130127,
      "step": 2543
    },
    {
      "epoch": 0.41045498547918685,
      "grad_norm": 1.3814376592636108,
      "learning_rate": 0.0005,
      "loss": 3.132689952850342,
      "step": 2544
    },
    {
      "epoch": 0.4106163278476928,
      "grad_norm": 1.246565341949463,
      "learning_rate": 0.0005,
      "loss": 3.0981106758117676,
      "step": 2545
    },
    {
      "epoch": 0.41077767021619876,
      "grad_norm": 1.2066044807434082,
      "learning_rate": 0.0005,
      "loss": 2.984272003173828,
      "step": 2546
    },
    {
      "epoch": 0.4109390125847047,
      "grad_norm": 1.9980913400650024,
      "learning_rate": 0.0005,
      "loss": 3.165285587310791,
      "step": 2547
    },
    {
      "epoch": 0.41110035495321073,
      "grad_norm": 1.8185515403747559,
      "learning_rate": 0.0005,
      "loss": 2.823883056640625,
      "step": 2548
    },
    {
      "epoch": 0.4112616973217167,
      "grad_norm": 1.6164257526397705,
      "learning_rate": 0.0005,
      "loss": 2.7600622177124023,
      "step": 2549
    },
    {
      "epoch": 0.41142303969022265,
      "grad_norm": 1.2023650407791138,
      "learning_rate": 0.0005,
      "loss": 2.7502403259277344,
      "step": 2550
    },
    {
      "epoch": 0.4115843820587286,
      "grad_norm": 1.2702642679214478,
      "learning_rate": 0.0005,
      "loss": 2.6219444274902344,
      "step": 2551
    },
    {
      "epoch": 0.4117457244272346,
      "grad_norm": 1.119624376296997,
      "learning_rate": 0.0005,
      "loss": 3.2149627208709717,
      "step": 2552
    },
    {
      "epoch": 0.4119070667957406,
      "grad_norm": 1.1183987855911255,
      "learning_rate": 0.0005,
      "loss": 3.2186856269836426,
      "step": 2553
    },
    {
      "epoch": 0.41206840916424653,
      "grad_norm": 1.9162622690200806,
      "learning_rate": 0.0005,
      "loss": 3.3757846355438232,
      "step": 2554
    },
    {
      "epoch": 0.4122297515327525,
      "grad_norm": 1.166976809501648,
      "learning_rate": 0.0005,
      "loss": 2.9368438720703125,
      "step": 2555
    },
    {
      "epoch": 0.41239109390125844,
      "grad_norm": 1.7317732572555542,
      "learning_rate": 0.0005,
      "loss": 3.5874314308166504,
      "step": 2556
    },
    {
      "epoch": 0.41255243626976446,
      "grad_norm": 1.8632445335388184,
      "learning_rate": 0.0005,
      "loss": 3.3821206092834473,
      "step": 2557
    },
    {
      "epoch": 0.4127137786382704,
      "grad_norm": 1.1716821193695068,
      "learning_rate": 0.0005,
      "loss": 3.2777740955352783,
      "step": 2558
    },
    {
      "epoch": 0.41287512100677637,
      "grad_norm": 1.5568666458129883,
      "learning_rate": 0.0005,
      "loss": 2.85368013381958,
      "step": 2559
    },
    {
      "epoch": 0.4130364633752823,
      "grad_norm": 1.4042550325393677,
      "learning_rate": 0.0005,
      "loss": 3.1312055587768555,
      "step": 2560
    },
    {
      "epoch": 0.41319780574378834,
      "grad_norm": 1.6366071701049805,
      "learning_rate": 0.0005,
      "loss": 3.1559128761291504,
      "step": 2561
    },
    {
      "epoch": 0.4133591481122943,
      "grad_norm": 1.2326431274414062,
      "learning_rate": 0.0005,
      "loss": 3.254795551300049,
      "step": 2562
    },
    {
      "epoch": 0.41352049048080025,
      "grad_norm": 3.057755470275879,
      "learning_rate": 0.0005,
      "loss": 3.310771942138672,
      "step": 2563
    },
    {
      "epoch": 0.4136818328493062,
      "grad_norm": 1.6019577980041504,
      "learning_rate": 0.0005,
      "loss": 2.8502840995788574,
      "step": 2564
    },
    {
      "epoch": 0.4138431752178122,
      "grad_norm": 1.4165303707122803,
      "learning_rate": 0.0005,
      "loss": 2.8435659408569336,
      "step": 2565
    },
    {
      "epoch": 0.4140045175863182,
      "grad_norm": 1.208389401435852,
      "learning_rate": 0.0005,
      "loss": 2.9752840995788574,
      "step": 2566
    },
    {
      "epoch": 0.41416585995482413,
      "grad_norm": 1.6180580854415894,
      "learning_rate": 0.0005,
      "loss": 2.8513050079345703,
      "step": 2567
    },
    {
      "epoch": 0.4143272023233301,
      "grad_norm": 1.084073543548584,
      "learning_rate": 0.0005,
      "loss": 2.4105470180511475,
      "step": 2568
    },
    {
      "epoch": 0.4144885446918361,
      "grad_norm": 1.1871542930603027,
      "learning_rate": 0.0005,
      "loss": 2.9507226943969727,
      "step": 2569
    },
    {
      "epoch": 0.41464988706034206,
      "grad_norm": 1.3759182691574097,
      "learning_rate": 0.0005,
      "loss": 3.0170912742614746,
      "step": 2570
    },
    {
      "epoch": 0.414811229428848,
      "grad_norm": 1.0844985246658325,
      "learning_rate": 0.0005,
      "loss": 2.726797580718994,
      "step": 2571
    },
    {
      "epoch": 0.414972571797354,
      "grad_norm": 2.1013824939727783,
      "learning_rate": 0.0005,
      "loss": 3.099113941192627,
      "step": 2572
    },
    {
      "epoch": 0.41513391416585993,
      "grad_norm": 2.424243688583374,
      "learning_rate": 0.0005,
      "loss": 2.722658157348633,
      "step": 2573
    },
    {
      "epoch": 0.41529525653436594,
      "grad_norm": 1.6322212219238281,
      "learning_rate": 0.0005,
      "loss": 2.9342875480651855,
      "step": 2574
    },
    {
      "epoch": 0.4154565989028719,
      "grad_norm": 2.0114190578460693,
      "learning_rate": 0.0005,
      "loss": 3.17745304107666,
      "step": 2575
    },
    {
      "epoch": 0.41561794127137786,
      "grad_norm": 1.2635339498519897,
      "learning_rate": 0.0005,
      "loss": 2.9568676948547363,
      "step": 2576
    },
    {
      "epoch": 0.4157792836398838,
      "grad_norm": 1.4312597513198853,
      "learning_rate": 0.0005,
      "loss": 2.816619873046875,
      "step": 2577
    },
    {
      "epoch": 0.4159406260083898,
      "grad_norm": 1.274693489074707,
      "learning_rate": 0.0005,
      "loss": 3.2165720462799072,
      "step": 2578
    },
    {
      "epoch": 0.4161019683768958,
      "grad_norm": 1.5287679433822632,
      "learning_rate": 0.0005,
      "loss": 2.8381433486938477,
      "step": 2579
    },
    {
      "epoch": 0.41626331074540174,
      "grad_norm": 1.5028884410858154,
      "learning_rate": 0.0005,
      "loss": 2.5951032638549805,
      "step": 2580
    },
    {
      "epoch": 0.4164246531139077,
      "grad_norm": 1.2047669887542725,
      "learning_rate": 0.0005,
      "loss": 2.645902156829834,
      "step": 2581
    },
    {
      "epoch": 0.4165859954824137,
      "grad_norm": 1.5705044269561768,
      "learning_rate": 0.0005,
      "loss": 3.103095531463623,
      "step": 2582
    },
    {
      "epoch": 0.41674733785091966,
      "grad_norm": 1.1150097846984863,
      "learning_rate": 0.0005,
      "loss": 3.113477945327759,
      "step": 2583
    },
    {
      "epoch": 0.4169086802194256,
      "grad_norm": 1.5975366830825806,
      "learning_rate": 0.0005,
      "loss": 3.3723397254943848,
      "step": 2584
    },
    {
      "epoch": 0.4170700225879316,
      "grad_norm": 1.8532590866088867,
      "learning_rate": 0.0005,
      "loss": 2.6367969512939453,
      "step": 2585
    },
    {
      "epoch": 0.41723136495643753,
      "grad_norm": 1.2847766876220703,
      "learning_rate": 0.0005,
      "loss": 2.64890193939209,
      "step": 2586
    },
    {
      "epoch": 0.41739270732494355,
      "grad_norm": 1.3895142078399658,
      "learning_rate": 0.0005,
      "loss": 3.14019513130188,
      "step": 2587
    },
    {
      "epoch": 0.4175540496934495,
      "grad_norm": 1.3310997486114502,
      "learning_rate": 0.0005,
      "loss": 2.5588693618774414,
      "step": 2588
    },
    {
      "epoch": 0.41771539206195546,
      "grad_norm": 1.717916488647461,
      "learning_rate": 0.0005,
      "loss": 3.001925230026245,
      "step": 2589
    },
    {
      "epoch": 0.4178767344304614,
      "grad_norm": 2.4902946949005127,
      "learning_rate": 0.0005,
      "loss": 3.533937454223633,
      "step": 2590
    },
    {
      "epoch": 0.41803807679896743,
      "grad_norm": 1.473880648612976,
      "learning_rate": 0.0005,
      "loss": 3.199657917022705,
      "step": 2591
    },
    {
      "epoch": 0.4181994191674734,
      "grad_norm": 1.7220197916030884,
      "learning_rate": 0.0005,
      "loss": 3.1004462242126465,
      "step": 2592
    },
    {
      "epoch": 0.41836076153597934,
      "grad_norm": 1.1802668571472168,
      "learning_rate": 0.0005,
      "loss": 3.5731091499328613,
      "step": 2593
    },
    {
      "epoch": 0.4185221039044853,
      "grad_norm": 1.2063710689544678,
      "learning_rate": 0.0005,
      "loss": 3.3240017890930176,
      "step": 2594
    },
    {
      "epoch": 0.4186834462729913,
      "grad_norm": 1.9711899757385254,
      "learning_rate": 0.0005,
      "loss": 2.952688694000244,
      "step": 2595
    },
    {
      "epoch": 0.41884478864149727,
      "grad_norm": 1.284153938293457,
      "learning_rate": 0.0005,
      "loss": 2.717036485671997,
      "step": 2596
    },
    {
      "epoch": 0.4190061310100032,
      "grad_norm": 1.5165972709655762,
      "learning_rate": 0.0005,
      "loss": 2.840341091156006,
      "step": 2597
    },
    {
      "epoch": 0.4191674733785092,
      "grad_norm": 1.504645586013794,
      "learning_rate": 0.0005,
      "loss": 3.142453670501709,
      "step": 2598
    },
    {
      "epoch": 0.4193288157470152,
      "grad_norm": 1.574733018875122,
      "learning_rate": 0.0005,
      "loss": 2.581389904022217,
      "step": 2599
    },
    {
      "epoch": 0.41949015811552115,
      "grad_norm": 1.3523390293121338,
      "learning_rate": 0.0005,
      "loss": 2.650132179260254,
      "step": 2600
    },
    {
      "epoch": 0.4196515004840271,
      "grad_norm": 1.3941278457641602,
      "learning_rate": 0.0005,
      "loss": 2.574244976043701,
      "step": 2601
    },
    {
      "epoch": 0.41981284285253306,
      "grad_norm": 1.503872275352478,
      "learning_rate": 0.0005,
      "loss": 3.3035378456115723,
      "step": 2602
    },
    {
      "epoch": 0.419974185221039,
      "grad_norm": 1.205197811126709,
      "learning_rate": 0.0005,
      "loss": 3.280245065689087,
      "step": 2603
    },
    {
      "epoch": 0.42013552758954503,
      "grad_norm": 1.296892523765564,
      "learning_rate": 0.0005,
      "loss": 2.9038302898406982,
      "step": 2604
    },
    {
      "epoch": 0.420296869958051,
      "grad_norm": 1.405094861984253,
      "learning_rate": 0.0005,
      "loss": 2.8034863471984863,
      "step": 2605
    },
    {
      "epoch": 0.42045821232655695,
      "grad_norm": 1.0095945596694946,
      "learning_rate": 0.0005,
      "loss": 3.2772786617279053,
      "step": 2606
    },
    {
      "epoch": 0.4206195546950629,
      "grad_norm": 1.2311152219772339,
      "learning_rate": 0.0005,
      "loss": 2.969111919403076,
      "step": 2607
    },
    {
      "epoch": 0.4207808970635689,
      "grad_norm": 1.4412966966629028,
      "learning_rate": 0.0005,
      "loss": 2.586013078689575,
      "step": 2608
    },
    {
      "epoch": 0.42094223943207487,
      "grad_norm": 1.4659245014190674,
      "learning_rate": 0.0005,
      "loss": 2.743847370147705,
      "step": 2609
    },
    {
      "epoch": 0.42110358180058083,
      "grad_norm": 1.214442491531372,
      "learning_rate": 0.0005,
      "loss": 2.905770778656006,
      "step": 2610
    },
    {
      "epoch": 0.4212649241690868,
      "grad_norm": 1.3926093578338623,
      "learning_rate": 0.0005,
      "loss": 2.7798519134521484,
      "step": 2611
    },
    {
      "epoch": 0.4214262665375928,
      "grad_norm": 1.3918037414550781,
      "learning_rate": 0.0005,
      "loss": 3.1923563480377197,
      "step": 2612
    },
    {
      "epoch": 0.42158760890609875,
      "grad_norm": 1.4049021005630493,
      "learning_rate": 0.0005,
      "loss": 2.836458206176758,
      "step": 2613
    },
    {
      "epoch": 0.4217489512746047,
      "grad_norm": 1.7638400793075562,
      "learning_rate": 0.0005,
      "loss": 2.680525779724121,
      "step": 2614
    },
    {
      "epoch": 0.42191029364311067,
      "grad_norm": 1.345224380493164,
      "learning_rate": 0.0005,
      "loss": 3.1489195823669434,
      "step": 2615
    },
    {
      "epoch": 0.4220716360116166,
      "grad_norm": 1.1484993696212769,
      "learning_rate": 0.0005,
      "loss": 2.9946436882019043,
      "step": 2616
    },
    {
      "epoch": 0.42223297838012264,
      "grad_norm": 1.365962028503418,
      "learning_rate": 0.0005,
      "loss": 3.2085766792297363,
      "step": 2617
    },
    {
      "epoch": 0.4223943207486286,
      "grad_norm": 1.2589548826217651,
      "learning_rate": 0.0005,
      "loss": 2.70090913772583,
      "step": 2618
    },
    {
      "epoch": 0.42255566311713455,
      "grad_norm": 1.4235554933547974,
      "learning_rate": 0.0005,
      "loss": 2.802238941192627,
      "step": 2619
    },
    {
      "epoch": 0.4227170054856405,
      "grad_norm": 1.3565229177474976,
      "learning_rate": 0.0005,
      "loss": 3.1524949073791504,
      "step": 2620
    },
    {
      "epoch": 0.4228783478541465,
      "grad_norm": 1.078755497932434,
      "learning_rate": 0.0005,
      "loss": 2.7334165573120117,
      "step": 2621
    },
    {
      "epoch": 0.4230396902226525,
      "grad_norm": 1.2135138511657715,
      "learning_rate": 0.0005,
      "loss": 2.5521435737609863,
      "step": 2622
    },
    {
      "epoch": 0.42320103259115843,
      "grad_norm": 1.2375121116638184,
      "learning_rate": 0.0005,
      "loss": 2.9875385761260986,
      "step": 2623
    },
    {
      "epoch": 0.4233623749596644,
      "grad_norm": 2.2508962154388428,
      "learning_rate": 0.0005,
      "loss": 3.1027565002441406,
      "step": 2624
    },
    {
      "epoch": 0.4235237173281704,
      "grad_norm": 1.109035611152649,
      "learning_rate": 0.0005,
      "loss": 2.8190717697143555,
      "step": 2625
    },
    {
      "epoch": 0.42368505969667636,
      "grad_norm": 1.1658802032470703,
      "learning_rate": 0.0005,
      "loss": 2.9056570529937744,
      "step": 2626
    },
    {
      "epoch": 0.4238464020651823,
      "grad_norm": 1.2639762163162231,
      "learning_rate": 0.0005,
      "loss": 2.9812862873077393,
      "step": 2627
    },
    {
      "epoch": 0.42400774443368827,
      "grad_norm": 1.8393927812576294,
      "learning_rate": 0.0005,
      "loss": 3.0826315879821777,
      "step": 2628
    },
    {
      "epoch": 0.4241690868021943,
      "grad_norm": 2.0055441856384277,
      "learning_rate": 0.0005,
      "loss": 2.929042100906372,
      "step": 2629
    },
    {
      "epoch": 0.42433042917070024,
      "grad_norm": 1.5147055387496948,
      "learning_rate": 0.0005,
      "loss": 2.810032367706299,
      "step": 2630
    },
    {
      "epoch": 0.4244917715392062,
      "grad_norm": 1.3150854110717773,
      "learning_rate": 0.0005,
      "loss": 3.21907377243042,
      "step": 2631
    },
    {
      "epoch": 0.42465311390771215,
      "grad_norm": 1.2451828718185425,
      "learning_rate": 0.0005,
      "loss": 2.637409210205078,
      "step": 2632
    },
    {
      "epoch": 0.4248144562762181,
      "grad_norm": 1.1802879571914673,
      "learning_rate": 0.0005,
      "loss": 2.5945708751678467,
      "step": 2633
    },
    {
      "epoch": 0.4249757986447241,
      "grad_norm": 1.3602017164230347,
      "learning_rate": 0.0005,
      "loss": 3.0192503929138184,
      "step": 2634
    },
    {
      "epoch": 0.4251371410132301,
      "grad_norm": 1.814509391784668,
      "learning_rate": 0.0005,
      "loss": 3.1324892044067383,
      "step": 2635
    },
    {
      "epoch": 0.42529848338173604,
      "grad_norm": 1.6023527383804321,
      "learning_rate": 0.0005,
      "loss": 2.9097039699554443,
      "step": 2636
    },
    {
      "epoch": 0.425459825750242,
      "grad_norm": 1.5023971796035767,
      "learning_rate": 0.0005,
      "loss": 3.4887795448303223,
      "step": 2637
    },
    {
      "epoch": 0.425621168118748,
      "grad_norm": 1.6277037858963013,
      "learning_rate": 0.0005,
      "loss": 2.7013843059539795,
      "step": 2638
    },
    {
      "epoch": 0.42578251048725396,
      "grad_norm": 1.689583420753479,
      "learning_rate": 0.0005,
      "loss": 3.094926595687866,
      "step": 2639
    },
    {
      "epoch": 0.4259438528557599,
      "grad_norm": 2.276951789855957,
      "learning_rate": 0.0005,
      "loss": 2.775834560394287,
      "step": 2640
    },
    {
      "epoch": 0.4261051952242659,
      "grad_norm": 1.218534231185913,
      "learning_rate": 0.0005,
      "loss": 2.5763983726501465,
      "step": 2641
    },
    {
      "epoch": 0.4262665375927719,
      "grad_norm": 1.179042935371399,
      "learning_rate": 0.0005,
      "loss": 3.5273661613464355,
      "step": 2642
    },
    {
      "epoch": 0.42642787996127784,
      "grad_norm": 1.135460376739502,
      "learning_rate": 0.0005,
      "loss": 3.057008743286133,
      "step": 2643
    },
    {
      "epoch": 0.4265892223297838,
      "grad_norm": 1.3399444818496704,
      "learning_rate": 0.0005,
      "loss": 2.634781837463379,
      "step": 2644
    },
    {
      "epoch": 0.42675056469828976,
      "grad_norm": 1.064852237701416,
      "learning_rate": 0.0005,
      "loss": 3.4702329635620117,
      "step": 2645
    },
    {
      "epoch": 0.4269119070667957,
      "grad_norm": 0.9381930828094482,
      "learning_rate": 0.0005,
      "loss": 2.8424363136291504,
      "step": 2646
    },
    {
      "epoch": 0.4270732494353017,
      "grad_norm": 1.845832347869873,
      "learning_rate": 0.0005,
      "loss": 2.587960720062256,
      "step": 2647
    },
    {
      "epoch": 0.4272345918038077,
      "grad_norm": 1.241658091545105,
      "learning_rate": 0.0005,
      "loss": 3.2094340324401855,
      "step": 2648
    },
    {
      "epoch": 0.42739593417231364,
      "grad_norm": 1.4638476371765137,
      "learning_rate": 0.0005,
      "loss": 3.0888752937316895,
      "step": 2649
    },
    {
      "epoch": 0.4275572765408196,
      "grad_norm": 1.0861674547195435,
      "learning_rate": 0.0005,
      "loss": 3.077530860900879,
      "step": 2650
    },
    {
      "epoch": 0.4277186189093256,
      "grad_norm": 1.1455506086349487,
      "learning_rate": 0.0005,
      "loss": 3.0106616020202637,
      "step": 2651
    },
    {
      "epoch": 0.42787996127783157,
      "grad_norm": 1.323952078819275,
      "learning_rate": 0.0005,
      "loss": 2.966412305831909,
      "step": 2652
    },
    {
      "epoch": 0.4280413036463375,
      "grad_norm": 1.2239530086517334,
      "learning_rate": 0.0005,
      "loss": 3.2418136596679688,
      "step": 2653
    },
    {
      "epoch": 0.4282026460148435,
      "grad_norm": 1.0227752923965454,
      "learning_rate": 0.0005,
      "loss": 2.766934871673584,
      "step": 2654
    },
    {
      "epoch": 0.4283639883833495,
      "grad_norm": 1.1418795585632324,
      "learning_rate": 0.0005,
      "loss": 2.331695079803467,
      "step": 2655
    },
    {
      "epoch": 0.42852533075185545,
      "grad_norm": 1.8324172496795654,
      "learning_rate": 0.0005,
      "loss": 2.951280117034912,
      "step": 2656
    },
    {
      "epoch": 0.4286866731203614,
      "grad_norm": 1.165477991104126,
      "learning_rate": 0.0005,
      "loss": 2.7446179389953613,
      "step": 2657
    },
    {
      "epoch": 0.42884801548886736,
      "grad_norm": 1.4513874053955078,
      "learning_rate": 0.0005,
      "loss": 3.4834084510803223,
      "step": 2658
    },
    {
      "epoch": 0.4290093578573733,
      "grad_norm": 1.1320067644119263,
      "learning_rate": 0.0005,
      "loss": 2.8896186351776123,
      "step": 2659
    },
    {
      "epoch": 0.42917070022587933,
      "grad_norm": 1.3020329475402832,
      "learning_rate": 0.0005,
      "loss": 3.0050432682037354,
      "step": 2660
    },
    {
      "epoch": 0.4293320425943853,
      "grad_norm": 1.4795392751693726,
      "learning_rate": 0.0005,
      "loss": 2.967601776123047,
      "step": 2661
    },
    {
      "epoch": 0.42949338496289124,
      "grad_norm": 1.7645753622055054,
      "learning_rate": 0.0005,
      "loss": 2.841209888458252,
      "step": 2662
    },
    {
      "epoch": 0.4296547273313972,
      "grad_norm": 1.4595263004302979,
      "learning_rate": 0.0005,
      "loss": 2.829098701477051,
      "step": 2663
    },
    {
      "epoch": 0.4298160696999032,
      "grad_norm": 1.4472628831863403,
      "learning_rate": 0.0005,
      "loss": 2.566741943359375,
      "step": 2664
    },
    {
      "epoch": 0.42997741206840917,
      "grad_norm": 1.392141580581665,
      "learning_rate": 0.0005,
      "loss": 2.924351692199707,
      "step": 2665
    },
    {
      "epoch": 0.4301387544369151,
      "grad_norm": 1.815312385559082,
      "learning_rate": 0.0005,
      "loss": 2.4969701766967773,
      "step": 2666
    },
    {
      "epoch": 0.4303000968054211,
      "grad_norm": 1.2996106147766113,
      "learning_rate": 0.0005,
      "loss": 2.9661638736724854,
      "step": 2667
    },
    {
      "epoch": 0.4304614391739271,
      "grad_norm": 1.6031346321105957,
      "learning_rate": 0.0005,
      "loss": 2.858579635620117,
      "step": 2668
    },
    {
      "epoch": 0.43062278154243305,
      "grad_norm": 1.7936428785324097,
      "learning_rate": 0.0005,
      "loss": 3.8033058643341064,
      "step": 2669
    },
    {
      "epoch": 0.430784123910939,
      "grad_norm": 1.1416449546813965,
      "learning_rate": 0.0005,
      "loss": 2.9845824241638184,
      "step": 2670
    },
    {
      "epoch": 0.43094546627944497,
      "grad_norm": 1.1555469036102295,
      "learning_rate": 0.0005,
      "loss": 3.0932087898254395,
      "step": 2671
    },
    {
      "epoch": 0.431106808647951,
      "grad_norm": 1.20041823387146,
      "learning_rate": 0.0005,
      "loss": 2.941460609436035,
      "step": 2672
    },
    {
      "epoch": 0.43126815101645694,
      "grad_norm": 1.2459884881973267,
      "learning_rate": 0.0005,
      "loss": 2.8978679180145264,
      "step": 2673
    },
    {
      "epoch": 0.4314294933849629,
      "grad_norm": 1.7898154258728027,
      "learning_rate": 0.0005,
      "loss": 2.5563578605651855,
      "step": 2674
    },
    {
      "epoch": 0.43159083575346885,
      "grad_norm": 1.271485447883606,
      "learning_rate": 0.0005,
      "loss": 3.0749635696411133,
      "step": 2675
    },
    {
      "epoch": 0.4317521781219748,
      "grad_norm": 1.3686635494232178,
      "learning_rate": 0.0005,
      "loss": 3.1023197174072266,
      "step": 2676
    },
    {
      "epoch": 0.4319135204904808,
      "grad_norm": 1.194162130355835,
      "learning_rate": 0.0005,
      "loss": 3.1212592124938965,
      "step": 2677
    },
    {
      "epoch": 0.4320748628589868,
      "grad_norm": 1.5747954845428467,
      "learning_rate": 0.0005,
      "loss": 2.760254144668579,
      "step": 2678
    },
    {
      "epoch": 0.43223620522749273,
      "grad_norm": 1.4452402591705322,
      "learning_rate": 0.0005,
      "loss": 2.920088291168213,
      "step": 2679
    },
    {
      "epoch": 0.4323975475959987,
      "grad_norm": 1.265262484550476,
      "learning_rate": 0.0005,
      "loss": 2.5194053649902344,
      "step": 2680
    },
    {
      "epoch": 0.4325588899645047,
      "grad_norm": 1.019905686378479,
      "learning_rate": 0.0005,
      "loss": 3.0307741165161133,
      "step": 2681
    },
    {
      "epoch": 0.43272023233301066,
      "grad_norm": 1.3703947067260742,
      "learning_rate": 0.0005,
      "loss": 3.2010669708251953,
      "step": 2682
    },
    {
      "epoch": 0.4328815747015166,
      "grad_norm": 1.1699349880218506,
      "learning_rate": 0.0005,
      "loss": 2.6464719772338867,
      "step": 2683
    },
    {
      "epoch": 0.43304291707002257,
      "grad_norm": 1.2433730363845825,
      "learning_rate": 0.0005,
      "loss": 2.9989778995513916,
      "step": 2684
    },
    {
      "epoch": 0.4332042594385286,
      "grad_norm": 1.6092594861984253,
      "learning_rate": 0.0005,
      "loss": 2.93464732170105,
      "step": 2685
    },
    {
      "epoch": 0.43336560180703454,
      "grad_norm": 1.2476142644882202,
      "learning_rate": 0.0005,
      "loss": 3.1005892753601074,
      "step": 2686
    },
    {
      "epoch": 0.4335269441755405,
      "grad_norm": 1.2409473657608032,
      "learning_rate": 0.0005,
      "loss": 2.898834705352783,
      "step": 2687
    },
    {
      "epoch": 0.43368828654404645,
      "grad_norm": 1.4205673933029175,
      "learning_rate": 0.0005,
      "loss": 2.7836058139801025,
      "step": 2688
    },
    {
      "epoch": 0.4338496289125524,
      "grad_norm": 1.7334394454956055,
      "learning_rate": 0.0005,
      "loss": 2.8572568893432617,
      "step": 2689
    },
    {
      "epoch": 0.4340109712810584,
      "grad_norm": 1.7402920722961426,
      "learning_rate": 0.0005,
      "loss": 2.8214645385742188,
      "step": 2690
    },
    {
      "epoch": 0.4341723136495644,
      "grad_norm": 1.4234973192214966,
      "learning_rate": 0.0005,
      "loss": 3.0923595428466797,
      "step": 2691
    },
    {
      "epoch": 0.43433365601807034,
      "grad_norm": 2.7251222133636475,
      "learning_rate": 0.0005,
      "loss": 2.796262741088867,
      "step": 2692
    },
    {
      "epoch": 0.4344949983865763,
      "grad_norm": 1.1080347299575806,
      "learning_rate": 0.0005,
      "loss": 2.644289016723633,
      "step": 2693
    },
    {
      "epoch": 0.4346563407550823,
      "grad_norm": 1.584294080734253,
      "learning_rate": 0.0005,
      "loss": 3.1654319763183594,
      "step": 2694
    },
    {
      "epoch": 0.43481768312358826,
      "grad_norm": 1.2521950006484985,
      "learning_rate": 0.0005,
      "loss": 3.0258243083953857,
      "step": 2695
    },
    {
      "epoch": 0.4349790254920942,
      "grad_norm": 1.4035712480545044,
      "learning_rate": 0.0005,
      "loss": 2.8504562377929688,
      "step": 2696
    },
    {
      "epoch": 0.4351403678606002,
      "grad_norm": 1.4151078462600708,
      "learning_rate": 0.0005,
      "loss": 3.118621349334717,
      "step": 2697
    },
    {
      "epoch": 0.4353017102291062,
      "grad_norm": 1.476434350013733,
      "learning_rate": 0.0005,
      "loss": 2.7929229736328125,
      "step": 2698
    },
    {
      "epoch": 0.43546305259761214,
      "grad_norm": 1.1663724184036255,
      "learning_rate": 0.0005,
      "loss": 2.675139904022217,
      "step": 2699
    },
    {
      "epoch": 0.4356243949661181,
      "grad_norm": 1.540806531906128,
      "learning_rate": 0.0005,
      "loss": 3.1336441040039062,
      "step": 2700
    },
    {
      "epoch": 0.43578573733462406,
      "grad_norm": 1.8350872993469238,
      "learning_rate": 0.0005,
      "loss": 2.640097141265869,
      "step": 2701
    },
    {
      "epoch": 0.43594707970313007,
      "grad_norm": 1.738525390625,
      "learning_rate": 0.0005,
      "loss": 2.901920795440674,
      "step": 2702
    },
    {
      "epoch": 0.436108422071636,
      "grad_norm": 1.3124017715454102,
      "learning_rate": 0.0005,
      "loss": 3.200779914855957,
      "step": 2703
    },
    {
      "epoch": 0.436269764440142,
      "grad_norm": 1.336068034172058,
      "learning_rate": 0.0005,
      "loss": 3.042649269104004,
      "step": 2704
    },
    {
      "epoch": 0.43643110680864794,
      "grad_norm": 1.6287055015563965,
      "learning_rate": 0.0005,
      "loss": 3.2945425510406494,
      "step": 2705
    },
    {
      "epoch": 0.4365924491771539,
      "grad_norm": 1.3651173114776611,
      "learning_rate": 0.0005,
      "loss": 3.378249168395996,
      "step": 2706
    },
    {
      "epoch": 0.4367537915456599,
      "grad_norm": 1.243394374847412,
      "learning_rate": 0.0005,
      "loss": 2.7620840072631836,
      "step": 2707
    },
    {
      "epoch": 0.43691513391416587,
      "grad_norm": 1.3454757928848267,
      "learning_rate": 0.0005,
      "loss": 3.3200535774230957,
      "step": 2708
    },
    {
      "epoch": 0.4370764762826718,
      "grad_norm": 1.4137746095657349,
      "learning_rate": 0.0005,
      "loss": 2.685026168823242,
      "step": 2709
    },
    {
      "epoch": 0.4372378186511778,
      "grad_norm": 1.242997407913208,
      "learning_rate": 0.0005,
      "loss": 3.1733641624450684,
      "step": 2710
    },
    {
      "epoch": 0.4373991610196838,
      "grad_norm": 1.2091721296310425,
      "learning_rate": 0.0005,
      "loss": 3.034430980682373,
      "step": 2711
    },
    {
      "epoch": 0.43756050338818975,
      "grad_norm": 1.3611878156661987,
      "learning_rate": 0.0005,
      "loss": 3.0628976821899414,
      "step": 2712
    },
    {
      "epoch": 0.4377218457566957,
      "grad_norm": 1.3988957405090332,
      "learning_rate": 0.0005,
      "loss": 2.7079968452453613,
      "step": 2713
    },
    {
      "epoch": 0.43788318812520166,
      "grad_norm": 1.431488037109375,
      "learning_rate": 0.0005,
      "loss": 2.9618518352508545,
      "step": 2714
    },
    {
      "epoch": 0.4380445304937077,
      "grad_norm": 1.8490045070648193,
      "learning_rate": 0.0005,
      "loss": 2.915548801422119,
      "step": 2715
    },
    {
      "epoch": 0.43820587286221363,
      "grad_norm": 1.2168322801589966,
      "learning_rate": 0.0005,
      "loss": 2.634646415710449,
      "step": 2716
    },
    {
      "epoch": 0.4383672152307196,
      "grad_norm": 1.1070668697357178,
      "learning_rate": 0.0005,
      "loss": 3.3532285690307617,
      "step": 2717
    },
    {
      "epoch": 0.43852855759922554,
      "grad_norm": 1.8760229349136353,
      "learning_rate": 0.0005,
      "loss": 2.6743390560150146,
      "step": 2718
    },
    {
      "epoch": 0.4386898999677315,
      "grad_norm": 1.1770466566085815,
      "learning_rate": 0.0005,
      "loss": 3.335310935974121,
      "step": 2719
    },
    {
      "epoch": 0.4388512423362375,
      "grad_norm": 1.3237206935882568,
      "learning_rate": 0.0005,
      "loss": 2.942821502685547,
      "step": 2720
    },
    {
      "epoch": 0.43901258470474347,
      "grad_norm": 1.218987226486206,
      "learning_rate": 0.0005,
      "loss": 2.567983627319336,
      "step": 2721
    },
    {
      "epoch": 0.4391739270732494,
      "grad_norm": 2.0920228958129883,
      "learning_rate": 0.0005,
      "loss": 3.258990526199341,
      "step": 2722
    },
    {
      "epoch": 0.4393352694417554,
      "grad_norm": 1.1531572341918945,
      "learning_rate": 0.0005,
      "loss": 2.9690585136413574,
      "step": 2723
    },
    {
      "epoch": 0.4394966118102614,
      "grad_norm": 1.4132459163665771,
      "learning_rate": 0.0005,
      "loss": 2.895291566848755,
      "step": 2724
    },
    {
      "epoch": 0.43965795417876735,
      "grad_norm": 1.1704156398773193,
      "learning_rate": 0.0005,
      "loss": 2.885202407836914,
      "step": 2725
    },
    {
      "epoch": 0.4398192965472733,
      "grad_norm": 1.368346929550171,
      "learning_rate": 0.0005,
      "loss": 2.5540387630462646,
      "step": 2726
    },
    {
      "epoch": 0.43998063891577927,
      "grad_norm": 1.430343747138977,
      "learning_rate": 0.0005,
      "loss": 3.0914535522460938,
      "step": 2727
    },
    {
      "epoch": 0.4401419812842853,
      "grad_norm": 1.2067018747329712,
      "learning_rate": 0.0005,
      "loss": 3.069843292236328,
      "step": 2728
    },
    {
      "epoch": 0.44030332365279123,
      "grad_norm": 1.118310809135437,
      "learning_rate": 0.0005,
      "loss": 2.293168067932129,
      "step": 2729
    },
    {
      "epoch": 0.4404646660212972,
      "grad_norm": 1.1442992687225342,
      "learning_rate": 0.0005,
      "loss": 2.6540844440460205,
      "step": 2730
    },
    {
      "epoch": 0.44062600838980315,
      "grad_norm": 1.7289209365844727,
      "learning_rate": 0.0005,
      "loss": 3.053464889526367,
      "step": 2731
    },
    {
      "epoch": 0.4407873507583091,
      "grad_norm": 1.2055338621139526,
      "learning_rate": 0.0005,
      "loss": 2.965092658996582,
      "step": 2732
    },
    {
      "epoch": 0.4409486931268151,
      "grad_norm": 1.8392505645751953,
      "learning_rate": 0.0005,
      "loss": 2.633270263671875,
      "step": 2733
    },
    {
      "epoch": 0.4411100354953211,
      "grad_norm": 1.4342352151870728,
      "learning_rate": 0.0005,
      "loss": 2.625598669052124,
      "step": 2734
    },
    {
      "epoch": 0.44127137786382703,
      "grad_norm": 1.8480138778686523,
      "learning_rate": 0.0005,
      "loss": 2.8597521781921387,
      "step": 2735
    },
    {
      "epoch": 0.441432720232333,
      "grad_norm": 1.4718267917633057,
      "learning_rate": 0.0005,
      "loss": 2.668456792831421,
      "step": 2736
    },
    {
      "epoch": 0.441594062600839,
      "grad_norm": 1.679790735244751,
      "learning_rate": 0.0005,
      "loss": 2.877542734146118,
      "step": 2737
    },
    {
      "epoch": 0.44175540496934496,
      "grad_norm": 0.9789018034934998,
      "learning_rate": 0.0005,
      "loss": 3.2560534477233887,
      "step": 2738
    },
    {
      "epoch": 0.4419167473378509,
      "grad_norm": 1.3766546249389648,
      "learning_rate": 0.0005,
      "loss": 2.640988349914551,
      "step": 2739
    },
    {
      "epoch": 0.44207808970635687,
      "grad_norm": 1.1854617595672607,
      "learning_rate": 0.0005,
      "loss": 2.900648593902588,
      "step": 2740
    },
    {
      "epoch": 0.4422394320748629,
      "grad_norm": 1.585888385772705,
      "learning_rate": 0.0005,
      "loss": 3.717130661010742,
      "step": 2741
    },
    {
      "epoch": 0.44240077444336884,
      "grad_norm": 1.2883107662200928,
      "learning_rate": 0.0005,
      "loss": 2.632218360900879,
      "step": 2742
    },
    {
      "epoch": 0.4425621168118748,
      "grad_norm": 1.219678521156311,
      "learning_rate": 0.0005,
      "loss": 2.728692054748535,
      "step": 2743
    },
    {
      "epoch": 0.44272345918038075,
      "grad_norm": 1.268942952156067,
      "learning_rate": 0.0005,
      "loss": 2.989896297454834,
      "step": 2744
    },
    {
      "epoch": 0.44288480154888676,
      "grad_norm": 0.9990659952163696,
      "learning_rate": 0.0005,
      "loss": 2.4167838096618652,
      "step": 2745
    },
    {
      "epoch": 0.4430461439173927,
      "grad_norm": 1.4575544595718384,
      "learning_rate": 0.0005,
      "loss": 3.1333866119384766,
      "step": 2746
    },
    {
      "epoch": 0.4432074862858987,
      "grad_norm": 1.892891764640808,
      "learning_rate": 0.0005,
      "loss": 2.9904794692993164,
      "step": 2747
    },
    {
      "epoch": 0.44336882865440463,
      "grad_norm": 1.4276926517486572,
      "learning_rate": 0.0005,
      "loss": 2.7116594314575195,
      "step": 2748
    },
    {
      "epoch": 0.4435301710229106,
      "grad_norm": 1.6432523727416992,
      "learning_rate": 0.0005,
      "loss": 2.788851261138916,
      "step": 2749
    },
    {
      "epoch": 0.4436915133914166,
      "grad_norm": 1.7003145217895508,
      "learning_rate": 0.0005,
      "loss": 2.9631078243255615,
      "step": 2750
    },
    {
      "epoch": 0.44385285575992256,
      "grad_norm": 1.2621574401855469,
      "learning_rate": 0.0005,
      "loss": 2.9755821228027344,
      "step": 2751
    },
    {
      "epoch": 0.4440141981284285,
      "grad_norm": 1.436302661895752,
      "learning_rate": 0.0005,
      "loss": 2.7782671451568604,
      "step": 2752
    },
    {
      "epoch": 0.4441755404969345,
      "grad_norm": 1.2655408382415771,
      "learning_rate": 0.0005,
      "loss": 2.8327243328094482,
      "step": 2753
    },
    {
      "epoch": 0.4443368828654405,
      "grad_norm": 1.7156732082366943,
      "learning_rate": 0.0005,
      "loss": 3.0730128288269043,
      "step": 2754
    },
    {
      "epoch": 0.44449822523394644,
      "grad_norm": 1.191158413887024,
      "learning_rate": 0.0005,
      "loss": 3.0200304985046387,
      "step": 2755
    },
    {
      "epoch": 0.4446595676024524,
      "grad_norm": 1.34764564037323,
      "learning_rate": 0.0005,
      "loss": 3.361921787261963,
      "step": 2756
    },
    {
      "epoch": 0.44482090997095836,
      "grad_norm": 1.7220948934555054,
      "learning_rate": 0.0005,
      "loss": 3.7564923763275146,
      "step": 2757
    },
    {
      "epoch": 0.44498225233946437,
      "grad_norm": 2.017691135406494,
      "learning_rate": 0.0005,
      "loss": 3.1237969398498535,
      "step": 2758
    },
    {
      "epoch": 0.4451435947079703,
      "grad_norm": 1.2330588102340698,
      "learning_rate": 0.0005,
      "loss": 2.65061616897583,
      "step": 2759
    },
    {
      "epoch": 0.4453049370764763,
      "grad_norm": 1.2389097213745117,
      "learning_rate": 0.0005,
      "loss": 3.232781410217285,
      "step": 2760
    },
    {
      "epoch": 0.44546627944498224,
      "grad_norm": 1.6367934942245483,
      "learning_rate": 0.0005,
      "loss": 3.218458414077759,
      "step": 2761
    },
    {
      "epoch": 0.4456276218134882,
      "grad_norm": 3.8302533626556396,
      "learning_rate": 0.0005,
      "loss": 3.211190700531006,
      "step": 2762
    },
    {
      "epoch": 0.4457889641819942,
      "grad_norm": 1.5461676120758057,
      "learning_rate": 0.0005,
      "loss": 2.452720880508423,
      "step": 2763
    },
    {
      "epoch": 0.44595030655050016,
      "grad_norm": 1.4129918813705444,
      "learning_rate": 0.0005,
      "loss": 2.632655620574951,
      "step": 2764
    },
    {
      "epoch": 0.4461116489190061,
      "grad_norm": 1.4971154928207397,
      "learning_rate": 0.0005,
      "loss": 2.9921884536743164,
      "step": 2765
    },
    {
      "epoch": 0.4462729912875121,
      "grad_norm": 1.81253182888031,
      "learning_rate": 0.0005,
      "loss": 3.0484886169433594,
      "step": 2766
    },
    {
      "epoch": 0.4464343336560181,
      "grad_norm": 1.1831318140029907,
      "learning_rate": 0.0005,
      "loss": 2.669090747833252,
      "step": 2767
    },
    {
      "epoch": 0.44659567602452405,
      "grad_norm": 1.129401445388794,
      "learning_rate": 0.0005,
      "loss": 2.6255509853363037,
      "step": 2768
    },
    {
      "epoch": 0.44675701839303,
      "grad_norm": 1.3075473308563232,
      "learning_rate": 0.0005,
      "loss": 2.899998664855957,
      "step": 2769
    },
    {
      "epoch": 0.44691836076153596,
      "grad_norm": 1.5356098413467407,
      "learning_rate": 0.0005,
      "loss": 3.281949520111084,
      "step": 2770
    },
    {
      "epoch": 0.44707970313004197,
      "grad_norm": 1.270932912826538,
      "learning_rate": 0.0005,
      "loss": 3.120616912841797,
      "step": 2771
    },
    {
      "epoch": 0.44724104549854793,
      "grad_norm": 1.1223809719085693,
      "learning_rate": 0.0005,
      "loss": 3.0230255126953125,
      "step": 2772
    },
    {
      "epoch": 0.4474023878670539,
      "grad_norm": 1.5969170331954956,
      "learning_rate": 0.0005,
      "loss": 3.507193088531494,
      "step": 2773
    },
    {
      "epoch": 0.44756373023555984,
      "grad_norm": 1.1048825979232788,
      "learning_rate": 0.0005,
      "loss": 3.0166139602661133,
      "step": 2774
    },
    {
      "epoch": 0.44772507260406585,
      "grad_norm": 1.0263019800186157,
      "learning_rate": 0.0005,
      "loss": 2.9662423133850098,
      "step": 2775
    },
    {
      "epoch": 0.4478864149725718,
      "grad_norm": 1.3312593698501587,
      "learning_rate": 0.0005,
      "loss": 3.0169363021850586,
      "step": 2776
    },
    {
      "epoch": 0.44804775734107777,
      "grad_norm": 1.2807220220565796,
      "learning_rate": 0.0005,
      "loss": 2.8705084323883057,
      "step": 2777
    },
    {
      "epoch": 0.4482090997095837,
      "grad_norm": 1.7213068008422852,
      "learning_rate": 0.0005,
      "loss": 2.8407106399536133,
      "step": 2778
    },
    {
      "epoch": 0.4483704420780897,
      "grad_norm": 1.6815892457962036,
      "learning_rate": 0.0005,
      "loss": 3.2018821239471436,
      "step": 2779
    },
    {
      "epoch": 0.4485317844465957,
      "grad_norm": 3.8674392700195312,
      "learning_rate": 0.0005,
      "loss": 2.6697611808776855,
      "step": 2780
    },
    {
      "epoch": 0.44869312681510165,
      "grad_norm": 1.4949228763580322,
      "learning_rate": 0.0005,
      "loss": 3.3448123931884766,
      "step": 2781
    },
    {
      "epoch": 0.4488544691836076,
      "grad_norm": 1.2388635873794556,
      "learning_rate": 0.0005,
      "loss": 3.1083552837371826,
      "step": 2782
    },
    {
      "epoch": 0.44901581155211356,
      "grad_norm": 1.42389714717865,
      "learning_rate": 0.0005,
      "loss": 3.023310422897339,
      "step": 2783
    },
    {
      "epoch": 0.4491771539206196,
      "grad_norm": 1.195517897605896,
      "learning_rate": 0.0005,
      "loss": 2.699718475341797,
      "step": 2784
    },
    {
      "epoch": 0.44933849628912553,
      "grad_norm": 1.3739359378814697,
      "learning_rate": 0.0005,
      "loss": 2.8165016174316406,
      "step": 2785
    },
    {
      "epoch": 0.4494998386576315,
      "grad_norm": 0.9698216319084167,
      "learning_rate": 0.0005,
      "loss": 2.6004793643951416,
      "step": 2786
    },
    {
      "epoch": 0.44966118102613745,
      "grad_norm": 1.5078339576721191,
      "learning_rate": 0.0005,
      "loss": 3.187570810317993,
      "step": 2787
    },
    {
      "epoch": 0.44982252339464346,
      "grad_norm": 1.3236961364746094,
      "learning_rate": 0.0005,
      "loss": 2.7020792961120605,
      "step": 2788
    },
    {
      "epoch": 0.4499838657631494,
      "grad_norm": 1.0965203046798706,
      "learning_rate": 0.0005,
      "loss": 2.508409023284912,
      "step": 2789
    },
    {
      "epoch": 0.45014520813165537,
      "grad_norm": 1.653975486755371,
      "learning_rate": 0.0005,
      "loss": 2.6513211727142334,
      "step": 2790
    },
    {
      "epoch": 0.45030655050016133,
      "grad_norm": 1.1691083908081055,
      "learning_rate": 0.0005,
      "loss": 2.7249393463134766,
      "step": 2791
    },
    {
      "epoch": 0.4504678928686673,
      "grad_norm": 1.2223150730133057,
      "learning_rate": 0.0005,
      "loss": 3.0772957801818848,
      "step": 2792
    },
    {
      "epoch": 0.4506292352371733,
      "grad_norm": 1.2242974042892456,
      "learning_rate": 0.0005,
      "loss": 3.283710479736328,
      "step": 2793
    },
    {
      "epoch": 0.45079057760567925,
      "grad_norm": 1.4207556247711182,
      "learning_rate": 0.0005,
      "loss": 2.891571044921875,
      "step": 2794
    },
    {
      "epoch": 0.4509519199741852,
      "grad_norm": 1.2548288106918335,
      "learning_rate": 0.0005,
      "loss": 3.2752718925476074,
      "step": 2795
    },
    {
      "epoch": 0.45111326234269117,
      "grad_norm": 1.3382455110549927,
      "learning_rate": 0.0005,
      "loss": 2.7096598148345947,
      "step": 2796
    },
    {
      "epoch": 0.4512746047111972,
      "grad_norm": 1.2003840208053589,
      "learning_rate": 0.0005,
      "loss": 2.5314879417419434,
      "step": 2797
    },
    {
      "epoch": 0.45143594707970314,
      "grad_norm": 1.1360993385314941,
      "learning_rate": 0.0005,
      "loss": 2.2290520668029785,
      "step": 2798
    },
    {
      "epoch": 0.4515972894482091,
      "grad_norm": 1.6705553531646729,
      "learning_rate": 0.0005,
      "loss": 3.1543967723846436,
      "step": 2799
    },
    {
      "epoch": 0.45175863181671505,
      "grad_norm": 1.6830005645751953,
      "learning_rate": 0.0005,
      "loss": 3.0697693824768066,
      "step": 2800
    },
    {
      "epoch": 0.45191997418522106,
      "grad_norm": 1.0009194612503052,
      "learning_rate": 0.0005,
      "loss": 2.931942939758301,
      "step": 2801
    },
    {
      "epoch": 0.452081316553727,
      "grad_norm": 1.4819494485855103,
      "learning_rate": 0.0005,
      "loss": 3.0885846614837646,
      "step": 2802
    },
    {
      "epoch": 0.452242658922233,
      "grad_norm": 1.3655768632888794,
      "learning_rate": 0.0005,
      "loss": 2.6353421211242676,
      "step": 2803
    },
    {
      "epoch": 0.45240400129073893,
      "grad_norm": 1.2806349992752075,
      "learning_rate": 0.0005,
      "loss": 3.0328378677368164,
      "step": 2804
    },
    {
      "epoch": 0.45256534365924495,
      "grad_norm": 1.220795750617981,
      "learning_rate": 0.0005,
      "loss": 2.7789015769958496,
      "step": 2805
    },
    {
      "epoch": 0.4527266860277509,
      "grad_norm": 1.099022388458252,
      "learning_rate": 0.0005,
      "loss": 3.035607099533081,
      "step": 2806
    },
    {
      "epoch": 0.45288802839625686,
      "grad_norm": 1.206615686416626,
      "learning_rate": 0.0005,
      "loss": 2.8381032943725586,
      "step": 2807
    },
    {
      "epoch": 0.4530493707647628,
      "grad_norm": 1.1222606897354126,
      "learning_rate": 0.0005,
      "loss": 2.5992794036865234,
      "step": 2808
    },
    {
      "epoch": 0.45321071313326877,
      "grad_norm": 1.4070756435394287,
      "learning_rate": 0.0005,
      "loss": 3.3670921325683594,
      "step": 2809
    },
    {
      "epoch": 0.4533720555017748,
      "grad_norm": 1.199668526649475,
      "learning_rate": 0.0005,
      "loss": 2.930044651031494,
      "step": 2810
    },
    {
      "epoch": 0.45353339787028074,
      "grad_norm": 1.2821824550628662,
      "learning_rate": 0.0005,
      "loss": 2.77132511138916,
      "step": 2811
    },
    {
      "epoch": 0.4536947402387867,
      "grad_norm": 1.2918519973754883,
      "learning_rate": 0.0005,
      "loss": 2.9472217559814453,
      "step": 2812
    },
    {
      "epoch": 0.45385608260729265,
      "grad_norm": 1.3052566051483154,
      "learning_rate": 0.0005,
      "loss": 2.9875168800354004,
      "step": 2813
    },
    {
      "epoch": 0.45401742497579867,
      "grad_norm": 1.3318766355514526,
      "learning_rate": 0.0005,
      "loss": 3.014280319213867,
      "step": 2814
    },
    {
      "epoch": 0.4541787673443046,
      "grad_norm": 1.4980446100234985,
      "learning_rate": 0.0005,
      "loss": 2.962679862976074,
      "step": 2815
    },
    {
      "epoch": 0.4543401097128106,
      "grad_norm": 1.1214956045150757,
      "learning_rate": 0.0005,
      "loss": 2.6333560943603516,
      "step": 2816
    },
    {
      "epoch": 0.45450145208131654,
      "grad_norm": 1.4134504795074463,
      "learning_rate": 0.0005,
      "loss": 2.8717048168182373,
      "step": 2817
    },
    {
      "epoch": 0.45466279444982255,
      "grad_norm": 0.9961474537849426,
      "learning_rate": 0.0005,
      "loss": 2.9089505672454834,
      "step": 2818
    },
    {
      "epoch": 0.4548241368183285,
      "grad_norm": 1.2721954584121704,
      "learning_rate": 0.0005,
      "loss": 3.1562089920043945,
      "step": 2819
    },
    {
      "epoch": 0.45498547918683446,
      "grad_norm": 1.3001482486724854,
      "learning_rate": 0.0005,
      "loss": 3.000819206237793,
      "step": 2820
    },
    {
      "epoch": 0.4551468215553404,
      "grad_norm": 1.3662279844284058,
      "learning_rate": 0.0005,
      "loss": 3.1989104747772217,
      "step": 2821
    },
    {
      "epoch": 0.4553081639238464,
      "grad_norm": 1.1528035402297974,
      "learning_rate": 0.0005,
      "loss": 2.7020959854125977,
      "step": 2822
    },
    {
      "epoch": 0.4554695062923524,
      "grad_norm": 2.156069755554199,
      "learning_rate": 0.0005,
      "loss": 3.155872106552124,
      "step": 2823
    },
    {
      "epoch": 0.45563084866085835,
      "grad_norm": 2.7095279693603516,
      "learning_rate": 0.0005,
      "loss": 3.0027427673339844,
      "step": 2824
    },
    {
      "epoch": 0.4557921910293643,
      "grad_norm": 1.2532525062561035,
      "learning_rate": 0.0005,
      "loss": 2.5116519927978516,
      "step": 2825
    },
    {
      "epoch": 0.45595353339787026,
      "grad_norm": 1.4059404134750366,
      "learning_rate": 0.0005,
      "loss": 2.540760040283203,
      "step": 2826
    },
    {
      "epoch": 0.45611487576637627,
      "grad_norm": 1.7325944900512695,
      "learning_rate": 0.0005,
      "loss": 2.7325947284698486,
      "step": 2827
    },
    {
      "epoch": 0.4562762181348822,
      "grad_norm": 1.1394423246383667,
      "learning_rate": 0.0005,
      "loss": 3.1060597896575928,
      "step": 2828
    },
    {
      "epoch": 0.4564375605033882,
      "grad_norm": 2.872399091720581,
      "learning_rate": 0.0005,
      "loss": 3.0583465099334717,
      "step": 2829
    },
    {
      "epoch": 0.45659890287189414,
      "grad_norm": 1.2390954494476318,
      "learning_rate": 0.0005,
      "loss": 2.584014892578125,
      "step": 2830
    },
    {
      "epoch": 0.45676024524040015,
      "grad_norm": 1.5578186511993408,
      "learning_rate": 0.0005,
      "loss": 3.1577696800231934,
      "step": 2831
    },
    {
      "epoch": 0.4569215876089061,
      "grad_norm": 1.1184916496276855,
      "learning_rate": 0.0005,
      "loss": 2.8819141387939453,
      "step": 2832
    },
    {
      "epoch": 0.45708292997741207,
      "grad_norm": 1.1341078281402588,
      "learning_rate": 0.0005,
      "loss": 3.0150694847106934,
      "step": 2833
    },
    {
      "epoch": 0.457244272345918,
      "grad_norm": 1.306192398071289,
      "learning_rate": 0.0005,
      "loss": 2.4993577003479004,
      "step": 2834
    },
    {
      "epoch": 0.457405614714424,
      "grad_norm": 1.6283655166625977,
      "learning_rate": 0.0005,
      "loss": 2.9453394412994385,
      "step": 2835
    },
    {
      "epoch": 0.45756695708293,
      "grad_norm": 1.384259819984436,
      "learning_rate": 0.0005,
      "loss": 3.2296366691589355,
      "step": 2836
    },
    {
      "epoch": 0.45772829945143595,
      "grad_norm": 1.1913124322891235,
      "learning_rate": 0.0005,
      "loss": 2.5415549278259277,
      "step": 2837
    },
    {
      "epoch": 0.4578896418199419,
      "grad_norm": 1.986061930656433,
      "learning_rate": 0.0005,
      "loss": 2.8868296146392822,
      "step": 2838
    },
    {
      "epoch": 0.45805098418844786,
      "grad_norm": 1.4647120237350464,
      "learning_rate": 0.0005,
      "loss": 2.6692543029785156,
      "step": 2839
    },
    {
      "epoch": 0.4582123265569539,
      "grad_norm": 1.0730576515197754,
      "learning_rate": 0.0005,
      "loss": 2.835470676422119,
      "step": 2840
    },
    {
      "epoch": 0.45837366892545983,
      "grad_norm": 1.296392560005188,
      "learning_rate": 0.0005,
      "loss": 3.1558260917663574,
      "step": 2841
    },
    {
      "epoch": 0.4585350112939658,
      "grad_norm": 1.342874526977539,
      "learning_rate": 0.0005,
      "loss": 2.482985734939575,
      "step": 2842
    },
    {
      "epoch": 0.45869635366247175,
      "grad_norm": 1.964586615562439,
      "learning_rate": 0.0005,
      "loss": 3.26594877243042,
      "step": 2843
    },
    {
      "epoch": 0.45885769603097776,
      "grad_norm": 1.1886236667633057,
      "learning_rate": 0.0005,
      "loss": 2.655386447906494,
      "step": 2844
    },
    {
      "epoch": 0.4590190383994837,
      "grad_norm": 1.5896717309951782,
      "learning_rate": 0.0005,
      "loss": 3.0233755111694336,
      "step": 2845
    },
    {
      "epoch": 0.45918038076798967,
      "grad_norm": 1.7124849557876587,
      "learning_rate": 0.0005,
      "loss": 2.9575741291046143,
      "step": 2846
    },
    {
      "epoch": 0.4593417231364956,
      "grad_norm": 1.347375750541687,
      "learning_rate": 0.0005,
      "loss": 3.4146926403045654,
      "step": 2847
    },
    {
      "epoch": 0.45950306550500164,
      "grad_norm": 1.1158781051635742,
      "learning_rate": 0.0005,
      "loss": 3.055098533630371,
      "step": 2848
    },
    {
      "epoch": 0.4596644078735076,
      "grad_norm": 1.7505414485931396,
      "learning_rate": 0.0005,
      "loss": 2.8955302238464355,
      "step": 2849
    },
    {
      "epoch": 0.45982575024201355,
      "grad_norm": 1.3952624797821045,
      "learning_rate": 0.0005,
      "loss": 2.8049583435058594,
      "step": 2850
    },
    {
      "epoch": 0.4599870926105195,
      "grad_norm": 1.3198785781860352,
      "learning_rate": 0.0005,
      "loss": 2.6949987411499023,
      "step": 2851
    },
    {
      "epoch": 0.46014843497902547,
      "grad_norm": 1.4160453081130981,
      "learning_rate": 0.0005,
      "loss": 2.913956642150879,
      "step": 2852
    },
    {
      "epoch": 0.4603097773475315,
      "grad_norm": 1.149835228919983,
      "learning_rate": 0.0005,
      "loss": 2.8344578742980957,
      "step": 2853
    },
    {
      "epoch": 0.46047111971603744,
      "grad_norm": 1.2183107137680054,
      "learning_rate": 0.0005,
      "loss": 2.7327489852905273,
      "step": 2854
    },
    {
      "epoch": 0.4606324620845434,
      "grad_norm": 1.505849003791809,
      "learning_rate": 0.0005,
      "loss": 3.2112321853637695,
      "step": 2855
    },
    {
      "epoch": 0.46079380445304935,
      "grad_norm": 1.2562175989151,
      "learning_rate": 0.0005,
      "loss": 3.206948757171631,
      "step": 2856
    },
    {
      "epoch": 0.46095514682155536,
      "grad_norm": 1.4749640226364136,
      "learning_rate": 0.0005,
      "loss": 2.7101938724517822,
      "step": 2857
    },
    {
      "epoch": 0.4611164891900613,
      "grad_norm": 1.5017211437225342,
      "learning_rate": 0.0005,
      "loss": 2.623725175857544,
      "step": 2858
    },
    {
      "epoch": 0.4612778315585673,
      "grad_norm": 1.1584384441375732,
      "learning_rate": 0.0005,
      "loss": 3.3057422637939453,
      "step": 2859
    },
    {
      "epoch": 0.46143917392707323,
      "grad_norm": 1.4290345907211304,
      "learning_rate": 0.0005,
      "loss": 3.024294853210449,
      "step": 2860
    },
    {
      "epoch": 0.46160051629557924,
      "grad_norm": 1.2376552820205688,
      "learning_rate": 0.0005,
      "loss": 3.017402172088623,
      "step": 2861
    },
    {
      "epoch": 0.4617618586640852,
      "grad_norm": 1.7360076904296875,
      "learning_rate": 0.0005,
      "loss": 2.7518093585968018,
      "step": 2862
    },
    {
      "epoch": 0.46192320103259116,
      "grad_norm": 1.5861172676086426,
      "learning_rate": 0.0005,
      "loss": 3.120027542114258,
      "step": 2863
    },
    {
      "epoch": 0.4620845434010971,
      "grad_norm": 1.3172098398208618,
      "learning_rate": 0.0005,
      "loss": 2.646538496017456,
      "step": 2864
    },
    {
      "epoch": 0.46224588576960307,
      "grad_norm": 1.4044039249420166,
      "learning_rate": 0.0005,
      "loss": 3.0792596340179443,
      "step": 2865
    },
    {
      "epoch": 0.4624072281381091,
      "grad_norm": 1.3260831832885742,
      "learning_rate": 0.0005,
      "loss": 3.464656114578247,
      "step": 2866
    },
    {
      "epoch": 0.46256857050661504,
      "grad_norm": 1.840267539024353,
      "learning_rate": 0.0005,
      "loss": 2.8611528873443604,
      "step": 2867
    },
    {
      "epoch": 0.462729912875121,
      "grad_norm": 1.1803752183914185,
      "learning_rate": 0.0005,
      "loss": 2.549015998840332,
      "step": 2868
    },
    {
      "epoch": 0.46289125524362695,
      "grad_norm": 1.1798930168151855,
      "learning_rate": 0.0005,
      "loss": 2.69635009765625,
      "step": 2869
    },
    {
      "epoch": 0.46305259761213297,
      "grad_norm": 1.143100380897522,
      "learning_rate": 0.0005,
      "loss": 2.9833736419677734,
      "step": 2870
    },
    {
      "epoch": 0.4632139399806389,
      "grad_norm": 1.0142000913619995,
      "learning_rate": 0.0005,
      "loss": 2.7238779067993164,
      "step": 2871
    },
    {
      "epoch": 0.4633752823491449,
      "grad_norm": 1.488213300704956,
      "learning_rate": 0.0005,
      "loss": 2.873546600341797,
      "step": 2872
    },
    {
      "epoch": 0.46353662471765084,
      "grad_norm": 1.083390235900879,
      "learning_rate": 0.0005,
      "loss": 2.974411964416504,
      "step": 2873
    },
    {
      "epoch": 0.46369796708615685,
      "grad_norm": 1.2278282642364502,
      "learning_rate": 0.0005,
      "loss": 2.3347651958465576,
      "step": 2874
    },
    {
      "epoch": 0.4638593094546628,
      "grad_norm": 2.1012866497039795,
      "learning_rate": 0.0005,
      "loss": 2.6533427238464355,
      "step": 2875
    },
    {
      "epoch": 0.46402065182316876,
      "grad_norm": 1.3135449886322021,
      "learning_rate": 0.0005,
      "loss": 2.741137981414795,
      "step": 2876
    },
    {
      "epoch": 0.4641819941916747,
      "grad_norm": 1.0174707174301147,
      "learning_rate": 0.0005,
      "loss": 2.5835041999816895,
      "step": 2877
    },
    {
      "epoch": 0.46434333656018073,
      "grad_norm": 1.1205998659133911,
      "learning_rate": 0.0005,
      "loss": 2.5727131366729736,
      "step": 2878
    },
    {
      "epoch": 0.4645046789286867,
      "grad_norm": 1.5949050188064575,
      "learning_rate": 0.0005,
      "loss": 2.996431350708008,
      "step": 2879
    },
    {
      "epoch": 0.46466602129719264,
      "grad_norm": 1.2274090051651,
      "learning_rate": 0.0005,
      "loss": 2.481602191925049,
      "step": 2880
    },
    {
      "epoch": 0.4648273636656986,
      "grad_norm": 1.5206631422042847,
      "learning_rate": 0.0005,
      "loss": 2.928469181060791,
      "step": 2881
    },
    {
      "epoch": 0.46498870603420456,
      "grad_norm": 1.44096839427948,
      "learning_rate": 0.0005,
      "loss": 3.0648207664489746,
      "step": 2882
    },
    {
      "epoch": 0.46515004840271057,
      "grad_norm": 1.3978477716445923,
      "learning_rate": 0.0005,
      "loss": 2.713628053665161,
      "step": 2883
    },
    {
      "epoch": 0.4653113907712165,
      "grad_norm": 1.3807435035705566,
      "learning_rate": 0.0005,
      "loss": 2.5452842712402344,
      "step": 2884
    },
    {
      "epoch": 0.4654727331397225,
      "grad_norm": 1.232608437538147,
      "learning_rate": 0.0005,
      "loss": 2.7905962467193604,
      "step": 2885
    },
    {
      "epoch": 0.46563407550822844,
      "grad_norm": 1.2556064128875732,
      "learning_rate": 0.0005,
      "loss": 2.8368992805480957,
      "step": 2886
    },
    {
      "epoch": 0.46579541787673445,
      "grad_norm": 1.126968502998352,
      "learning_rate": 0.0005,
      "loss": 2.1957995891571045,
      "step": 2887
    },
    {
      "epoch": 0.4659567602452404,
      "grad_norm": 1.1007319688796997,
      "learning_rate": 0.0005,
      "loss": 2.890366315841675,
      "step": 2888
    },
    {
      "epoch": 0.46611810261374637,
      "grad_norm": 1.2729274034500122,
      "learning_rate": 0.0005,
      "loss": 2.500253677368164,
      "step": 2889
    },
    {
      "epoch": 0.4662794449822523,
      "grad_norm": 1.2741413116455078,
      "learning_rate": 0.0005,
      "loss": 2.8536500930786133,
      "step": 2890
    },
    {
      "epoch": 0.46644078735075833,
      "grad_norm": 1.3608840703964233,
      "learning_rate": 0.0005,
      "loss": 3.29425048828125,
      "step": 2891
    },
    {
      "epoch": 0.4666021297192643,
      "grad_norm": 1.113453984260559,
      "learning_rate": 0.0005,
      "loss": 2.6773338317871094,
      "step": 2892
    },
    {
      "epoch": 0.46676347208777025,
      "grad_norm": 1.4544798135757446,
      "learning_rate": 0.0005,
      "loss": 3.5104007720947266,
      "step": 2893
    },
    {
      "epoch": 0.4669248144562762,
      "grad_norm": 1.7085634469985962,
      "learning_rate": 0.0005,
      "loss": 2.4876253604888916,
      "step": 2894
    },
    {
      "epoch": 0.46708615682478216,
      "grad_norm": 1.5672467947006226,
      "learning_rate": 0.0005,
      "loss": 3.192178249359131,
      "step": 2895
    },
    {
      "epoch": 0.4672474991932882,
      "grad_norm": 1.4596108198165894,
      "learning_rate": 0.0005,
      "loss": 3.098940849304199,
      "step": 2896
    },
    {
      "epoch": 0.46740884156179413,
      "grad_norm": 1.3967911005020142,
      "learning_rate": 0.0005,
      "loss": 2.867961883544922,
      "step": 2897
    },
    {
      "epoch": 0.4675701839303001,
      "grad_norm": 2.2104568481445312,
      "learning_rate": 0.0005,
      "loss": 3.1330761909484863,
      "step": 2898
    },
    {
      "epoch": 0.46773152629880604,
      "grad_norm": 1.4249587059020996,
      "learning_rate": 0.0005,
      "loss": 3.16334867477417,
      "step": 2899
    },
    {
      "epoch": 0.46789286866731206,
      "grad_norm": 1.4308512210845947,
      "learning_rate": 0.0005,
      "loss": 2.7069263458251953,
      "step": 2900
    },
    {
      "epoch": 0.468054211035818,
      "grad_norm": 1.2911667823791504,
      "learning_rate": 0.0005,
      "loss": 2.676790714263916,
      "step": 2901
    },
    {
      "epoch": 0.46821555340432397,
      "grad_norm": 1.2277969121932983,
      "learning_rate": 0.0005,
      "loss": 2.844715118408203,
      "step": 2902
    },
    {
      "epoch": 0.4683768957728299,
      "grad_norm": 1.5194958448410034,
      "learning_rate": 0.0005,
      "loss": 3.033627510070801,
      "step": 2903
    },
    {
      "epoch": 0.46853823814133594,
      "grad_norm": 1.1935127973556519,
      "learning_rate": 0.0005,
      "loss": 2.6566998958587646,
      "step": 2904
    },
    {
      "epoch": 0.4686995805098419,
      "grad_norm": 1.0270153284072876,
      "learning_rate": 0.0005,
      "loss": 2.979747772216797,
      "step": 2905
    },
    {
      "epoch": 0.46886092287834785,
      "grad_norm": 1.5506412982940674,
      "learning_rate": 0.0005,
      "loss": 2.994493007659912,
      "step": 2906
    },
    {
      "epoch": 0.4690222652468538,
      "grad_norm": 1.5762650966644287,
      "learning_rate": 0.0005,
      "loss": 2.9776577949523926,
      "step": 2907
    },
    {
      "epoch": 0.4691836076153598,
      "grad_norm": 1.4513899087905884,
      "learning_rate": 0.0005,
      "loss": 2.9730305671691895,
      "step": 2908
    },
    {
      "epoch": 0.4693449499838658,
      "grad_norm": 1.3297359943389893,
      "learning_rate": 0.0005,
      "loss": 3.1019341945648193,
      "step": 2909
    },
    {
      "epoch": 0.46950629235237173,
      "grad_norm": 1.2120654582977295,
      "learning_rate": 0.0005,
      "loss": 2.8858463764190674,
      "step": 2910
    },
    {
      "epoch": 0.4696676347208777,
      "grad_norm": 1.0338008403778076,
      "learning_rate": 0.0005,
      "loss": 2.1824169158935547,
      "step": 2911
    },
    {
      "epoch": 0.46982897708938365,
      "grad_norm": 1.279533863067627,
      "learning_rate": 0.0005,
      "loss": 3.1647329330444336,
      "step": 2912
    },
    {
      "epoch": 0.46999031945788966,
      "grad_norm": 1.007703423500061,
      "learning_rate": 0.0005,
      "loss": 2.513044834136963,
      "step": 2913
    },
    {
      "epoch": 0.4701516618263956,
      "grad_norm": 1.0753239393234253,
      "learning_rate": 0.0005,
      "loss": 2.4375178813934326,
      "step": 2914
    },
    {
      "epoch": 0.4703130041949016,
      "grad_norm": 1.3013802766799927,
      "learning_rate": 0.0005,
      "loss": 2.8103220462799072,
      "step": 2915
    },
    {
      "epoch": 0.47047434656340753,
      "grad_norm": 1.5017465353012085,
      "learning_rate": 0.0005,
      "loss": 2.991931915283203,
      "step": 2916
    },
    {
      "epoch": 0.47063568893191354,
      "grad_norm": 1.5798600912094116,
      "learning_rate": 0.0005,
      "loss": 3.101199150085449,
      "step": 2917
    },
    {
      "epoch": 0.4707970313004195,
      "grad_norm": 1.4193450212478638,
      "learning_rate": 0.0005,
      "loss": 3.1810202598571777,
      "step": 2918
    },
    {
      "epoch": 0.47095837366892546,
      "grad_norm": 8.90064811706543,
      "learning_rate": 0.0005,
      "loss": 2.670017719268799,
      "step": 2919
    },
    {
      "epoch": 0.4711197160374314,
      "grad_norm": 1.8235715627670288,
      "learning_rate": 0.0005,
      "loss": 3.293335199356079,
      "step": 2920
    },
    {
      "epoch": 0.4712810584059374,
      "grad_norm": 2.5947811603546143,
      "learning_rate": 0.0005,
      "loss": 2.9498181343078613,
      "step": 2921
    },
    {
      "epoch": 0.4714424007744434,
      "grad_norm": 1.3490748405456543,
      "learning_rate": 0.0005,
      "loss": 2.887863874435425,
      "step": 2922
    },
    {
      "epoch": 0.47160374314294934,
      "grad_norm": 1.2409307956695557,
      "learning_rate": 0.0005,
      "loss": 3.3771004676818848,
      "step": 2923
    },
    {
      "epoch": 0.4717650855114553,
      "grad_norm": 1.1582609415054321,
      "learning_rate": 0.0005,
      "loss": 3.131369113922119,
      "step": 2924
    },
    {
      "epoch": 0.47192642787996125,
      "grad_norm": 1.2182387113571167,
      "learning_rate": 0.0005,
      "loss": 3.1682755947113037,
      "step": 2925
    },
    {
      "epoch": 0.47208777024846726,
      "grad_norm": 1.4110175371170044,
      "learning_rate": 0.0005,
      "loss": 3.083688735961914,
      "step": 2926
    },
    {
      "epoch": 0.4722491126169732,
      "grad_norm": 1.1329851150512695,
      "learning_rate": 0.0005,
      "loss": 2.6337666511535645,
      "step": 2927
    },
    {
      "epoch": 0.4724104549854792,
      "grad_norm": 1.2692028284072876,
      "learning_rate": 0.0005,
      "loss": 2.7826390266418457,
      "step": 2928
    },
    {
      "epoch": 0.47257179735398513,
      "grad_norm": 1.3295300006866455,
      "learning_rate": 0.0005,
      "loss": 2.8382065296173096,
      "step": 2929
    },
    {
      "epoch": 0.47273313972249115,
      "grad_norm": 1.4795974493026733,
      "learning_rate": 0.0005,
      "loss": 2.8403310775756836,
      "step": 2930
    },
    {
      "epoch": 0.4728944820909971,
      "grad_norm": 1.3288031816482544,
      "learning_rate": 0.0005,
      "loss": 3.0354695320129395,
      "step": 2931
    },
    {
      "epoch": 0.47305582445950306,
      "grad_norm": 1.2463103532791138,
      "learning_rate": 0.0005,
      "loss": 2.9232282638549805,
      "step": 2932
    },
    {
      "epoch": 0.473217166828009,
      "grad_norm": 1.19490385055542,
      "learning_rate": 0.0005,
      "loss": 2.959048271179199,
      "step": 2933
    },
    {
      "epoch": 0.47337850919651503,
      "grad_norm": 1.2888848781585693,
      "learning_rate": 0.0005,
      "loss": 2.5450985431671143,
      "step": 2934
    },
    {
      "epoch": 0.473539851565021,
      "grad_norm": 2.574958562850952,
      "learning_rate": 0.0005,
      "loss": 2.528287410736084,
      "step": 2935
    },
    {
      "epoch": 0.47370119393352694,
      "grad_norm": 1.3324742317199707,
      "learning_rate": 0.0005,
      "loss": 3.035144805908203,
      "step": 2936
    },
    {
      "epoch": 0.4738625363020329,
      "grad_norm": 1.4000152349472046,
      "learning_rate": 0.0005,
      "loss": 3.054029703140259,
      "step": 2937
    },
    {
      "epoch": 0.47402387867053886,
      "grad_norm": 1.1137312650680542,
      "learning_rate": 0.0005,
      "loss": 2.850884199142456,
      "step": 2938
    },
    {
      "epoch": 0.47418522103904487,
      "grad_norm": 1.1871708631515503,
      "learning_rate": 0.0005,
      "loss": 2.9656074047088623,
      "step": 2939
    },
    {
      "epoch": 0.4743465634075508,
      "grad_norm": 1.211626648902893,
      "learning_rate": 0.0005,
      "loss": 2.9913902282714844,
      "step": 2940
    },
    {
      "epoch": 0.4745079057760568,
      "grad_norm": 1.174285650253296,
      "learning_rate": 0.0005,
      "loss": 2.842031240463257,
      "step": 2941
    },
    {
      "epoch": 0.47466924814456274,
      "grad_norm": 1.4350438117980957,
      "learning_rate": 0.0005,
      "loss": 3.2243118286132812,
      "step": 2942
    },
    {
      "epoch": 0.47483059051306875,
      "grad_norm": 1.354228138923645,
      "learning_rate": 0.0005,
      "loss": 2.668325185775757,
      "step": 2943
    },
    {
      "epoch": 0.4749919328815747,
      "grad_norm": 1.3445215225219727,
      "learning_rate": 0.0005,
      "loss": 3.0269668102264404,
      "step": 2944
    },
    {
      "epoch": 0.47515327525008066,
      "grad_norm": 1.3084750175476074,
      "learning_rate": 0.0005,
      "loss": 2.4055593013763428,
      "step": 2945
    },
    {
      "epoch": 0.4753146176185866,
      "grad_norm": 1.521888256072998,
      "learning_rate": 0.0005,
      "loss": 2.9003660678863525,
      "step": 2946
    },
    {
      "epoch": 0.47547595998709263,
      "grad_norm": 1.3096305131912231,
      "learning_rate": 0.0005,
      "loss": 2.957045316696167,
      "step": 2947
    },
    {
      "epoch": 0.4756373023555986,
      "grad_norm": 1.4158658981323242,
      "learning_rate": 0.0005,
      "loss": 2.463470220565796,
      "step": 2948
    },
    {
      "epoch": 0.47579864472410455,
      "grad_norm": 1.4377110004425049,
      "learning_rate": 0.0005,
      "loss": 2.8891549110412598,
      "step": 2949
    },
    {
      "epoch": 0.4759599870926105,
      "grad_norm": 1.2192997932434082,
      "learning_rate": 0.0005,
      "loss": 3.0084962844848633,
      "step": 2950
    },
    {
      "epoch": 0.4761213294611165,
      "grad_norm": 1.1514029502868652,
      "learning_rate": 0.0005,
      "loss": 2.831376552581787,
      "step": 2951
    },
    {
      "epoch": 0.4762826718296225,
      "grad_norm": 2.133303642272949,
      "learning_rate": 0.0005,
      "loss": 3.2963333129882812,
      "step": 2952
    },
    {
      "epoch": 0.47644401419812843,
      "grad_norm": 1.1599533557891846,
      "learning_rate": 0.0005,
      "loss": 2.6554479598999023,
      "step": 2953
    },
    {
      "epoch": 0.4766053565666344,
      "grad_norm": 1.3069052696228027,
      "learning_rate": 0.0005,
      "loss": 2.7138545513153076,
      "step": 2954
    },
    {
      "epoch": 0.47676669893514034,
      "grad_norm": 1.0409296751022339,
      "learning_rate": 0.0005,
      "loss": 3.0238137245178223,
      "step": 2955
    },
    {
      "epoch": 0.47692804130364636,
      "grad_norm": 1.3583974838256836,
      "learning_rate": 0.0005,
      "loss": 2.283005714416504,
      "step": 2956
    },
    {
      "epoch": 0.4770893836721523,
      "grad_norm": 1.216620683670044,
      "learning_rate": 0.0005,
      "loss": 2.8521132469177246,
      "step": 2957
    },
    {
      "epoch": 0.47725072604065827,
      "grad_norm": 1.5922894477844238,
      "learning_rate": 0.0005,
      "loss": 3.5067601203918457,
      "step": 2958
    },
    {
      "epoch": 0.4774120684091642,
      "grad_norm": 1.2449880838394165,
      "learning_rate": 0.0005,
      "loss": 2.834763526916504,
      "step": 2959
    },
    {
      "epoch": 0.47757341077767024,
      "grad_norm": 1.4887343645095825,
      "learning_rate": 0.0005,
      "loss": 3.2800796031951904,
      "step": 2960
    },
    {
      "epoch": 0.4777347531461762,
      "grad_norm": 1.3435132503509521,
      "learning_rate": 0.0005,
      "loss": 2.6081247329711914,
      "step": 2961
    },
    {
      "epoch": 0.47789609551468215,
      "grad_norm": 1.2964568138122559,
      "learning_rate": 0.0005,
      "loss": 2.805948257446289,
      "step": 2962
    },
    {
      "epoch": 0.4780574378831881,
      "grad_norm": 1.263645887374878,
      "learning_rate": 0.0005,
      "loss": 2.399085521697998,
      "step": 2963
    },
    {
      "epoch": 0.4782187802516941,
      "grad_norm": 1.6251996755599976,
      "learning_rate": 0.0005,
      "loss": 2.8774142265319824,
      "step": 2964
    },
    {
      "epoch": 0.4783801226202001,
      "grad_norm": 1.2214683294296265,
      "learning_rate": 0.0005,
      "loss": 2.3342456817626953,
      "step": 2965
    },
    {
      "epoch": 0.47854146498870603,
      "grad_norm": 1.2237532138824463,
      "learning_rate": 0.0005,
      "loss": 2.951265811920166,
      "step": 2966
    },
    {
      "epoch": 0.478702807357212,
      "grad_norm": 1.061077356338501,
      "learning_rate": 0.0005,
      "loss": 2.821385383605957,
      "step": 2967
    },
    {
      "epoch": 0.47886414972571795,
      "grad_norm": 1.102501392364502,
      "learning_rate": 0.0005,
      "loss": 2.570089340209961,
      "step": 2968
    },
    {
      "epoch": 0.47902549209422396,
      "grad_norm": 1.3067283630371094,
      "learning_rate": 0.0005,
      "loss": 3.2582857608795166,
      "step": 2969
    },
    {
      "epoch": 0.4791868344627299,
      "grad_norm": 1.0734108686447144,
      "learning_rate": 0.0005,
      "loss": 2.9007487297058105,
      "step": 2970
    },
    {
      "epoch": 0.4793481768312359,
      "grad_norm": 1.2130153179168701,
      "learning_rate": 0.0005,
      "loss": 3.0307717323303223,
      "step": 2971
    },
    {
      "epoch": 0.47950951919974183,
      "grad_norm": 1.439435362815857,
      "learning_rate": 0.0005,
      "loss": 2.5248372554779053,
      "step": 2972
    },
    {
      "epoch": 0.47967086156824784,
      "grad_norm": 1.2084373235702515,
      "learning_rate": 0.0005,
      "loss": 3.130436420440674,
      "step": 2973
    },
    {
      "epoch": 0.4798322039367538,
      "grad_norm": 1.7398334741592407,
      "learning_rate": 0.0005,
      "loss": 3.1129350662231445,
      "step": 2974
    },
    {
      "epoch": 0.47999354630525976,
      "grad_norm": 1.2654715776443481,
      "learning_rate": 0.0005,
      "loss": 2.6957716941833496,
      "step": 2975
    },
    {
      "epoch": 0.4801548886737657,
      "grad_norm": 1.2055109739303589,
      "learning_rate": 0.0005,
      "loss": 2.9027628898620605,
      "step": 2976
    },
    {
      "epoch": 0.4803162310422717,
      "grad_norm": 1.2184174060821533,
      "learning_rate": 0.0005,
      "loss": 3.058864116668701,
      "step": 2977
    },
    {
      "epoch": 0.4804775734107777,
      "grad_norm": 1.621392846107483,
      "learning_rate": 0.0005,
      "loss": 3.0271785259246826,
      "step": 2978
    },
    {
      "epoch": 0.48063891577928364,
      "grad_norm": 1.41385817527771,
      "learning_rate": 0.0005,
      "loss": 2.7981796264648438,
      "step": 2979
    },
    {
      "epoch": 0.4808002581477896,
      "grad_norm": 1.2603731155395508,
      "learning_rate": 0.0005,
      "loss": 3.0016696453094482,
      "step": 2980
    },
    {
      "epoch": 0.4809616005162956,
      "grad_norm": 1.4056645631790161,
      "learning_rate": 0.0005,
      "loss": 2.824944019317627,
      "step": 2981
    },
    {
      "epoch": 0.48112294288480156,
      "grad_norm": 1.546054720878601,
      "learning_rate": 0.0005,
      "loss": 2.6580491065979004,
      "step": 2982
    },
    {
      "epoch": 0.4812842852533075,
      "grad_norm": 1.3794279098510742,
      "learning_rate": 0.0005,
      "loss": 2.6446778774261475,
      "step": 2983
    },
    {
      "epoch": 0.4814456276218135,
      "grad_norm": 1.303795337677002,
      "learning_rate": 0.0005,
      "loss": 2.758664608001709,
      "step": 2984
    },
    {
      "epoch": 0.48160696999031943,
      "grad_norm": 1.6784536838531494,
      "learning_rate": 0.0005,
      "loss": 3.1601147651672363,
      "step": 2985
    },
    {
      "epoch": 0.48176831235882545,
      "grad_norm": 1.5952266454696655,
      "learning_rate": 0.0005,
      "loss": 3.06535005569458,
      "step": 2986
    },
    {
      "epoch": 0.4819296547273314,
      "grad_norm": 1.5365574359893799,
      "learning_rate": 0.0005,
      "loss": 2.764765501022339,
      "step": 2987
    },
    {
      "epoch": 0.48209099709583736,
      "grad_norm": 1.6589869260787964,
      "learning_rate": 0.0005,
      "loss": 2.8383452892303467,
      "step": 2988
    },
    {
      "epoch": 0.4822523394643433,
      "grad_norm": 1.2020851373672485,
      "learning_rate": 0.0005,
      "loss": 3.1842873096466064,
      "step": 2989
    },
    {
      "epoch": 0.48241368183284933,
      "grad_norm": 1.1405247449874878,
      "learning_rate": 0.0005,
      "loss": 2.8332059383392334,
      "step": 2990
    },
    {
      "epoch": 0.4825750242013553,
      "grad_norm": 1.3112285137176514,
      "learning_rate": 0.0005,
      "loss": 2.5452518463134766,
      "step": 2991
    },
    {
      "epoch": 0.48273636656986124,
      "grad_norm": 1.2420135736465454,
      "learning_rate": 0.0005,
      "loss": 2.7970895767211914,
      "step": 2992
    },
    {
      "epoch": 0.4828977089383672,
      "grad_norm": 1.0967658758163452,
      "learning_rate": 0.0005,
      "loss": 2.8437323570251465,
      "step": 2993
    },
    {
      "epoch": 0.4830590513068732,
      "grad_norm": 1.129788875579834,
      "learning_rate": 0.0005,
      "loss": 2.8975038528442383,
      "step": 2994
    },
    {
      "epoch": 0.48322039367537917,
      "grad_norm": 1.0299317836761475,
      "learning_rate": 0.0005,
      "loss": 2.3984761238098145,
      "step": 2995
    },
    {
      "epoch": 0.4833817360438851,
      "grad_norm": 1.154751181602478,
      "learning_rate": 0.0005,
      "loss": 3.1395888328552246,
      "step": 2996
    },
    {
      "epoch": 0.4835430784123911,
      "grad_norm": 1.1712549924850464,
      "learning_rate": 0.0005,
      "loss": 2.729562282562256,
      "step": 2997
    },
    {
      "epoch": 0.48370442078089704,
      "grad_norm": 1.5152552127838135,
      "learning_rate": 0.0005,
      "loss": 3.1084699630737305,
      "step": 2998
    },
    {
      "epoch": 0.48386576314940305,
      "grad_norm": 1.1183319091796875,
      "learning_rate": 0.0005,
      "loss": 3.0134992599487305,
      "step": 2999
    },
    {
      "epoch": 0.484027105517909,
      "grad_norm": 2.246180772781372,
      "learning_rate": 0.0005,
      "loss": 3.1693272590637207,
      "step": 3000
    },
    {
      "epoch": 0.48418844788641496,
      "grad_norm": 1.7616894245147705,
      "learning_rate": 0.0005,
      "loss": 2.6715683937072754,
      "step": 3001
    },
    {
      "epoch": 0.4843497902549209,
      "grad_norm": 1.1193206310272217,
      "learning_rate": 0.0005,
      "loss": 3.139380931854248,
      "step": 3002
    },
    {
      "epoch": 0.48451113262342693,
      "grad_norm": 1.289214849472046,
      "learning_rate": 0.0005,
      "loss": 2.5783159732818604,
      "step": 3003
    },
    {
      "epoch": 0.4846724749919329,
      "grad_norm": 1.5763462781906128,
      "learning_rate": 0.0005,
      "loss": 2.8205530643463135,
      "step": 3004
    },
    {
      "epoch": 0.48483381736043885,
      "grad_norm": 1.074110507965088,
      "learning_rate": 0.0005,
      "loss": 2.927365779876709,
      "step": 3005
    },
    {
      "epoch": 0.4849951597289448,
      "grad_norm": 1.120815634727478,
      "learning_rate": 0.0005,
      "loss": 3.0124926567077637,
      "step": 3006
    },
    {
      "epoch": 0.4851565020974508,
      "grad_norm": 1.3375020027160645,
      "learning_rate": 0.0005,
      "loss": 2.825416088104248,
      "step": 3007
    },
    {
      "epoch": 0.48531784446595677,
      "grad_norm": 1.365431308746338,
      "learning_rate": 0.0005,
      "loss": 2.9820823669433594,
      "step": 3008
    },
    {
      "epoch": 0.48547918683446273,
      "grad_norm": 1.2382222414016724,
      "learning_rate": 0.0005,
      "loss": 2.5321850776672363,
      "step": 3009
    },
    {
      "epoch": 0.4856405292029687,
      "grad_norm": 1.119741439819336,
      "learning_rate": 0.0005,
      "loss": 3.1591737270355225,
      "step": 3010
    },
    {
      "epoch": 0.48580187157147464,
      "grad_norm": 1.0144517421722412,
      "learning_rate": 0.0005,
      "loss": 2.993380069732666,
      "step": 3011
    },
    {
      "epoch": 0.48596321393998065,
      "grad_norm": 1.2114337682724,
      "learning_rate": 0.0005,
      "loss": 2.695193290710449,
      "step": 3012
    },
    {
      "epoch": 0.4861245563084866,
      "grad_norm": 1.4813145399093628,
      "learning_rate": 0.0005,
      "loss": 3.003201484680176,
      "step": 3013
    },
    {
      "epoch": 0.48628589867699257,
      "grad_norm": 1.1273741722106934,
      "learning_rate": 0.0005,
      "loss": 3.1964898109436035,
      "step": 3014
    },
    {
      "epoch": 0.4864472410454985,
      "grad_norm": 1.0867241621017456,
      "learning_rate": 0.0005,
      "loss": 2.585756301879883,
      "step": 3015
    },
    {
      "epoch": 0.48660858341400454,
      "grad_norm": 1.2769032716751099,
      "learning_rate": 0.0005,
      "loss": 2.6969308853149414,
      "step": 3016
    },
    {
      "epoch": 0.4867699257825105,
      "grad_norm": 1.3604295253753662,
      "learning_rate": 0.0005,
      "loss": 2.9469094276428223,
      "step": 3017
    },
    {
      "epoch": 0.48693126815101645,
      "grad_norm": 1.5700689554214478,
      "learning_rate": 0.0005,
      "loss": 2.621835947036743,
      "step": 3018
    },
    {
      "epoch": 0.4870926105195224,
      "grad_norm": 1.7971384525299072,
      "learning_rate": 0.0005,
      "loss": 2.824821949005127,
      "step": 3019
    },
    {
      "epoch": 0.4872539528880284,
      "grad_norm": 1.2875564098358154,
      "learning_rate": 0.0005,
      "loss": 3.110016345977783,
      "step": 3020
    },
    {
      "epoch": 0.4874152952565344,
      "grad_norm": 1.1624784469604492,
      "learning_rate": 0.0005,
      "loss": 2.7987732887268066,
      "step": 3021
    },
    {
      "epoch": 0.48757663762504033,
      "grad_norm": 1.106075644493103,
      "learning_rate": 0.0005,
      "loss": 2.7284352779388428,
      "step": 3022
    },
    {
      "epoch": 0.4877379799935463,
      "grad_norm": 1.4568065404891968,
      "learning_rate": 0.0005,
      "loss": 2.9146785736083984,
      "step": 3023
    },
    {
      "epoch": 0.4878993223620523,
      "grad_norm": 1.397938847541809,
      "learning_rate": 0.0005,
      "loss": 3.0642242431640625,
      "step": 3024
    },
    {
      "epoch": 0.48806066473055826,
      "grad_norm": 1.319544792175293,
      "learning_rate": 0.0005,
      "loss": 2.8563857078552246,
      "step": 3025
    },
    {
      "epoch": 0.4882220070990642,
      "grad_norm": 1.0810542106628418,
      "learning_rate": 0.0005,
      "loss": 2.6962716579437256,
      "step": 3026
    },
    {
      "epoch": 0.48838334946757017,
      "grad_norm": 2.5621798038482666,
      "learning_rate": 0.0005,
      "loss": 2.653780460357666,
      "step": 3027
    },
    {
      "epoch": 0.48854469183607613,
      "grad_norm": 3.841545820236206,
      "learning_rate": 0.0005,
      "loss": 2.7981932163238525,
      "step": 3028
    },
    {
      "epoch": 0.48870603420458214,
      "grad_norm": 1.2672841548919678,
      "learning_rate": 0.0005,
      "loss": 3.2826201915740967,
      "step": 3029
    },
    {
      "epoch": 0.4888673765730881,
      "grad_norm": 1.0507770776748657,
      "learning_rate": 0.0005,
      "loss": 2.7900187969207764,
      "step": 3030
    },
    {
      "epoch": 0.48902871894159405,
      "grad_norm": 1.162688970565796,
      "learning_rate": 0.0005,
      "loss": 2.9217216968536377,
      "step": 3031
    },
    {
      "epoch": 0.4891900613101,
      "grad_norm": 1.2301948070526123,
      "learning_rate": 0.0005,
      "loss": 3.3148999214172363,
      "step": 3032
    },
    {
      "epoch": 0.489351403678606,
      "grad_norm": 1.4073081016540527,
      "learning_rate": 0.0005,
      "loss": 2.3707611560821533,
      "step": 3033
    },
    {
      "epoch": 0.489512746047112,
      "grad_norm": 1.3823496103286743,
      "learning_rate": 0.0005,
      "loss": 2.62046217918396,
      "step": 3034
    },
    {
      "epoch": 0.48967408841561794,
      "grad_norm": 1.7864879369735718,
      "learning_rate": 0.0005,
      "loss": 3.222874164581299,
      "step": 3035
    },
    {
      "epoch": 0.4898354307841239,
      "grad_norm": 1.7173436880111694,
      "learning_rate": 0.0005,
      "loss": 3.0944013595581055,
      "step": 3036
    },
    {
      "epoch": 0.4899967731526299,
      "grad_norm": 2.369205951690674,
      "learning_rate": 0.0005,
      "loss": 2.951444149017334,
      "step": 3037
    },
    {
      "epoch": 0.49015811552113586,
      "grad_norm": 1.2203928232192993,
      "learning_rate": 0.0005,
      "loss": 2.865767478942871,
      "step": 3038
    },
    {
      "epoch": 0.4903194578896418,
      "grad_norm": 1.2184878587722778,
      "learning_rate": 0.0005,
      "loss": 3.0441575050354004,
      "step": 3039
    },
    {
      "epoch": 0.4904808002581478,
      "grad_norm": 1.7564862966537476,
      "learning_rate": 0.0005,
      "loss": 3.3369412422180176,
      "step": 3040
    },
    {
      "epoch": 0.49064214262665373,
      "grad_norm": 1.6725213527679443,
      "learning_rate": 0.0005,
      "loss": 3.108250617980957,
      "step": 3041
    },
    {
      "epoch": 0.49080348499515974,
      "grad_norm": 1.7322649955749512,
      "learning_rate": 0.0005,
      "loss": 3.395832061767578,
      "step": 3042
    },
    {
      "epoch": 0.4909648273636657,
      "grad_norm": 1.3819894790649414,
      "learning_rate": 0.0005,
      "loss": 2.444093704223633,
      "step": 3043
    },
    {
      "epoch": 0.49112616973217166,
      "grad_norm": 1.2080482244491577,
      "learning_rate": 0.0005,
      "loss": 2.932694673538208,
      "step": 3044
    },
    {
      "epoch": 0.4912875121006776,
      "grad_norm": 1.003051996231079,
      "learning_rate": 0.0005,
      "loss": 2.6685025691986084,
      "step": 3045
    },
    {
      "epoch": 0.4914488544691836,
      "grad_norm": 1.047318935394287,
      "learning_rate": 0.0005,
      "loss": 2.8229799270629883,
      "step": 3046
    },
    {
      "epoch": 0.4916101968376896,
      "grad_norm": 0.9860501289367676,
      "learning_rate": 0.0005,
      "loss": 2.748812437057495,
      "step": 3047
    },
    {
      "epoch": 0.49177153920619554,
      "grad_norm": 1.4048532247543335,
      "learning_rate": 0.0005,
      "loss": 3.057274103164673,
      "step": 3048
    },
    {
      "epoch": 0.4919328815747015,
      "grad_norm": 1.3658568859100342,
      "learning_rate": 0.0005,
      "loss": 2.8203701972961426,
      "step": 3049
    },
    {
      "epoch": 0.4920942239432075,
      "grad_norm": 1.191063404083252,
      "learning_rate": 0.0005,
      "loss": 2.390857458114624,
      "step": 3050
    },
    {
      "epoch": 0.49225556631171347,
      "grad_norm": 1.285979151725769,
      "learning_rate": 0.0005,
      "loss": 2.8595030307769775,
      "step": 3051
    },
    {
      "epoch": 0.4924169086802194,
      "grad_norm": 1.744713544845581,
      "learning_rate": 0.0005,
      "loss": 2.824737071990967,
      "step": 3052
    },
    {
      "epoch": 0.4925782510487254,
      "grad_norm": 1.1710220575332642,
      "learning_rate": 0.0005,
      "loss": 2.7893941402435303,
      "step": 3053
    },
    {
      "epoch": 0.4927395934172314,
      "grad_norm": 1.4304349422454834,
      "learning_rate": 0.0005,
      "loss": 2.6261191368103027,
      "step": 3054
    },
    {
      "epoch": 0.49290093578573735,
      "grad_norm": 1.3354332447052002,
      "learning_rate": 0.0005,
      "loss": 2.9007694721221924,
      "step": 3055
    },
    {
      "epoch": 0.4930622781542433,
      "grad_norm": 1.4792665243148804,
      "learning_rate": 0.0005,
      "loss": 2.9343395233154297,
      "step": 3056
    },
    {
      "epoch": 0.49322362052274926,
      "grad_norm": 1.185591220855713,
      "learning_rate": 0.0005,
      "loss": 2.6569132804870605,
      "step": 3057
    },
    {
      "epoch": 0.4933849628912552,
      "grad_norm": 1.4242914915084839,
      "learning_rate": 0.0005,
      "loss": 2.5072505474090576,
      "step": 3058
    },
    {
      "epoch": 0.49354630525976123,
      "grad_norm": 15.585787773132324,
      "learning_rate": 0.0005,
      "loss": 2.6014504432678223,
      "step": 3059
    },
    {
      "epoch": 0.4937076476282672,
      "grad_norm": 0.9752811789512634,
      "learning_rate": 0.0005,
      "loss": 2.591238021850586,
      "step": 3060
    },
    {
      "epoch": 0.49386898999677314,
      "grad_norm": 1.7127306461334229,
      "learning_rate": 0.0005,
      "loss": 3.0420475006103516,
      "step": 3061
    },
    {
      "epoch": 0.4940303323652791,
      "grad_norm": 1.0714104175567627,
      "learning_rate": 0.0005,
      "loss": 2.682703971862793,
      "step": 3062
    },
    {
      "epoch": 0.4941916747337851,
      "grad_norm": 1.1625237464904785,
      "learning_rate": 0.0005,
      "loss": 3.0433576107025146,
      "step": 3063
    },
    {
      "epoch": 0.49435301710229107,
      "grad_norm": 1.4402861595153809,
      "learning_rate": 0.0005,
      "loss": 3.210724353790283,
      "step": 3064
    },
    {
      "epoch": 0.494514359470797,
      "grad_norm": 1.369302749633789,
      "learning_rate": 0.0005,
      "loss": 3.028608798980713,
      "step": 3065
    },
    {
      "epoch": 0.494675701839303,
      "grad_norm": 1.0676276683807373,
      "learning_rate": 0.0005,
      "loss": 2.7304491996765137,
      "step": 3066
    },
    {
      "epoch": 0.494837044207809,
      "grad_norm": 1.136577844619751,
      "learning_rate": 0.0005,
      "loss": 3.012756109237671,
      "step": 3067
    },
    {
      "epoch": 0.49499838657631495,
      "grad_norm": 1.9430052042007446,
      "learning_rate": 0.0005,
      "loss": 2.6533291339874268,
      "step": 3068
    },
    {
      "epoch": 0.4951597289448209,
      "grad_norm": 1.4287605285644531,
      "learning_rate": 0.0005,
      "loss": 2.8309054374694824,
      "step": 3069
    },
    {
      "epoch": 0.49532107131332687,
      "grad_norm": 1.3504395484924316,
      "learning_rate": 0.0005,
      "loss": 3.1061458587646484,
      "step": 3070
    },
    {
      "epoch": 0.4954824136818328,
      "grad_norm": 1.188687801361084,
      "learning_rate": 0.0005,
      "loss": 2.401217460632324,
      "step": 3071
    },
    {
      "epoch": 0.49564375605033884,
      "grad_norm": 1.101745367050171,
      "learning_rate": 0.0005,
      "loss": 2.3116631507873535,
      "step": 3072
    },
    {
      "epoch": 0.4958050984188448,
      "grad_norm": 1.3911538124084473,
      "learning_rate": 0.0005,
      "loss": 2.5706729888916016,
      "step": 3073
    },
    {
      "epoch": 0.49596644078735075,
      "grad_norm": 1.495682716369629,
      "learning_rate": 0.0005,
      "loss": 3.0906577110290527,
      "step": 3074
    },
    {
      "epoch": 0.4961277831558567,
      "grad_norm": 1.3674389123916626,
      "learning_rate": 0.0005,
      "loss": 3.043722152709961,
      "step": 3075
    },
    {
      "epoch": 0.4962891255243627,
      "grad_norm": 1.1459801197052002,
      "learning_rate": 0.0005,
      "loss": 2.8395609855651855,
      "step": 3076
    },
    {
      "epoch": 0.4964504678928687,
      "grad_norm": 1.8321616649627686,
      "learning_rate": 0.0005,
      "loss": 2.8130269050598145,
      "step": 3077
    },
    {
      "epoch": 0.49661181026137463,
      "grad_norm": 1.0174264907836914,
      "learning_rate": 0.0005,
      "loss": 2.9045042991638184,
      "step": 3078
    },
    {
      "epoch": 0.4967731526298806,
      "grad_norm": 1.2307212352752686,
      "learning_rate": 0.0005,
      "loss": 2.5361084938049316,
      "step": 3079
    },
    {
      "epoch": 0.4969344949983866,
      "grad_norm": 0.9405462741851807,
      "learning_rate": 0.0005,
      "loss": 2.481006622314453,
      "step": 3080
    },
    {
      "epoch": 0.49709583736689256,
      "grad_norm": 1.354447603225708,
      "learning_rate": 0.0005,
      "loss": 2.5059823989868164,
      "step": 3081
    },
    {
      "epoch": 0.4972571797353985,
      "grad_norm": 1.8631596565246582,
      "learning_rate": 0.0005,
      "loss": 3.2146801948547363,
      "step": 3082
    },
    {
      "epoch": 0.49741852210390447,
      "grad_norm": 1.3233450651168823,
      "learning_rate": 0.0005,
      "loss": 2.9830312728881836,
      "step": 3083
    },
    {
      "epoch": 0.4975798644724105,
      "grad_norm": 1.0968163013458252,
      "learning_rate": 0.0005,
      "loss": 2.6587841510772705,
      "step": 3084
    },
    {
      "epoch": 0.49774120684091644,
      "grad_norm": 1.2321032285690308,
      "learning_rate": 0.0005,
      "loss": 2.6790595054626465,
      "step": 3085
    },
    {
      "epoch": 0.4979025492094224,
      "grad_norm": 1.7584378719329834,
      "learning_rate": 0.0005,
      "loss": 2.7307956218719482,
      "step": 3086
    },
    {
      "epoch": 0.49806389157792835,
      "grad_norm": 1.7704838514328003,
      "learning_rate": 0.0005,
      "loss": 3.0275661945343018,
      "step": 3087
    },
    {
      "epoch": 0.4982252339464343,
      "grad_norm": 1.548384666442871,
      "learning_rate": 0.0005,
      "loss": 2.9033381938934326,
      "step": 3088
    },
    {
      "epoch": 0.4983865763149403,
      "grad_norm": 1.228184700012207,
      "learning_rate": 0.0005,
      "loss": 2.8175101280212402,
      "step": 3089
    },
    {
      "epoch": 0.4985479186834463,
      "grad_norm": 1.227756142616272,
      "learning_rate": 0.0005,
      "loss": 3.009909152984619,
      "step": 3090
    },
    {
      "epoch": 0.49870926105195224,
      "grad_norm": 1.1697410345077515,
      "learning_rate": 0.0005,
      "loss": 2.553443431854248,
      "step": 3091
    },
    {
      "epoch": 0.4988706034204582,
      "grad_norm": 1.2116577625274658,
      "learning_rate": 0.0005,
      "loss": 2.3190226554870605,
      "step": 3092
    },
    {
      "epoch": 0.4990319457889642,
      "grad_norm": 1.1326755285263062,
      "learning_rate": 0.0005,
      "loss": 2.8269400596618652,
      "step": 3093
    },
    {
      "epoch": 0.49919328815747016,
      "grad_norm": 1.125635027885437,
      "learning_rate": 0.0005,
      "loss": 2.764512062072754,
      "step": 3094
    },
    {
      "epoch": 0.4993546305259761,
      "grad_norm": 1.8876850605010986,
      "learning_rate": 0.0005,
      "loss": 2.9201455116271973,
      "step": 3095
    },
    {
      "epoch": 0.4995159728944821,
      "grad_norm": 1.4504270553588867,
      "learning_rate": 0.0005,
      "loss": 2.8671419620513916,
      "step": 3096
    },
    {
      "epoch": 0.4996773152629881,
      "grad_norm": 1.2434617280960083,
      "learning_rate": 0.0005,
      "loss": 3.004587173461914,
      "step": 3097
    },
    {
      "epoch": 0.49983865763149404,
      "grad_norm": 1.546420693397522,
      "learning_rate": 0.0005,
      "loss": 2.9833016395568848,
      "step": 3098
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1573269367218018,
      "learning_rate": 0.0005,
      "loss": 2.8313257694244385,
      "step": 3099
    },
    {
      "epoch": 0.500161342368506,
      "grad_norm": 1.605425477027893,
      "learning_rate": 0.0005,
      "loss": 2.4286599159240723,
      "step": 3100
    },
    {
      "epoch": 0.5003226847370119,
      "grad_norm": 1.9062656164169312,
      "learning_rate": 0.0005,
      "loss": 2.6564900875091553,
      "step": 3101
    },
    {
      "epoch": 0.5004840271055179,
      "grad_norm": 1.4022223949432373,
      "learning_rate": 0.0005,
      "loss": 2.87880802154541,
      "step": 3102
    },
    {
      "epoch": 0.5006453694740238,
      "grad_norm": 1.3247463703155518,
      "learning_rate": 0.0005,
      "loss": 2.8485965728759766,
      "step": 3103
    },
    {
      "epoch": 0.5008067118425299,
      "grad_norm": 1.4347304105758667,
      "learning_rate": 0.0005,
      "loss": 3.0645341873168945,
      "step": 3104
    },
    {
      "epoch": 0.5009680542110359,
      "grad_norm": 1.5208868980407715,
      "learning_rate": 0.0005,
      "loss": 3.1174676418304443,
      "step": 3105
    },
    {
      "epoch": 0.5011293965795418,
      "grad_norm": 1.1843634843826294,
      "learning_rate": 0.0005,
      "loss": 2.8269147872924805,
      "step": 3106
    },
    {
      "epoch": 0.5012907389480478,
      "grad_norm": 1.4298964738845825,
      "learning_rate": 0.0005,
      "loss": 2.9901838302612305,
      "step": 3107
    },
    {
      "epoch": 0.5014520813165537,
      "grad_norm": 1.574588656425476,
      "learning_rate": 0.0005,
      "loss": 3.3955254554748535,
      "step": 3108
    },
    {
      "epoch": 0.5016134236850597,
      "grad_norm": 1.2078664302825928,
      "learning_rate": 0.0005,
      "loss": 2.886547088623047,
      "step": 3109
    },
    {
      "epoch": 0.5017747660535656,
      "grad_norm": 1.6433982849121094,
      "learning_rate": 0.0005,
      "loss": 2.69692325592041,
      "step": 3110
    },
    {
      "epoch": 0.5019361084220716,
      "grad_norm": 1.450730800628662,
      "learning_rate": 0.0005,
      "loss": 2.660036325454712,
      "step": 3111
    },
    {
      "epoch": 0.5020974507905777,
      "grad_norm": 1.5044525861740112,
      "learning_rate": 0.0005,
      "loss": 2.3215491771698,
      "step": 3112
    },
    {
      "epoch": 0.5022587931590836,
      "grad_norm": 1.0957757234573364,
      "learning_rate": 0.0005,
      "loss": 2.740218162536621,
      "step": 3113
    },
    {
      "epoch": 0.5024201355275896,
      "grad_norm": 1.1085797548294067,
      "learning_rate": 0.0005,
      "loss": 3.176438808441162,
      "step": 3114
    },
    {
      "epoch": 0.5025814778960955,
      "grad_norm": 1.3047491312026978,
      "learning_rate": 0.0005,
      "loss": 2.7923882007598877,
      "step": 3115
    },
    {
      "epoch": 0.5027428202646015,
      "grad_norm": 1.243712067604065,
      "learning_rate": 0.0005,
      "loss": 2.604038715362549,
      "step": 3116
    },
    {
      "epoch": 0.5029041626331074,
      "grad_norm": 1.2759231328964233,
      "learning_rate": 0.0005,
      "loss": 3.3586959838867188,
      "step": 3117
    },
    {
      "epoch": 0.5030655050016134,
      "grad_norm": 1.131783127784729,
      "learning_rate": 0.0005,
      "loss": 2.769901752471924,
      "step": 3118
    },
    {
      "epoch": 0.5032268473701194,
      "grad_norm": 1.189190149307251,
      "learning_rate": 0.0005,
      "loss": 2.358769655227661,
      "step": 3119
    },
    {
      "epoch": 0.5033881897386253,
      "grad_norm": 1.4965007305145264,
      "learning_rate": 0.0005,
      "loss": 3.177948474884033,
      "step": 3120
    },
    {
      "epoch": 0.5035495321071314,
      "grad_norm": 1.5343823432922363,
      "learning_rate": 0.0005,
      "loss": 2.6955480575561523,
      "step": 3121
    },
    {
      "epoch": 0.5037108744756373,
      "grad_norm": 1.6010643243789673,
      "learning_rate": 0.0005,
      "loss": 3.027592182159424,
      "step": 3122
    },
    {
      "epoch": 0.5038722168441433,
      "grad_norm": 1.1245092153549194,
      "learning_rate": 0.0005,
      "loss": 2.5464162826538086,
      "step": 3123
    },
    {
      "epoch": 0.5040335592126493,
      "grad_norm": 1.3218104839324951,
      "learning_rate": 0.0005,
      "loss": 3.064751386642456,
      "step": 3124
    },
    {
      "epoch": 0.5041949015811552,
      "grad_norm": 1.5437016487121582,
      "learning_rate": 0.0005,
      "loss": 3.007843494415283,
      "step": 3125
    },
    {
      "epoch": 0.5043562439496612,
      "grad_norm": 2.521967887878418,
      "learning_rate": 0.0005,
      "loss": 3.5187511444091797,
      "step": 3126
    },
    {
      "epoch": 0.5045175863181671,
      "grad_norm": 1.0907719135284424,
      "learning_rate": 0.0005,
      "loss": 2.8387932777404785,
      "step": 3127
    },
    {
      "epoch": 0.5046789286866731,
      "grad_norm": 1.4148029088974,
      "learning_rate": 0.0005,
      "loss": 3.0161516666412354,
      "step": 3128
    },
    {
      "epoch": 0.504840271055179,
      "grad_norm": 1.999545931816101,
      "learning_rate": 0.0005,
      "loss": 2.6137609481811523,
      "step": 3129
    },
    {
      "epoch": 0.5050016134236851,
      "grad_norm": 1.1666659116744995,
      "learning_rate": 0.0005,
      "loss": 2.7662322521209717,
      "step": 3130
    },
    {
      "epoch": 0.5051629557921911,
      "grad_norm": 1.4366977214813232,
      "learning_rate": 0.0005,
      "loss": 2.8649306297302246,
      "step": 3131
    },
    {
      "epoch": 0.505324298160697,
      "grad_norm": 1.108373999595642,
      "learning_rate": 0.0005,
      "loss": 2.6991779804229736,
      "step": 3132
    },
    {
      "epoch": 0.505485640529203,
      "grad_norm": 1.5219976902008057,
      "learning_rate": 0.0005,
      "loss": 3.2984275817871094,
      "step": 3133
    },
    {
      "epoch": 0.5056469828977089,
      "grad_norm": 1.1764541864395142,
      "learning_rate": 0.0005,
      "loss": 2.8838677406311035,
      "step": 3134
    },
    {
      "epoch": 0.5058083252662149,
      "grad_norm": 1.637916088104248,
      "learning_rate": 0.0005,
      "loss": 2.743443250656128,
      "step": 3135
    },
    {
      "epoch": 0.5059696676347208,
      "grad_norm": 1.0258738994598389,
      "learning_rate": 0.0005,
      "loss": 2.9046852588653564,
      "step": 3136
    },
    {
      "epoch": 0.5061310100032268,
      "grad_norm": 1.2553709745407104,
      "learning_rate": 0.0005,
      "loss": 3.2589263916015625,
      "step": 3137
    },
    {
      "epoch": 0.5062923523717329,
      "grad_norm": 1.1962817907333374,
      "learning_rate": 0.0005,
      "loss": 2.878690242767334,
      "step": 3138
    },
    {
      "epoch": 0.5064536947402388,
      "grad_norm": 0.8445382118225098,
      "learning_rate": 0.0005,
      "loss": 2.3618879318237305,
      "step": 3139
    },
    {
      "epoch": 0.5066150371087448,
      "grad_norm": 0.9931569695472717,
      "learning_rate": 0.0005,
      "loss": 2.821892738342285,
      "step": 3140
    },
    {
      "epoch": 0.5067763794772507,
      "grad_norm": 1.7311639785766602,
      "learning_rate": 0.0005,
      "loss": 2.3502144813537598,
      "step": 3141
    },
    {
      "epoch": 0.5069377218457567,
      "grad_norm": 1.353798747062683,
      "learning_rate": 0.0005,
      "loss": 2.505702495574951,
      "step": 3142
    },
    {
      "epoch": 0.5070990642142627,
      "grad_norm": 12.244100570678711,
      "learning_rate": 0.0005,
      "loss": 2.990746021270752,
      "step": 3143
    },
    {
      "epoch": 0.5072604065827686,
      "grad_norm": 1.1583595275878906,
      "learning_rate": 0.0005,
      "loss": 2.767848014831543,
      "step": 3144
    },
    {
      "epoch": 0.5074217489512746,
      "grad_norm": 1.3560656309127808,
      "learning_rate": 0.0005,
      "loss": 2.441027879714966,
      "step": 3145
    },
    {
      "epoch": 0.5075830913197805,
      "grad_norm": 2.0431911945343018,
      "learning_rate": 0.0005,
      "loss": 2.6703009605407715,
      "step": 3146
    },
    {
      "epoch": 0.5077444336882866,
      "grad_norm": 1.5971591472625732,
      "learning_rate": 0.0005,
      "loss": 2.8790135383605957,
      "step": 3147
    },
    {
      "epoch": 0.5079057760567925,
      "grad_norm": 1.4012303352355957,
      "learning_rate": 0.0005,
      "loss": 3.072572708129883,
      "step": 3148
    },
    {
      "epoch": 0.5080671184252985,
      "grad_norm": 1.4662972688674927,
      "learning_rate": 0.0005,
      "loss": 2.957552194595337,
      "step": 3149
    },
    {
      "epoch": 0.5082284607938045,
      "grad_norm": 1.0112133026123047,
      "learning_rate": 0.0005,
      "loss": 2.9625871181488037,
      "step": 3150
    },
    {
      "epoch": 0.5083898031623104,
      "grad_norm": 1.1768195629119873,
      "learning_rate": 0.0005,
      "loss": 2.5712647438049316,
      "step": 3151
    },
    {
      "epoch": 0.5085511455308164,
      "grad_norm": 1.4273124933242798,
      "learning_rate": 0.0005,
      "loss": 2.687218427658081,
      "step": 3152
    },
    {
      "epoch": 0.5087124878993223,
      "grad_norm": 2.257244825363159,
      "learning_rate": 0.0005,
      "loss": 2.7948789596557617,
      "step": 3153
    },
    {
      "epoch": 0.5088738302678283,
      "grad_norm": 1.3027119636535645,
      "learning_rate": 0.0005,
      "loss": 3.168945550918579,
      "step": 3154
    },
    {
      "epoch": 0.5090351726363344,
      "grad_norm": 1.3195654153823853,
      "learning_rate": 0.0005,
      "loss": 2.135969400405884,
      "step": 3155
    },
    {
      "epoch": 0.5091965150048403,
      "grad_norm": 1.4225056171417236,
      "learning_rate": 0.0005,
      "loss": 2.769920825958252,
      "step": 3156
    },
    {
      "epoch": 0.5093578573733463,
      "grad_norm": 1.46295964717865,
      "learning_rate": 0.0005,
      "loss": 2.9274985790252686,
      "step": 3157
    },
    {
      "epoch": 0.5095191997418522,
      "grad_norm": 1.3230788707733154,
      "learning_rate": 0.0005,
      "loss": 2.844411849975586,
      "step": 3158
    },
    {
      "epoch": 0.5096805421103582,
      "grad_norm": 1.4171098470687866,
      "learning_rate": 0.0005,
      "loss": 2.5807294845581055,
      "step": 3159
    },
    {
      "epoch": 0.5098418844788641,
      "grad_norm": 1.5257047414779663,
      "learning_rate": 0.0005,
      "loss": 3.0183725357055664,
      "step": 3160
    },
    {
      "epoch": 0.5100032268473701,
      "grad_norm": 1.6618942022323608,
      "learning_rate": 0.0005,
      "loss": 2.269282341003418,
      "step": 3161
    },
    {
      "epoch": 0.510164569215876,
      "grad_norm": 1.4319610595703125,
      "learning_rate": 0.0005,
      "loss": 3.0274133682250977,
      "step": 3162
    },
    {
      "epoch": 0.510325911584382,
      "grad_norm": 1.144068717956543,
      "learning_rate": 0.0005,
      "loss": 2.7185745239257812,
      "step": 3163
    },
    {
      "epoch": 0.5104872539528881,
      "grad_norm": 1.2068339586257935,
      "learning_rate": 0.0005,
      "loss": 3.01981258392334,
      "step": 3164
    },
    {
      "epoch": 0.510648596321394,
      "grad_norm": 1.3345619440078735,
      "learning_rate": 0.0005,
      "loss": 2.922140598297119,
      "step": 3165
    },
    {
      "epoch": 0.5108099386899,
      "grad_norm": 1.063295841217041,
      "learning_rate": 0.0005,
      "loss": 3.076941967010498,
      "step": 3166
    },
    {
      "epoch": 0.510971281058406,
      "grad_norm": 1.3112066984176636,
      "learning_rate": 0.0005,
      "loss": 2.6323413848876953,
      "step": 3167
    },
    {
      "epoch": 0.5111326234269119,
      "grad_norm": 1.136697769165039,
      "learning_rate": 0.0005,
      "loss": 2.585841178894043,
      "step": 3168
    },
    {
      "epoch": 0.5112939657954179,
      "grad_norm": 1.2844822406768799,
      "learning_rate": 0.0005,
      "loss": 2.821850061416626,
      "step": 3169
    },
    {
      "epoch": 0.5114553081639238,
      "grad_norm": 1.0537421703338623,
      "learning_rate": 0.0005,
      "loss": 2.6236605644226074,
      "step": 3170
    },
    {
      "epoch": 0.5116166505324298,
      "grad_norm": 0.9716245532035828,
      "learning_rate": 0.0005,
      "loss": 2.620220422744751,
      "step": 3171
    },
    {
      "epoch": 0.5117779929009358,
      "grad_norm": 1.4089205265045166,
      "learning_rate": 0.0005,
      "loss": 2.7866129875183105,
      "step": 3172
    },
    {
      "epoch": 0.5119393352694418,
      "grad_norm": 1.1468465328216553,
      "learning_rate": 0.0005,
      "loss": 2.8865714073181152,
      "step": 3173
    },
    {
      "epoch": 0.5121006776379478,
      "grad_norm": 1.203222632408142,
      "learning_rate": 0.0005,
      "loss": 2.6609973907470703,
      "step": 3174
    },
    {
      "epoch": 0.5122620200064537,
      "grad_norm": 1.3074103593826294,
      "learning_rate": 0.0005,
      "loss": 3.1153321266174316,
      "step": 3175
    },
    {
      "epoch": 0.5124233623749597,
      "grad_norm": 1.1768361330032349,
      "learning_rate": 0.0005,
      "loss": 2.880718231201172,
      "step": 3176
    },
    {
      "epoch": 0.5125847047434656,
      "grad_norm": 1.4424419403076172,
      "learning_rate": 0.0005,
      "loss": 2.825214385986328,
      "step": 3177
    },
    {
      "epoch": 0.5127460471119716,
      "grad_norm": 1.5190849304199219,
      "learning_rate": 0.0005,
      "loss": 2.6469831466674805,
      "step": 3178
    },
    {
      "epoch": 0.5129073894804775,
      "grad_norm": 1.0940207242965698,
      "learning_rate": 0.0005,
      "loss": 2.7347888946533203,
      "step": 3179
    },
    {
      "epoch": 0.5130687318489835,
      "grad_norm": 1.4126676321029663,
      "learning_rate": 0.0005,
      "loss": 2.504711627960205,
      "step": 3180
    },
    {
      "epoch": 0.5132300742174896,
      "grad_norm": 1.11753511428833,
      "learning_rate": 0.0005,
      "loss": 2.66796612739563,
      "step": 3181
    },
    {
      "epoch": 0.5133914165859955,
      "grad_norm": 1.6347777843475342,
      "learning_rate": 0.0005,
      "loss": 2.8257718086242676,
      "step": 3182
    },
    {
      "epoch": 0.5135527589545015,
      "grad_norm": 1.4984585046768188,
      "learning_rate": 0.0005,
      "loss": 2.527315616607666,
      "step": 3183
    },
    {
      "epoch": 0.5137141013230074,
      "grad_norm": 1.2309526205062866,
      "learning_rate": 0.0005,
      "loss": 2.834712266921997,
      "step": 3184
    },
    {
      "epoch": 0.5138754436915134,
      "grad_norm": 1.2122554779052734,
      "learning_rate": 0.0005,
      "loss": 2.9654440879821777,
      "step": 3185
    },
    {
      "epoch": 0.5140367860600193,
      "grad_norm": 1.1129215955734253,
      "learning_rate": 0.0005,
      "loss": 2.992177724838257,
      "step": 3186
    },
    {
      "epoch": 0.5141981284285253,
      "grad_norm": 1.2963366508483887,
      "learning_rate": 0.0005,
      "loss": 2.8802342414855957,
      "step": 3187
    },
    {
      "epoch": 0.5143594707970313,
      "grad_norm": 1.3112740516662598,
      "learning_rate": 0.0005,
      "loss": 2.9879655838012695,
      "step": 3188
    },
    {
      "epoch": 0.5145208131655372,
      "grad_norm": 1.4885015487670898,
      "learning_rate": 0.0005,
      "loss": 2.74503493309021,
      "step": 3189
    },
    {
      "epoch": 0.5146821555340433,
      "grad_norm": 1.3349266052246094,
      "learning_rate": 0.0005,
      "loss": 2.5297768115997314,
      "step": 3190
    },
    {
      "epoch": 0.5148434979025492,
      "grad_norm": 1.081287145614624,
      "learning_rate": 0.0005,
      "loss": 2.952847480773926,
      "step": 3191
    },
    {
      "epoch": 0.5150048402710552,
      "grad_norm": 1.4894932508468628,
      "learning_rate": 0.0005,
      "loss": 2.9122512340545654,
      "step": 3192
    },
    {
      "epoch": 0.5151661826395612,
      "grad_norm": 1.1013411283493042,
      "learning_rate": 0.0005,
      "loss": 3.022152900695801,
      "step": 3193
    },
    {
      "epoch": 0.5153275250080671,
      "grad_norm": 1.2273163795471191,
      "learning_rate": 0.0005,
      "loss": 2.5652835369110107,
      "step": 3194
    },
    {
      "epoch": 0.5154888673765731,
      "grad_norm": 1.2192741632461548,
      "learning_rate": 0.0005,
      "loss": 2.680574417114258,
      "step": 3195
    },
    {
      "epoch": 0.515650209745079,
      "grad_norm": 1.2598897218704224,
      "learning_rate": 0.0005,
      "loss": 2.6018829345703125,
      "step": 3196
    },
    {
      "epoch": 0.515811552113585,
      "grad_norm": 1.0986523628234863,
      "learning_rate": 0.0005,
      "loss": 3.02262020111084,
      "step": 3197
    },
    {
      "epoch": 0.515972894482091,
      "grad_norm": 1.1270577907562256,
      "learning_rate": 0.0005,
      "loss": 2.830097198486328,
      "step": 3198
    },
    {
      "epoch": 0.516134236850597,
      "grad_norm": 1.4149030447006226,
      "learning_rate": 0.0005,
      "loss": 2.7791433334350586,
      "step": 3199
    },
    {
      "epoch": 0.516295579219103,
      "grad_norm": 1.8919135332107544,
      "learning_rate": 0.0005,
      "loss": 3.0394976139068604,
      "step": 3200
    },
    {
      "epoch": 0.5164569215876089,
      "grad_norm": 1.060890793800354,
      "learning_rate": 0.0005,
      "loss": 2.4331722259521484,
      "step": 3201
    },
    {
      "epoch": 0.5166182639561149,
      "grad_norm": 1.4834848642349243,
      "learning_rate": 0.0005,
      "loss": 2.330254077911377,
      "step": 3202
    },
    {
      "epoch": 0.5167796063246208,
      "grad_norm": 1.1973990201950073,
      "learning_rate": 0.0005,
      "loss": 2.926335334777832,
      "step": 3203
    },
    {
      "epoch": 0.5169409486931268,
      "grad_norm": 1.6596192121505737,
      "learning_rate": 0.0005,
      "loss": 3.169387102127075,
      "step": 3204
    },
    {
      "epoch": 0.5171022910616327,
      "grad_norm": 1.5260287523269653,
      "learning_rate": 0.0005,
      "loss": 2.8274598121643066,
      "step": 3205
    },
    {
      "epoch": 0.5172636334301387,
      "grad_norm": 1.3700172901153564,
      "learning_rate": 0.0005,
      "loss": 2.759265899658203,
      "step": 3206
    },
    {
      "epoch": 0.5174249757986448,
      "grad_norm": 1.140438199043274,
      "learning_rate": 0.0005,
      "loss": 2.944587230682373,
      "step": 3207
    },
    {
      "epoch": 0.5175863181671507,
      "grad_norm": 1.2562360763549805,
      "learning_rate": 0.0005,
      "loss": 3.1176562309265137,
      "step": 3208
    },
    {
      "epoch": 0.5177476605356567,
      "grad_norm": 2.087918281555176,
      "learning_rate": 0.0005,
      "loss": 2.595494031906128,
      "step": 3209
    },
    {
      "epoch": 0.5179090029041626,
      "grad_norm": 1.185288429260254,
      "learning_rate": 0.0005,
      "loss": 2.817263126373291,
      "step": 3210
    },
    {
      "epoch": 0.5180703452726686,
      "grad_norm": 1.2940934896469116,
      "learning_rate": 0.0005,
      "loss": 2.7708678245544434,
      "step": 3211
    },
    {
      "epoch": 0.5182316876411746,
      "grad_norm": 1.6224843263626099,
      "learning_rate": 0.0005,
      "loss": 2.35408616065979,
      "step": 3212
    },
    {
      "epoch": 0.5183930300096805,
      "grad_norm": 1.847082495689392,
      "learning_rate": 0.0005,
      "loss": 2.626594305038452,
      "step": 3213
    },
    {
      "epoch": 0.5185543723781865,
      "grad_norm": 1.2434415817260742,
      "learning_rate": 0.0005,
      "loss": 3.067528247833252,
      "step": 3214
    },
    {
      "epoch": 0.5187157147466925,
      "grad_norm": 1.2762573957443237,
      "learning_rate": 0.0005,
      "loss": 2.6411638259887695,
      "step": 3215
    },
    {
      "epoch": 0.5188770571151985,
      "grad_norm": 1.2145763635635376,
      "learning_rate": 0.0005,
      "loss": 2.68377685546875,
      "step": 3216
    },
    {
      "epoch": 0.5190383994837044,
      "grad_norm": 1.679675817489624,
      "learning_rate": 0.0005,
      "loss": 3.4044039249420166,
      "step": 3217
    },
    {
      "epoch": 0.5191997418522104,
      "grad_norm": 1.1544095277786255,
      "learning_rate": 0.0005,
      "loss": 2.6753926277160645,
      "step": 3218
    },
    {
      "epoch": 0.5193610842207164,
      "grad_norm": 1.2607871294021606,
      "learning_rate": 0.0005,
      "loss": 2.516444683074951,
      "step": 3219
    },
    {
      "epoch": 0.5195224265892223,
      "grad_norm": 1.285603404045105,
      "learning_rate": 0.0005,
      "loss": 2.9213743209838867,
      "step": 3220
    },
    {
      "epoch": 0.5196837689577283,
      "grad_norm": 1.0905760526657104,
      "learning_rate": 0.0005,
      "loss": 2.780569314956665,
      "step": 3221
    },
    {
      "epoch": 0.5198451113262342,
      "grad_norm": 1.8156092166900635,
      "learning_rate": 0.0005,
      "loss": 2.9915008544921875,
      "step": 3222
    },
    {
      "epoch": 0.5200064536947402,
      "grad_norm": 1.2367315292358398,
      "learning_rate": 0.0005,
      "loss": 2.8326172828674316,
      "step": 3223
    },
    {
      "epoch": 0.5201677960632463,
      "grad_norm": 1.2929192781448364,
      "learning_rate": 0.0005,
      "loss": 3.020516872406006,
      "step": 3224
    },
    {
      "epoch": 0.5203291384317522,
      "grad_norm": 1.1166459321975708,
      "learning_rate": 0.0005,
      "loss": 2.561465263366699,
      "step": 3225
    },
    {
      "epoch": 0.5204904808002582,
      "grad_norm": 1.4219266176223755,
      "learning_rate": 0.0005,
      "loss": 2.5222067832946777,
      "step": 3226
    },
    {
      "epoch": 0.5206518231687641,
      "grad_norm": 1.0043225288391113,
      "learning_rate": 0.0005,
      "loss": 2.291752576828003,
      "step": 3227
    },
    {
      "epoch": 0.5208131655372701,
      "grad_norm": 1.295067548751831,
      "learning_rate": 0.0005,
      "loss": 2.754448175430298,
      "step": 3228
    },
    {
      "epoch": 0.520974507905776,
      "grad_norm": 1.495819330215454,
      "learning_rate": 0.0005,
      "loss": 2.36674165725708,
      "step": 3229
    },
    {
      "epoch": 0.521135850274282,
      "grad_norm": 1.216709017753601,
      "learning_rate": 0.0005,
      "loss": 2.7149338722229004,
      "step": 3230
    },
    {
      "epoch": 0.521297192642788,
      "grad_norm": 1.7027777433395386,
      "learning_rate": 0.0005,
      "loss": 2.8068058490753174,
      "step": 3231
    },
    {
      "epoch": 0.5214585350112939,
      "grad_norm": 1.7144945859909058,
      "learning_rate": 0.0005,
      "loss": 2.512303113937378,
      "step": 3232
    },
    {
      "epoch": 0.5216198773798,
      "grad_norm": 1.4658112525939941,
      "learning_rate": 0.0005,
      "loss": 2.966809034347534,
      "step": 3233
    },
    {
      "epoch": 0.5217812197483059,
      "grad_norm": 0.9763604998588562,
      "learning_rate": 0.0005,
      "loss": 2.696087121963501,
      "step": 3234
    },
    {
      "epoch": 0.5219425621168119,
      "grad_norm": 1.0295965671539307,
      "learning_rate": 0.0005,
      "loss": 2.9674859046936035,
      "step": 3235
    },
    {
      "epoch": 0.5221039044853178,
      "grad_norm": 1.344717264175415,
      "learning_rate": 0.0005,
      "loss": 3.17364239692688,
      "step": 3236
    },
    {
      "epoch": 0.5222652468538238,
      "grad_norm": 0.93802410364151,
      "learning_rate": 0.0005,
      "loss": 2.6697933673858643,
      "step": 3237
    },
    {
      "epoch": 0.5224265892223298,
      "grad_norm": 1.6368696689605713,
      "learning_rate": 0.0005,
      "loss": 2.625648021697998,
      "step": 3238
    },
    {
      "epoch": 0.5225879315908357,
      "grad_norm": 1.5508962869644165,
      "learning_rate": 0.0005,
      "loss": 3.0377421379089355,
      "step": 3239
    },
    {
      "epoch": 0.5227492739593417,
      "grad_norm": 1.1407283544540405,
      "learning_rate": 0.0005,
      "loss": 2.8189291954040527,
      "step": 3240
    },
    {
      "epoch": 0.5229106163278477,
      "grad_norm": 1.0608160495758057,
      "learning_rate": 0.0005,
      "loss": 2.726348400115967,
      "step": 3241
    },
    {
      "epoch": 0.5230719586963537,
      "grad_norm": 1.1865657567977905,
      "learning_rate": 0.0005,
      "loss": 2.8280768394470215,
      "step": 3242
    },
    {
      "epoch": 0.5232333010648597,
      "grad_norm": 0.9775597453117371,
      "learning_rate": 0.0005,
      "loss": 2.8741979598999023,
      "step": 3243
    },
    {
      "epoch": 0.5233946434333656,
      "grad_norm": 1.236153483390808,
      "learning_rate": 0.0005,
      "loss": 2.76102352142334,
      "step": 3244
    },
    {
      "epoch": 0.5235559858018716,
      "grad_norm": 1.4407275915145874,
      "learning_rate": 0.0005,
      "loss": 2.9799394607543945,
      "step": 3245
    },
    {
      "epoch": 0.5237173281703775,
      "grad_norm": 1.396437644958496,
      "learning_rate": 0.0005,
      "loss": 3.038372278213501,
      "step": 3246
    },
    {
      "epoch": 0.5238786705388835,
      "grad_norm": 1.4419384002685547,
      "learning_rate": 0.0005,
      "loss": 2.912287473678589,
      "step": 3247
    },
    {
      "epoch": 0.5240400129073894,
      "grad_norm": 1.8562378883361816,
      "learning_rate": 0.0005,
      "loss": 2.8406336307525635,
      "step": 3248
    },
    {
      "epoch": 0.5242013552758954,
      "grad_norm": 1.3362253904342651,
      "learning_rate": 0.0005,
      "loss": 2.7601253986358643,
      "step": 3249
    },
    {
      "epoch": 0.5243626976444015,
      "grad_norm": 1.5636026859283447,
      "learning_rate": 0.0005,
      "loss": 2.799135208129883,
      "step": 3250
    },
    {
      "epoch": 0.5245240400129074,
      "grad_norm": 1.247008204460144,
      "learning_rate": 0.0005,
      "loss": 2.821441173553467,
      "step": 3251
    },
    {
      "epoch": 0.5246853823814134,
      "grad_norm": 1.8028701543807983,
      "learning_rate": 0.0005,
      "loss": 2.8892385959625244,
      "step": 3252
    },
    {
      "epoch": 0.5248467247499193,
      "grad_norm": 1.215376377105713,
      "learning_rate": 0.0005,
      "loss": 2.737724542617798,
      "step": 3253
    },
    {
      "epoch": 0.5250080671184253,
      "grad_norm": 1.1408414840698242,
      "learning_rate": 0.0005,
      "loss": 2.3939013481140137,
      "step": 3254
    },
    {
      "epoch": 0.5251694094869312,
      "grad_norm": 1.1657183170318604,
      "learning_rate": 0.0005,
      "loss": 2.677053451538086,
      "step": 3255
    },
    {
      "epoch": 0.5253307518554372,
      "grad_norm": 1.2213317155838013,
      "learning_rate": 0.0005,
      "loss": 2.982783317565918,
      "step": 3256
    },
    {
      "epoch": 0.5254920942239432,
      "grad_norm": 1.0156773328781128,
      "learning_rate": 0.0005,
      "loss": 2.295712947845459,
      "step": 3257
    },
    {
      "epoch": 0.5256534365924492,
      "grad_norm": 3.634052038192749,
      "learning_rate": 0.0005,
      "loss": 2.9759037494659424,
      "step": 3258
    },
    {
      "epoch": 0.5258147789609552,
      "grad_norm": 1.29420006275177,
      "learning_rate": 0.0005,
      "loss": 3.032111406326294,
      "step": 3259
    },
    {
      "epoch": 0.5259761213294611,
      "grad_norm": 1.1701102256774902,
      "learning_rate": 0.0005,
      "loss": 2.6917309761047363,
      "step": 3260
    },
    {
      "epoch": 0.5261374636979671,
      "grad_norm": 3.054263114929199,
      "learning_rate": 0.0005,
      "loss": 3.216081380844116,
      "step": 3261
    },
    {
      "epoch": 0.5262988060664731,
      "grad_norm": 1.3028334379196167,
      "learning_rate": 0.0005,
      "loss": 3.2695834636688232,
      "step": 3262
    },
    {
      "epoch": 0.526460148434979,
      "grad_norm": 1.394893765449524,
      "learning_rate": 0.0005,
      "loss": 2.3209614753723145,
      "step": 3263
    },
    {
      "epoch": 0.526621490803485,
      "grad_norm": 1.3584085702896118,
      "learning_rate": 0.0005,
      "loss": 2.7089595794677734,
      "step": 3264
    },
    {
      "epoch": 0.5267828331719909,
      "grad_norm": 1.2442517280578613,
      "learning_rate": 0.0005,
      "loss": 2.816742420196533,
      "step": 3265
    },
    {
      "epoch": 0.5269441755404969,
      "grad_norm": 1.5948988199234009,
      "learning_rate": 0.0005,
      "loss": 2.436277151107788,
      "step": 3266
    },
    {
      "epoch": 0.527105517909003,
      "grad_norm": 1.192642331123352,
      "learning_rate": 0.0005,
      "loss": 3.188534736633301,
      "step": 3267
    },
    {
      "epoch": 0.5272668602775089,
      "grad_norm": 1.1375370025634766,
      "learning_rate": 0.0005,
      "loss": 2.843066453933716,
      "step": 3268
    },
    {
      "epoch": 0.5274282026460149,
      "grad_norm": 1.577014446258545,
      "learning_rate": 0.0005,
      "loss": 3.2260754108428955,
      "step": 3269
    },
    {
      "epoch": 0.5275895450145208,
      "grad_norm": 1.1540026664733887,
      "learning_rate": 0.0005,
      "loss": 2.8131818771362305,
      "step": 3270
    },
    {
      "epoch": 0.5277508873830268,
      "grad_norm": 1.1116559505462646,
      "learning_rate": 0.0005,
      "loss": 2.4633631706237793,
      "step": 3271
    },
    {
      "epoch": 0.5279122297515327,
      "grad_norm": 1.39287269115448,
      "learning_rate": 0.0005,
      "loss": 3.3163833618164062,
      "step": 3272
    },
    {
      "epoch": 0.5280735721200387,
      "grad_norm": 1.0623233318328857,
      "learning_rate": 0.0005,
      "loss": 2.8226027488708496,
      "step": 3273
    },
    {
      "epoch": 0.5282349144885446,
      "grad_norm": 1.1820908784866333,
      "learning_rate": 0.0005,
      "loss": 2.7600579261779785,
      "step": 3274
    },
    {
      "epoch": 0.5283962568570507,
      "grad_norm": 1.1913648843765259,
      "learning_rate": 0.0005,
      "loss": 2.562554359436035,
      "step": 3275
    },
    {
      "epoch": 0.5285575992255567,
      "grad_norm": 1.1664361953735352,
      "learning_rate": 0.0005,
      "loss": 2.7823290824890137,
      "step": 3276
    },
    {
      "epoch": 0.5287189415940626,
      "grad_norm": 1.457578182220459,
      "learning_rate": 0.0005,
      "loss": 2.4935498237609863,
      "step": 3277
    },
    {
      "epoch": 0.5288802839625686,
      "grad_norm": 1.6478639841079712,
      "learning_rate": 0.0005,
      "loss": 2.567258834838867,
      "step": 3278
    },
    {
      "epoch": 0.5290416263310745,
      "grad_norm": 1.3136029243469238,
      "learning_rate": 0.0005,
      "loss": 2.5948946475982666,
      "step": 3279
    },
    {
      "epoch": 0.5292029686995805,
      "grad_norm": 1.2033913135528564,
      "learning_rate": 0.0005,
      "loss": 2.81343936920166,
      "step": 3280
    },
    {
      "epoch": 0.5293643110680865,
      "grad_norm": 1.4668763875961304,
      "learning_rate": 0.0005,
      "loss": 2.7969276905059814,
      "step": 3281
    },
    {
      "epoch": 0.5295256534365924,
      "grad_norm": 1.337954044342041,
      "learning_rate": 0.0005,
      "loss": 2.8869576454162598,
      "step": 3282
    },
    {
      "epoch": 0.5296869958050984,
      "grad_norm": 1.1238998174667358,
      "learning_rate": 0.0005,
      "loss": 2.799825429916382,
      "step": 3283
    },
    {
      "epoch": 0.5298483381736044,
      "grad_norm": 6.1161394119262695,
      "learning_rate": 0.0005,
      "loss": 2.528687000274658,
      "step": 3284
    },
    {
      "epoch": 0.5300096805421104,
      "grad_norm": 1.3528591394424438,
      "learning_rate": 0.0005,
      "loss": 2.8062057495117188,
      "step": 3285
    },
    {
      "epoch": 0.5301710229106164,
      "grad_norm": 1.119889497756958,
      "learning_rate": 0.0005,
      "loss": 2.3647356033325195,
      "step": 3286
    },
    {
      "epoch": 0.5303323652791223,
      "grad_norm": 1.1351889371871948,
      "learning_rate": 0.0005,
      "loss": 2.5898380279541016,
      "step": 3287
    },
    {
      "epoch": 0.5304937076476283,
      "grad_norm": 1.311161994934082,
      "learning_rate": 0.0005,
      "loss": 2.6631555557250977,
      "step": 3288
    },
    {
      "epoch": 0.5306550500161342,
      "grad_norm": 1.6726586818695068,
      "learning_rate": 0.0005,
      "loss": 2.767327308654785,
      "step": 3289
    },
    {
      "epoch": 0.5308163923846402,
      "grad_norm": 1.2873245477676392,
      "learning_rate": 0.0005,
      "loss": 2.8738627433776855,
      "step": 3290
    },
    {
      "epoch": 0.5309777347531461,
      "grad_norm": 1.2107206583023071,
      "learning_rate": 0.0005,
      "loss": 2.8445286750793457,
      "step": 3291
    },
    {
      "epoch": 0.5311390771216521,
      "grad_norm": 1.6140315532684326,
      "learning_rate": 0.0005,
      "loss": 2.520331382751465,
      "step": 3292
    },
    {
      "epoch": 0.5313004194901582,
      "grad_norm": 1.0505390167236328,
      "learning_rate": 0.0005,
      "loss": 2.929009437561035,
      "step": 3293
    },
    {
      "epoch": 0.5314617618586641,
      "grad_norm": 1.3108280897140503,
      "learning_rate": 0.0005,
      "loss": 2.83394193649292,
      "step": 3294
    },
    {
      "epoch": 0.5316231042271701,
      "grad_norm": 1.2637999057769775,
      "learning_rate": 0.0005,
      "loss": 2.9295449256896973,
      "step": 3295
    },
    {
      "epoch": 0.531784446595676,
      "grad_norm": 1.2055338621139526,
      "learning_rate": 0.0005,
      "loss": 2.654684066772461,
      "step": 3296
    },
    {
      "epoch": 0.531945788964182,
      "grad_norm": 1.0008282661437988,
      "learning_rate": 0.0005,
      "loss": 2.9222402572631836,
      "step": 3297
    },
    {
      "epoch": 0.5321071313326879,
      "grad_norm": 1.560360074043274,
      "learning_rate": 0.0005,
      "loss": 2.6167681217193604,
      "step": 3298
    },
    {
      "epoch": 0.5322684737011939,
      "grad_norm": 1.21744966506958,
      "learning_rate": 0.0005,
      "loss": 2.6417975425720215,
      "step": 3299
    },
    {
      "epoch": 0.5324298160696999,
      "grad_norm": 1.2984797954559326,
      "learning_rate": 0.0005,
      "loss": 2.5135464668273926,
      "step": 3300
    },
    {
      "epoch": 0.5325911584382059,
      "grad_norm": 1.2727266550064087,
      "learning_rate": 0.0005,
      "loss": 2.2649435997009277,
      "step": 3301
    },
    {
      "epoch": 0.5327525008067119,
      "grad_norm": 1.1667349338531494,
      "learning_rate": 0.0005,
      "loss": 3.2037713527679443,
      "step": 3302
    },
    {
      "epoch": 0.5329138431752178,
      "grad_norm": 1.4820513725280762,
      "learning_rate": 0.0005,
      "loss": 3.0871691703796387,
      "step": 3303
    },
    {
      "epoch": 0.5330751855437238,
      "grad_norm": 1.3995461463928223,
      "learning_rate": 0.0005,
      "loss": 3.2877817153930664,
      "step": 3304
    },
    {
      "epoch": 0.5332365279122298,
      "grad_norm": 1.2102478742599487,
      "learning_rate": 0.0005,
      "loss": 2.5184500217437744,
      "step": 3305
    },
    {
      "epoch": 0.5333978702807357,
      "grad_norm": 1.50407874584198,
      "learning_rate": 0.0005,
      "loss": 2.8348770141601562,
      "step": 3306
    },
    {
      "epoch": 0.5335592126492417,
      "grad_norm": 1.105975866317749,
      "learning_rate": 0.0005,
      "loss": 2.833768844604492,
      "step": 3307
    },
    {
      "epoch": 0.5337205550177476,
      "grad_norm": 2.3266584873199463,
      "learning_rate": 0.0005,
      "loss": 2.871134042739868,
      "step": 3308
    },
    {
      "epoch": 0.5338818973862536,
      "grad_norm": 0.8832978010177612,
      "learning_rate": 0.0005,
      "loss": 2.9408693313598633,
      "step": 3309
    },
    {
      "epoch": 0.5340432397547596,
      "grad_norm": 1.1064256429672241,
      "learning_rate": 0.0005,
      "loss": 3.065484046936035,
      "step": 3310
    },
    {
      "epoch": 0.5342045821232656,
      "grad_norm": 1.493016242980957,
      "learning_rate": 0.0005,
      "loss": 3.057701349258423,
      "step": 3311
    },
    {
      "epoch": 0.5343659244917716,
      "grad_norm": 1.2795171737670898,
      "learning_rate": 0.0005,
      "loss": 2.6980514526367188,
      "step": 3312
    },
    {
      "epoch": 0.5345272668602775,
      "grad_norm": 1.2965118885040283,
      "learning_rate": 0.0005,
      "loss": 2.5404672622680664,
      "step": 3313
    },
    {
      "epoch": 0.5346886092287835,
      "grad_norm": 1.3087832927703857,
      "learning_rate": 0.0005,
      "loss": 2.6903369426727295,
      "step": 3314
    },
    {
      "epoch": 0.5348499515972894,
      "grad_norm": 1.2146754264831543,
      "learning_rate": 0.0005,
      "loss": 2.714534282684326,
      "step": 3315
    },
    {
      "epoch": 0.5350112939657954,
      "grad_norm": 1.2936787605285645,
      "learning_rate": 0.0005,
      "loss": 3.052542209625244,
      "step": 3316
    },
    {
      "epoch": 0.5351726363343013,
      "grad_norm": 3.8192639350891113,
      "learning_rate": 0.0005,
      "loss": 2.771373748779297,
      "step": 3317
    },
    {
      "epoch": 0.5353339787028074,
      "grad_norm": 1.1944340467453003,
      "learning_rate": 0.0005,
      "loss": 2.631517171859741,
      "step": 3318
    },
    {
      "epoch": 0.5354953210713134,
      "grad_norm": 1.4755702018737793,
      "learning_rate": 0.0005,
      "loss": 2.69053053855896,
      "step": 3319
    },
    {
      "epoch": 0.5356566634398193,
      "grad_norm": 1.272847056388855,
      "learning_rate": 0.0005,
      "loss": 2.8442797660827637,
      "step": 3320
    },
    {
      "epoch": 0.5358180058083253,
      "grad_norm": 1.2332090139389038,
      "learning_rate": 0.0005,
      "loss": 2.970334053039551,
      "step": 3321
    },
    {
      "epoch": 0.5359793481768312,
      "grad_norm": 1.3864290714263916,
      "learning_rate": 0.0005,
      "loss": 2.748687982559204,
      "step": 3322
    },
    {
      "epoch": 0.5361406905453372,
      "grad_norm": 1.5976964235305786,
      "learning_rate": 0.0005,
      "loss": 3.025172710418701,
      "step": 3323
    },
    {
      "epoch": 0.5363020329138432,
      "grad_norm": 1.3461776971817017,
      "learning_rate": 0.0005,
      "loss": 2.7624316215515137,
      "step": 3324
    },
    {
      "epoch": 0.5364633752823491,
      "grad_norm": 0.9587190747261047,
      "learning_rate": 0.0005,
      "loss": 2.4083900451660156,
      "step": 3325
    },
    {
      "epoch": 0.5366247176508551,
      "grad_norm": 1.001613736152649,
      "learning_rate": 0.0005,
      "loss": 2.7486493587493896,
      "step": 3326
    },
    {
      "epoch": 0.5367860600193611,
      "grad_norm": 1.2049150466918945,
      "learning_rate": 0.0005,
      "loss": 2.634995698928833,
      "step": 3327
    },
    {
      "epoch": 0.5369474023878671,
      "grad_norm": 1.3583260774612427,
      "learning_rate": 0.0005,
      "loss": 2.630065679550171,
      "step": 3328
    },
    {
      "epoch": 0.537108744756373,
      "grad_norm": 1.8159282207489014,
      "learning_rate": 0.0005,
      "loss": 3.425779342651367,
      "step": 3329
    },
    {
      "epoch": 0.537270087124879,
      "grad_norm": 1.9975683689117432,
      "learning_rate": 0.0005,
      "loss": 3.465179443359375,
      "step": 3330
    },
    {
      "epoch": 0.537431429493385,
      "grad_norm": 1.5932191610336304,
      "learning_rate": 0.0005,
      "loss": 2.505188226699829,
      "step": 3331
    },
    {
      "epoch": 0.5375927718618909,
      "grad_norm": 1.4421063661575317,
      "learning_rate": 0.0005,
      "loss": 2.712698459625244,
      "step": 3332
    },
    {
      "epoch": 0.5377541142303969,
      "grad_norm": 3.8379905223846436,
      "learning_rate": 0.0005,
      "loss": 2.4564571380615234,
      "step": 3333
    },
    {
      "epoch": 0.5379154565989028,
      "grad_norm": 1.4593054056167603,
      "learning_rate": 0.0005,
      "loss": 2.567378520965576,
      "step": 3334
    },
    {
      "epoch": 0.5380767989674088,
      "grad_norm": 1.581032395362854,
      "learning_rate": 0.0005,
      "loss": 2.807432174682617,
      "step": 3335
    },
    {
      "epoch": 0.5382381413359149,
      "grad_norm": 1.498443603515625,
      "learning_rate": 0.0005,
      "loss": 2.7955026626586914,
      "step": 3336
    },
    {
      "epoch": 0.5383994837044208,
      "grad_norm": 1.3056950569152832,
      "learning_rate": 0.0005,
      "loss": 2.5889782905578613,
      "step": 3337
    },
    {
      "epoch": 0.5385608260729268,
      "grad_norm": 1.1068379878997803,
      "learning_rate": 0.0005,
      "loss": 2.832980155944824,
      "step": 3338
    },
    {
      "epoch": 0.5387221684414327,
      "grad_norm": 1.2540128231048584,
      "learning_rate": 0.0005,
      "loss": 2.886960506439209,
      "step": 3339
    },
    {
      "epoch": 0.5388835108099387,
      "grad_norm": 1.1224907636642456,
      "learning_rate": 0.0005,
      "loss": 2.9143130779266357,
      "step": 3340
    },
    {
      "epoch": 0.5390448531784446,
      "grad_norm": 1.239626169204712,
      "learning_rate": 0.0005,
      "loss": 2.888887643814087,
      "step": 3341
    },
    {
      "epoch": 0.5392061955469506,
      "grad_norm": 1.6021184921264648,
      "learning_rate": 0.0005,
      "loss": 2.9187850952148438,
      "step": 3342
    },
    {
      "epoch": 0.5393675379154566,
      "grad_norm": 1.2692407369613647,
      "learning_rate": 0.0005,
      "loss": 2.665191173553467,
      "step": 3343
    },
    {
      "epoch": 0.5395288802839626,
      "grad_norm": 1.4038934707641602,
      "learning_rate": 0.0005,
      "loss": 2.697209358215332,
      "step": 3344
    },
    {
      "epoch": 0.5396902226524686,
      "grad_norm": 1.1990125179290771,
      "learning_rate": 0.0005,
      "loss": 2.568769931793213,
      "step": 3345
    },
    {
      "epoch": 0.5398515650209745,
      "grad_norm": 1.2228409051895142,
      "learning_rate": 0.0005,
      "loss": 2.4583566188812256,
      "step": 3346
    },
    {
      "epoch": 0.5400129073894805,
      "grad_norm": 1.192710041999817,
      "learning_rate": 0.0005,
      "loss": 2.494072198867798,
      "step": 3347
    },
    {
      "epoch": 0.5401742497579864,
      "grad_norm": 1.0596518516540527,
      "learning_rate": 0.0005,
      "loss": 2.7782182693481445,
      "step": 3348
    },
    {
      "epoch": 0.5403355921264924,
      "grad_norm": 1.167181372642517,
      "learning_rate": 0.0005,
      "loss": 2.5494022369384766,
      "step": 3349
    },
    {
      "epoch": 0.5404969344949984,
      "grad_norm": 1.2604554891586304,
      "learning_rate": 0.0005,
      "loss": 3.214999198913574,
      "step": 3350
    },
    {
      "epoch": 0.5406582768635043,
      "grad_norm": 1.3505085706710815,
      "learning_rate": 0.0005,
      "loss": 2.613910675048828,
      "step": 3351
    },
    {
      "epoch": 0.5408196192320103,
      "grad_norm": 1.123223066329956,
      "learning_rate": 0.0005,
      "loss": 2.6918835639953613,
      "step": 3352
    },
    {
      "epoch": 0.5409809616005163,
      "grad_norm": 1.213391900062561,
      "learning_rate": 0.0005,
      "loss": 2.4880623817443848,
      "step": 3353
    },
    {
      "epoch": 0.5411423039690223,
      "grad_norm": 1.207493543624878,
      "learning_rate": 0.0005,
      "loss": 2.4335291385650635,
      "step": 3354
    },
    {
      "epoch": 0.5413036463375283,
      "grad_norm": 1.3486568927764893,
      "learning_rate": 0.0005,
      "loss": 3.1264381408691406,
      "step": 3355
    },
    {
      "epoch": 0.5414649887060342,
      "grad_norm": 1.3881405591964722,
      "learning_rate": 0.0005,
      "loss": 3.0122337341308594,
      "step": 3356
    },
    {
      "epoch": 0.5416263310745402,
      "grad_norm": 1.2779858112335205,
      "learning_rate": 0.0005,
      "loss": 2.731839418411255,
      "step": 3357
    },
    {
      "epoch": 0.5417876734430461,
      "grad_norm": 0.9886285066604614,
      "learning_rate": 0.0005,
      "loss": 2.5917067527770996,
      "step": 3358
    },
    {
      "epoch": 0.5419490158115521,
      "grad_norm": 1.3690906763076782,
      "learning_rate": 0.0005,
      "loss": 2.413708448410034,
      "step": 3359
    },
    {
      "epoch": 0.542110358180058,
      "grad_norm": 1.1125462055206299,
      "learning_rate": 0.0005,
      "loss": 2.5773754119873047,
      "step": 3360
    },
    {
      "epoch": 0.5422717005485641,
      "grad_norm": 2.4122314453125,
      "learning_rate": 0.0005,
      "loss": 2.9367074966430664,
      "step": 3361
    },
    {
      "epoch": 0.5424330429170701,
      "grad_norm": 1.3637651205062866,
      "learning_rate": 0.0005,
      "loss": 2.863447666168213,
      "step": 3362
    },
    {
      "epoch": 0.542594385285576,
      "grad_norm": 1.6279923915863037,
      "learning_rate": 0.0005,
      "loss": 2.3803224563598633,
      "step": 3363
    },
    {
      "epoch": 0.542755727654082,
      "grad_norm": 1.309301733970642,
      "learning_rate": 0.0005,
      "loss": 3.3292503356933594,
      "step": 3364
    },
    {
      "epoch": 0.5429170700225879,
      "grad_norm": 1.3535754680633545,
      "learning_rate": 0.0005,
      "loss": 2.8246636390686035,
      "step": 3365
    },
    {
      "epoch": 0.5430784123910939,
      "grad_norm": 1.243022084236145,
      "learning_rate": 0.0005,
      "loss": 2.837094306945801,
      "step": 3366
    },
    {
      "epoch": 0.5432397547595998,
      "grad_norm": 1.5051480531692505,
      "learning_rate": 0.0005,
      "loss": 2.613368511199951,
      "step": 3367
    },
    {
      "epoch": 0.5434010971281058,
      "grad_norm": 1.085871696472168,
      "learning_rate": 0.0005,
      "loss": 2.6004531383514404,
      "step": 3368
    },
    {
      "epoch": 0.5435624394966118,
      "grad_norm": 1.6559597253799438,
      "learning_rate": 0.0005,
      "loss": 2.600193977355957,
      "step": 3369
    },
    {
      "epoch": 0.5437237818651178,
      "grad_norm": 1.2424598932266235,
      "learning_rate": 0.0005,
      "loss": 2.6302294731140137,
      "step": 3370
    },
    {
      "epoch": 0.5438851242336238,
      "grad_norm": 1.117038369178772,
      "learning_rate": 0.0005,
      "loss": 2.9182868003845215,
      "step": 3371
    },
    {
      "epoch": 0.5440464666021297,
      "grad_norm": 1.3318971395492554,
      "learning_rate": 0.0005,
      "loss": 3.430354595184326,
      "step": 3372
    },
    {
      "epoch": 0.5442078089706357,
      "grad_norm": 2.2033193111419678,
      "learning_rate": 0.0005,
      "loss": 2.9677977561950684,
      "step": 3373
    },
    {
      "epoch": 0.5443691513391417,
      "grad_norm": 1.2138813734054565,
      "learning_rate": 0.0005,
      "loss": 2.664933681488037,
      "step": 3374
    },
    {
      "epoch": 0.5445304937076476,
      "grad_norm": 1.2128117084503174,
      "learning_rate": 0.0005,
      "loss": 2.330501079559326,
      "step": 3375
    },
    {
      "epoch": 0.5446918360761536,
      "grad_norm": 1.6107168197631836,
      "learning_rate": 0.0005,
      "loss": 2.6044278144836426,
      "step": 3376
    },
    {
      "epoch": 0.5448531784446595,
      "grad_norm": 1.6778892278671265,
      "learning_rate": 0.0005,
      "loss": 2.957111358642578,
      "step": 3377
    },
    {
      "epoch": 0.5450145208131656,
      "grad_norm": 1.0697147846221924,
      "learning_rate": 0.0005,
      "loss": 2.990593433380127,
      "step": 3378
    },
    {
      "epoch": 0.5451758631816715,
      "grad_norm": 1.4322525262832642,
      "learning_rate": 0.0005,
      "loss": 2.6859588623046875,
      "step": 3379
    },
    {
      "epoch": 0.5453372055501775,
      "grad_norm": 1.2808524370193481,
      "learning_rate": 0.0005,
      "loss": 2.8160300254821777,
      "step": 3380
    },
    {
      "epoch": 0.5454985479186835,
      "grad_norm": 1.2147167921066284,
      "learning_rate": 0.0005,
      "loss": 2.252371072769165,
      "step": 3381
    },
    {
      "epoch": 0.5456598902871894,
      "grad_norm": 1.14993155002594,
      "learning_rate": 0.0005,
      "loss": 2.75828218460083,
      "step": 3382
    },
    {
      "epoch": 0.5458212326556954,
      "grad_norm": 1.1108397245407104,
      "learning_rate": 0.0005,
      "loss": 2.9656877517700195,
      "step": 3383
    },
    {
      "epoch": 0.5459825750242013,
      "grad_norm": 1.2816150188446045,
      "learning_rate": 0.0005,
      "loss": 2.860128402709961,
      "step": 3384
    },
    {
      "epoch": 0.5461439173927073,
      "grad_norm": 2.340909481048584,
      "learning_rate": 0.0005,
      "loss": 2.7014551162719727,
      "step": 3385
    },
    {
      "epoch": 0.5463052597612132,
      "grad_norm": 1.2355811595916748,
      "learning_rate": 0.0005,
      "loss": 3.160294771194458,
      "step": 3386
    },
    {
      "epoch": 0.5464666021297193,
      "grad_norm": 1.5961135625839233,
      "learning_rate": 0.0005,
      "loss": 2.1645827293395996,
      "step": 3387
    },
    {
      "epoch": 0.5466279444982253,
      "grad_norm": 1.00643789768219,
      "learning_rate": 0.0005,
      "loss": 2.765784502029419,
      "step": 3388
    },
    {
      "epoch": 0.5467892868667312,
      "grad_norm": 1.2869744300842285,
      "learning_rate": 0.0005,
      "loss": 2.767308235168457,
      "step": 3389
    },
    {
      "epoch": 0.5469506292352372,
      "grad_norm": 0.9845778346061707,
      "learning_rate": 0.0005,
      "loss": 3.3052196502685547,
      "step": 3390
    },
    {
      "epoch": 0.5471119716037431,
      "grad_norm": 1.0797312259674072,
      "learning_rate": 0.0005,
      "loss": 2.9972877502441406,
      "step": 3391
    },
    {
      "epoch": 0.5472733139722491,
      "grad_norm": 1.2149786949157715,
      "learning_rate": 0.0005,
      "loss": 2.8500030040740967,
      "step": 3392
    },
    {
      "epoch": 0.547434656340755,
      "grad_norm": 1.1545326709747314,
      "learning_rate": 0.0005,
      "loss": 2.63375186920166,
      "step": 3393
    },
    {
      "epoch": 0.547595998709261,
      "grad_norm": 1.4454933404922485,
      "learning_rate": 0.0005,
      "loss": 2.614121913909912,
      "step": 3394
    },
    {
      "epoch": 0.547757341077767,
      "grad_norm": 1.1696524620056152,
      "learning_rate": 0.0005,
      "loss": 3.0529730319976807,
      "step": 3395
    },
    {
      "epoch": 0.547918683446273,
      "grad_norm": 1.1072715520858765,
      "learning_rate": 0.0005,
      "loss": 2.7420995235443115,
      "step": 3396
    },
    {
      "epoch": 0.548080025814779,
      "grad_norm": 1.3719414472579956,
      "learning_rate": 0.0005,
      "loss": 2.994162082672119,
      "step": 3397
    },
    {
      "epoch": 0.548241368183285,
      "grad_norm": 1.2859573364257812,
      "learning_rate": 0.0005,
      "loss": 3.046480178833008,
      "step": 3398
    },
    {
      "epoch": 0.5484027105517909,
      "grad_norm": 1.2156373262405396,
      "learning_rate": 0.0005,
      "loss": 2.900566339492798,
      "step": 3399
    },
    {
      "epoch": 0.5485640529202969,
      "grad_norm": 1.224384069442749,
      "learning_rate": 0.0005,
      "loss": 2.8131890296936035,
      "step": 3400
    },
    {
      "epoch": 0.5487253952888028,
      "grad_norm": 1.3895516395568848,
      "learning_rate": 0.0005,
      "loss": 2.6686575412750244,
      "step": 3401
    },
    {
      "epoch": 0.5488867376573088,
      "grad_norm": 1.2940168380737305,
      "learning_rate": 0.0005,
      "loss": 2.951810836791992,
      "step": 3402
    },
    {
      "epoch": 0.5490480800258147,
      "grad_norm": 1.2625255584716797,
      "learning_rate": 0.0005,
      "loss": 2.7217764854431152,
      "step": 3403
    },
    {
      "epoch": 0.5492094223943208,
      "grad_norm": 1.220548391342163,
      "learning_rate": 0.0005,
      "loss": 2.757481098175049,
      "step": 3404
    },
    {
      "epoch": 0.5493707647628268,
      "grad_norm": 1.6542153358459473,
      "learning_rate": 0.0005,
      "loss": 2.831051826477051,
      "step": 3405
    },
    {
      "epoch": 0.5495321071313327,
      "grad_norm": 1.2208657264709473,
      "learning_rate": 0.0005,
      "loss": 3.4160592555999756,
      "step": 3406
    },
    {
      "epoch": 0.5496934494998387,
      "grad_norm": 1.0424458980560303,
      "learning_rate": 0.0005,
      "loss": 2.490525722503662,
      "step": 3407
    },
    {
      "epoch": 0.5498547918683446,
      "grad_norm": 1.2938555479049683,
      "learning_rate": 0.0005,
      "loss": 2.8972558975219727,
      "step": 3408
    },
    {
      "epoch": 0.5500161342368506,
      "grad_norm": 1.5270365476608276,
      "learning_rate": 0.0005,
      "loss": 2.625990867614746,
      "step": 3409
    },
    {
      "epoch": 0.5501774766053565,
      "grad_norm": 1.257349967956543,
      "learning_rate": 0.0005,
      "loss": 2.774077892303467,
      "step": 3410
    },
    {
      "epoch": 0.5503388189738625,
      "grad_norm": 1.1858822107315063,
      "learning_rate": 0.0005,
      "loss": 2.872443675994873,
      "step": 3411
    },
    {
      "epoch": 0.5505001613423685,
      "grad_norm": 1.1606504917144775,
      "learning_rate": 0.0005,
      "loss": 2.7246413230895996,
      "step": 3412
    },
    {
      "epoch": 0.5506615037108745,
      "grad_norm": 1.2188892364501953,
      "learning_rate": 0.0005,
      "loss": 2.948793888092041,
      "step": 3413
    },
    {
      "epoch": 0.5508228460793805,
      "grad_norm": 1.2485363483428955,
      "learning_rate": 0.0005,
      "loss": 2.5450055599212646,
      "step": 3414
    },
    {
      "epoch": 0.5509841884478864,
      "grad_norm": 1.2846084833145142,
      "learning_rate": 0.0005,
      "loss": 2.973637342453003,
      "step": 3415
    },
    {
      "epoch": 0.5511455308163924,
      "grad_norm": 1.124588966369629,
      "learning_rate": 0.0005,
      "loss": 2.6953744888305664,
      "step": 3416
    },
    {
      "epoch": 0.5513068731848983,
      "grad_norm": 1.0811762809753418,
      "learning_rate": 0.0005,
      "loss": 2.335486888885498,
      "step": 3417
    },
    {
      "epoch": 0.5514682155534043,
      "grad_norm": 1.0710573196411133,
      "learning_rate": 0.0005,
      "loss": 2.7618443965911865,
      "step": 3418
    },
    {
      "epoch": 0.5516295579219103,
      "grad_norm": 0.9337418675422668,
      "learning_rate": 0.0005,
      "loss": 2.443209409713745,
      "step": 3419
    },
    {
      "epoch": 0.5517909002904162,
      "grad_norm": 1.2118054628372192,
      "learning_rate": 0.0005,
      "loss": 2.5820913314819336,
      "step": 3420
    },
    {
      "epoch": 0.5519522426589223,
      "grad_norm": 1.3286494016647339,
      "learning_rate": 0.0005,
      "loss": 3.1548752784729004,
      "step": 3421
    },
    {
      "epoch": 0.5521135850274282,
      "grad_norm": 1.1202038526535034,
      "learning_rate": 0.0005,
      "loss": 2.4649510383605957,
      "step": 3422
    },
    {
      "epoch": 0.5522749273959342,
      "grad_norm": 1.2035831212997437,
      "learning_rate": 0.0005,
      "loss": 2.967128276824951,
      "step": 3423
    },
    {
      "epoch": 0.5524362697644402,
      "grad_norm": 1.3175523281097412,
      "learning_rate": 0.0005,
      "loss": 2.508633613586426,
      "step": 3424
    },
    {
      "epoch": 0.5525976121329461,
      "grad_norm": 1.5188161134719849,
      "learning_rate": 0.0005,
      "loss": 3.1149282455444336,
      "step": 3425
    },
    {
      "epoch": 0.5527589545014521,
      "grad_norm": 1.0957828760147095,
      "learning_rate": 0.0005,
      "loss": 2.3381943702697754,
      "step": 3426
    },
    {
      "epoch": 0.552920296869958,
      "grad_norm": 1.2566639184951782,
      "learning_rate": 0.0005,
      "loss": 2.8473377227783203,
      "step": 3427
    },
    {
      "epoch": 0.553081639238464,
      "grad_norm": 1.3422421216964722,
      "learning_rate": 0.0005,
      "loss": 2.68302059173584,
      "step": 3428
    },
    {
      "epoch": 0.5532429816069699,
      "grad_norm": 1.2721519470214844,
      "learning_rate": 0.0005,
      "loss": 3.139410972595215,
      "step": 3429
    },
    {
      "epoch": 0.553404323975476,
      "grad_norm": 1.2927900552749634,
      "learning_rate": 0.0005,
      "loss": 3.031583309173584,
      "step": 3430
    },
    {
      "epoch": 0.553565666343982,
      "grad_norm": 1.2701001167297363,
      "learning_rate": 0.0005,
      "loss": 2.621140480041504,
      "step": 3431
    },
    {
      "epoch": 0.5537270087124879,
      "grad_norm": 1.1782350540161133,
      "learning_rate": 0.0005,
      "loss": 2.615863561630249,
      "step": 3432
    },
    {
      "epoch": 0.5538883510809939,
      "grad_norm": 1.3131927251815796,
      "learning_rate": 0.0005,
      "loss": 2.834084987640381,
      "step": 3433
    },
    {
      "epoch": 0.5540496934494998,
      "grad_norm": 1.1718801259994507,
      "learning_rate": 0.0005,
      "loss": 2.730562210083008,
      "step": 3434
    },
    {
      "epoch": 0.5542110358180058,
      "grad_norm": 1.1087995767593384,
      "learning_rate": 0.0005,
      "loss": 2.730349063873291,
      "step": 3435
    },
    {
      "epoch": 0.5543723781865117,
      "grad_norm": 1.140883445739746,
      "learning_rate": 0.0005,
      "loss": 2.9301114082336426,
      "step": 3436
    },
    {
      "epoch": 0.5545337205550177,
      "grad_norm": 1.3168308734893799,
      "learning_rate": 0.0005,
      "loss": 2.6607186794281006,
      "step": 3437
    },
    {
      "epoch": 0.5546950629235237,
      "grad_norm": 1.0159552097320557,
      "learning_rate": 0.0005,
      "loss": 2.706223726272583,
      "step": 3438
    },
    {
      "epoch": 0.5548564052920297,
      "grad_norm": 1.1266120672225952,
      "learning_rate": 0.0005,
      "loss": 2.474461078643799,
      "step": 3439
    },
    {
      "epoch": 0.5550177476605357,
      "grad_norm": 2.3059604167938232,
      "learning_rate": 0.0005,
      "loss": 2.3178906440734863,
      "step": 3440
    },
    {
      "epoch": 0.5551790900290416,
      "grad_norm": 1.2999471426010132,
      "learning_rate": 0.0005,
      "loss": 2.9928297996520996,
      "step": 3441
    },
    {
      "epoch": 0.5553404323975476,
      "grad_norm": 1.2594430446624756,
      "learning_rate": 0.0005,
      "loss": 3.1174705028533936,
      "step": 3442
    },
    {
      "epoch": 0.5555017747660536,
      "grad_norm": 1.340552806854248,
      "learning_rate": 0.0005,
      "loss": 3.2346110343933105,
      "step": 3443
    },
    {
      "epoch": 0.5556631171345595,
      "grad_norm": 1.3717511892318726,
      "learning_rate": 0.0005,
      "loss": 2.5341434478759766,
      "step": 3444
    },
    {
      "epoch": 0.5558244595030655,
      "grad_norm": 1.1049940586090088,
      "learning_rate": 0.0005,
      "loss": 2.4895734786987305,
      "step": 3445
    },
    {
      "epoch": 0.5559858018715714,
      "grad_norm": 1.32472825050354,
      "learning_rate": 0.0005,
      "loss": 3.071497917175293,
      "step": 3446
    },
    {
      "epoch": 0.5561471442400775,
      "grad_norm": 1.4109506607055664,
      "learning_rate": 0.0005,
      "loss": 2.9654879570007324,
      "step": 3447
    },
    {
      "epoch": 0.5563084866085835,
      "grad_norm": 2.3540849685668945,
      "learning_rate": 0.0005,
      "loss": 2.5154287815093994,
      "step": 3448
    },
    {
      "epoch": 0.5564698289770894,
      "grad_norm": 0.9818688631057739,
      "learning_rate": 0.0005,
      "loss": 2.631140947341919,
      "step": 3449
    },
    {
      "epoch": 0.5566311713455954,
      "grad_norm": 1.375884771347046,
      "learning_rate": 0.0005,
      "loss": 2.643437623977661,
      "step": 3450
    },
    {
      "epoch": 0.5567925137141013,
      "grad_norm": 1.3017232418060303,
      "learning_rate": 0.0005,
      "loss": 2.582190990447998,
      "step": 3451
    },
    {
      "epoch": 0.5569538560826073,
      "grad_norm": 1.2481660842895508,
      "learning_rate": 0.0005,
      "loss": 3.093822479248047,
      "step": 3452
    },
    {
      "epoch": 0.5571151984511132,
      "grad_norm": 1.288214921951294,
      "learning_rate": 0.0005,
      "loss": 2.9836318492889404,
      "step": 3453
    },
    {
      "epoch": 0.5572765408196192,
      "grad_norm": 1.0252866744995117,
      "learning_rate": 0.0005,
      "loss": 2.6255407333374023,
      "step": 3454
    },
    {
      "epoch": 0.5574378831881251,
      "grad_norm": 1.3989075422286987,
      "learning_rate": 0.0005,
      "loss": 2.502312183380127,
      "step": 3455
    },
    {
      "epoch": 0.5575992255566312,
      "grad_norm": 1.4842902421951294,
      "learning_rate": 0.0005,
      "loss": 2.801640033721924,
      "step": 3456
    },
    {
      "epoch": 0.5577605679251372,
      "grad_norm": 1.4936332702636719,
      "learning_rate": 0.0005,
      "loss": 2.7058286666870117,
      "step": 3457
    },
    {
      "epoch": 0.5579219102936431,
      "grad_norm": 1.5408217906951904,
      "learning_rate": 0.0005,
      "loss": 2.760319709777832,
      "step": 3458
    },
    {
      "epoch": 0.5580832526621491,
      "grad_norm": 2.2100729942321777,
      "learning_rate": 0.0005,
      "loss": 3.3171207904815674,
      "step": 3459
    },
    {
      "epoch": 0.558244595030655,
      "grad_norm": 1.5173615217208862,
      "learning_rate": 0.0005,
      "loss": 3.2419474124908447,
      "step": 3460
    },
    {
      "epoch": 0.558405937399161,
      "grad_norm": 1.4925947189331055,
      "learning_rate": 0.0005,
      "loss": 3.133732795715332,
      "step": 3461
    },
    {
      "epoch": 0.558567279767667,
      "grad_norm": 1.204013466835022,
      "learning_rate": 0.0005,
      "loss": 2.7332570552825928,
      "step": 3462
    },
    {
      "epoch": 0.5587286221361729,
      "grad_norm": 1.5994337797164917,
      "learning_rate": 0.0005,
      "loss": 3.0737733840942383,
      "step": 3463
    },
    {
      "epoch": 0.558889964504679,
      "grad_norm": 1.126192569732666,
      "learning_rate": 0.0005,
      "loss": 2.8097238540649414,
      "step": 3464
    },
    {
      "epoch": 0.5590513068731849,
      "grad_norm": 1.1624329090118408,
      "learning_rate": 0.0005,
      "loss": 2.792175769805908,
      "step": 3465
    },
    {
      "epoch": 0.5592126492416909,
      "grad_norm": 1.198028564453125,
      "learning_rate": 0.0005,
      "loss": 2.125330686569214,
      "step": 3466
    },
    {
      "epoch": 0.5593739916101969,
      "grad_norm": 1.4865490198135376,
      "learning_rate": 0.0005,
      "loss": 3.032125473022461,
      "step": 3467
    },
    {
      "epoch": 0.5595353339787028,
      "grad_norm": 1.387412190437317,
      "learning_rate": 0.0005,
      "loss": 2.696495771408081,
      "step": 3468
    },
    {
      "epoch": 0.5596966763472088,
      "grad_norm": 1.3844060897827148,
      "learning_rate": 0.0005,
      "loss": 3.1381256580352783,
      "step": 3469
    },
    {
      "epoch": 0.5598580187157147,
      "grad_norm": 1.3711720705032349,
      "learning_rate": 0.0005,
      "loss": 2.510010242462158,
      "step": 3470
    },
    {
      "epoch": 0.5600193610842207,
      "grad_norm": 1.2582875490188599,
      "learning_rate": 0.0005,
      "loss": 2.781891345977783,
      "step": 3471
    },
    {
      "epoch": 0.5601807034527266,
      "grad_norm": 1.4562268257141113,
      "learning_rate": 0.0005,
      "loss": 2.49180269241333,
      "step": 3472
    },
    {
      "epoch": 0.5603420458212327,
      "grad_norm": 1.5608888864517212,
      "learning_rate": 0.0005,
      "loss": 2.5432019233703613,
      "step": 3473
    },
    {
      "epoch": 0.5605033881897387,
      "grad_norm": 1.1025739908218384,
      "learning_rate": 0.0005,
      "loss": 2.623000144958496,
      "step": 3474
    },
    {
      "epoch": 0.5606647305582446,
      "grad_norm": 1.0886387825012207,
      "learning_rate": 0.0005,
      "loss": 2.41702938079834,
      "step": 3475
    },
    {
      "epoch": 0.5608260729267506,
      "grad_norm": 1.8162516355514526,
      "learning_rate": 0.0005,
      "loss": 2.7566940784454346,
      "step": 3476
    },
    {
      "epoch": 0.5609874152952565,
      "grad_norm": 1.8331032991409302,
      "learning_rate": 0.0005,
      "loss": 2.660198211669922,
      "step": 3477
    },
    {
      "epoch": 0.5611487576637625,
      "grad_norm": 0.9715525507926941,
      "learning_rate": 0.0005,
      "loss": 2.967860221862793,
      "step": 3478
    },
    {
      "epoch": 0.5613101000322684,
      "grad_norm": 1.066386103630066,
      "learning_rate": 0.0005,
      "loss": 2.718696117401123,
      "step": 3479
    },
    {
      "epoch": 0.5614714424007744,
      "grad_norm": 1.0951260328292847,
      "learning_rate": 0.0005,
      "loss": 2.6882472038269043,
      "step": 3480
    },
    {
      "epoch": 0.5616327847692804,
      "grad_norm": 0.9857366681098938,
      "learning_rate": 0.0005,
      "loss": 2.7244768142700195,
      "step": 3481
    },
    {
      "epoch": 0.5617941271377864,
      "grad_norm": 1.4118579626083374,
      "learning_rate": 0.0005,
      "loss": 2.5989432334899902,
      "step": 3482
    },
    {
      "epoch": 0.5619554695062924,
      "grad_norm": 1.0461612939834595,
      "learning_rate": 0.0005,
      "loss": 2.7707362174987793,
      "step": 3483
    },
    {
      "epoch": 0.5621168118747983,
      "grad_norm": 0.8557549715042114,
      "learning_rate": 0.0005,
      "loss": 2.719726800918579,
      "step": 3484
    },
    {
      "epoch": 0.5622781542433043,
      "grad_norm": 1.0620653629302979,
      "learning_rate": 0.0005,
      "loss": 2.078623056411743,
      "step": 3485
    },
    {
      "epoch": 0.5624394966118103,
      "grad_norm": 1.0980638265609741,
      "learning_rate": 0.0005,
      "loss": 2.70957612991333,
      "step": 3486
    },
    {
      "epoch": 0.5626008389803162,
      "grad_norm": 1.2554727792739868,
      "learning_rate": 0.0005,
      "loss": 2.9074387550354004,
      "step": 3487
    },
    {
      "epoch": 0.5627621813488222,
      "grad_norm": 1.0785847902297974,
      "learning_rate": 0.0005,
      "loss": 2.6869711875915527,
      "step": 3488
    },
    {
      "epoch": 0.5629235237173281,
      "grad_norm": 1.408583641052246,
      "learning_rate": 0.0005,
      "loss": 2.8167591094970703,
      "step": 3489
    },
    {
      "epoch": 0.5630848660858342,
      "grad_norm": 1.5287505388259888,
      "learning_rate": 0.0005,
      "loss": 2.5186305046081543,
      "step": 3490
    },
    {
      "epoch": 0.5632462084543401,
      "grad_norm": 1.2735295295715332,
      "learning_rate": 0.0005,
      "loss": 2.99375057220459,
      "step": 3491
    },
    {
      "epoch": 0.5634075508228461,
      "grad_norm": 1.3209130764007568,
      "learning_rate": 0.0005,
      "loss": 3.166757345199585,
      "step": 3492
    },
    {
      "epoch": 0.5635688931913521,
      "grad_norm": 1.1764470338821411,
      "learning_rate": 0.0005,
      "loss": 2.814791679382324,
      "step": 3493
    },
    {
      "epoch": 0.563730235559858,
      "grad_norm": 1.512463092803955,
      "learning_rate": 0.0005,
      "loss": 2.981180191040039,
      "step": 3494
    },
    {
      "epoch": 0.563891577928364,
      "grad_norm": 1.0470097064971924,
      "learning_rate": 0.0005,
      "loss": 2.638406753540039,
      "step": 3495
    },
    {
      "epoch": 0.5640529202968699,
      "grad_norm": 0.9572147727012634,
      "learning_rate": 0.0005,
      "loss": 2.5625133514404297,
      "step": 3496
    },
    {
      "epoch": 0.5642142626653759,
      "grad_norm": 1.0734397172927856,
      "learning_rate": 0.0005,
      "loss": 2.411776065826416,
      "step": 3497
    },
    {
      "epoch": 0.5643756050338818,
      "grad_norm": 1.3169010877609253,
      "learning_rate": 0.0005,
      "loss": 2.553417921066284,
      "step": 3498
    },
    {
      "epoch": 0.5645369474023879,
      "grad_norm": 1.2720128297805786,
      "learning_rate": 0.0005,
      "loss": 3.0229616165161133,
      "step": 3499
    },
    {
      "epoch": 0.5646982897708939,
      "grad_norm": 1.351808786392212,
      "learning_rate": 0.0005,
      "loss": 2.8567869663238525,
      "step": 3500
    },
    {
      "epoch": 0.5648596321393998,
      "grad_norm": 1.1768698692321777,
      "learning_rate": 0.0005,
      "loss": 2.762418031692505,
      "step": 3501
    },
    {
      "epoch": 0.5650209745079058,
      "grad_norm": 1.2149196863174438,
      "learning_rate": 0.0005,
      "loss": 2.8886406421661377,
      "step": 3502
    },
    {
      "epoch": 0.5651823168764117,
      "grad_norm": 1.833866000175476,
      "learning_rate": 0.0005,
      "loss": 2.926612615585327,
      "step": 3503
    },
    {
      "epoch": 0.5653436592449177,
      "grad_norm": 1.3041565418243408,
      "learning_rate": 0.0005,
      "loss": 2.8777875900268555,
      "step": 3504
    },
    {
      "epoch": 0.5655050016134237,
      "grad_norm": 1.2950538396835327,
      "learning_rate": 0.0005,
      "loss": 2.8455469608306885,
      "step": 3505
    },
    {
      "epoch": 0.5656663439819296,
      "grad_norm": 1.2703050374984741,
      "learning_rate": 0.0005,
      "loss": 2.9896960258483887,
      "step": 3506
    },
    {
      "epoch": 0.5658276863504357,
      "grad_norm": 1.1909626722335815,
      "learning_rate": 0.0005,
      "loss": 2.5174002647399902,
      "step": 3507
    },
    {
      "epoch": 0.5659890287189416,
      "grad_norm": 1.3173266649246216,
      "learning_rate": 0.0005,
      "loss": 2.409722328186035,
      "step": 3508
    },
    {
      "epoch": 0.5661503710874476,
      "grad_norm": 1.143396019935608,
      "learning_rate": 0.0005,
      "loss": 2.514793634414673,
      "step": 3509
    },
    {
      "epoch": 0.5663117134559535,
      "grad_norm": 1.1172564029693604,
      "learning_rate": 0.0005,
      "loss": 2.919586658477783,
      "step": 3510
    },
    {
      "epoch": 0.5664730558244595,
      "grad_norm": 1.1791032552719116,
      "learning_rate": 0.0005,
      "loss": 2.7624754905700684,
      "step": 3511
    },
    {
      "epoch": 0.5666343981929655,
      "grad_norm": 1.4151794910430908,
      "learning_rate": 0.0005,
      "loss": 2.5469541549682617,
      "step": 3512
    },
    {
      "epoch": 0.5667957405614714,
      "grad_norm": 1.3562560081481934,
      "learning_rate": 0.0005,
      "loss": 2.596027374267578,
      "step": 3513
    },
    {
      "epoch": 0.5669570829299774,
      "grad_norm": 1.4788724184036255,
      "learning_rate": 0.0005,
      "loss": 2.829949140548706,
      "step": 3514
    },
    {
      "epoch": 0.5671184252984833,
      "grad_norm": 1.0262075662612915,
      "learning_rate": 0.0005,
      "loss": 3.1381149291992188,
      "step": 3515
    },
    {
      "epoch": 0.5672797676669894,
      "grad_norm": 1.1539928913116455,
      "learning_rate": 0.0005,
      "loss": 2.9384312629699707,
      "step": 3516
    },
    {
      "epoch": 0.5674411100354954,
      "grad_norm": 1.1573702096939087,
      "learning_rate": 0.0005,
      "loss": 2.7480082511901855,
      "step": 3517
    },
    {
      "epoch": 0.5676024524040013,
      "grad_norm": 1.1241718530654907,
      "learning_rate": 0.0005,
      "loss": 2.510829210281372,
      "step": 3518
    },
    {
      "epoch": 0.5677637947725073,
      "grad_norm": 1.210105299949646,
      "learning_rate": 0.0005,
      "loss": 2.545377016067505,
      "step": 3519
    },
    {
      "epoch": 0.5679251371410132,
      "grad_norm": 17.851966857910156,
      "learning_rate": 0.0005,
      "loss": 2.6369409561157227,
      "step": 3520
    },
    {
      "epoch": 0.5680864795095192,
      "grad_norm": 1.3037109375,
      "learning_rate": 0.0005,
      "loss": 2.413048267364502,
      "step": 3521
    },
    {
      "epoch": 0.5682478218780251,
      "grad_norm": 1.4990724325180054,
      "learning_rate": 0.0005,
      "loss": 2.658064842224121,
      "step": 3522
    },
    {
      "epoch": 0.5684091642465311,
      "grad_norm": 1.5792239904403687,
      "learning_rate": 0.0005,
      "loss": 2.8429596424102783,
      "step": 3523
    },
    {
      "epoch": 0.5685705066150372,
      "grad_norm": 1.008297085762024,
      "learning_rate": 0.0005,
      "loss": 2.518104076385498,
      "step": 3524
    },
    {
      "epoch": 0.5687318489835431,
      "grad_norm": 1.2825899124145508,
      "learning_rate": 0.0005,
      "loss": 2.7115869522094727,
      "step": 3525
    },
    {
      "epoch": 0.5688931913520491,
      "grad_norm": 2.4766693115234375,
      "learning_rate": 0.0005,
      "loss": 2.7367303371429443,
      "step": 3526
    },
    {
      "epoch": 0.569054533720555,
      "grad_norm": 1.132886528968811,
      "learning_rate": 0.0005,
      "loss": 2.7700889110565186,
      "step": 3527
    },
    {
      "epoch": 0.569215876089061,
      "grad_norm": 1.1802215576171875,
      "learning_rate": 0.0005,
      "loss": 2.748201370239258,
      "step": 3528
    },
    {
      "epoch": 0.569377218457567,
      "grad_norm": 1.1817189455032349,
      "learning_rate": 0.0005,
      "loss": 2.5698697566986084,
      "step": 3529
    },
    {
      "epoch": 0.5695385608260729,
      "grad_norm": 1.330692172050476,
      "learning_rate": 0.0005,
      "loss": 2.453434467315674,
      "step": 3530
    },
    {
      "epoch": 0.5696999031945789,
      "grad_norm": 1.400850534439087,
      "learning_rate": 0.0005,
      "loss": 2.95870041847229,
      "step": 3531
    },
    {
      "epoch": 0.5698612455630848,
      "grad_norm": 1.7221407890319824,
      "learning_rate": 0.0005,
      "loss": 2.7285327911376953,
      "step": 3532
    },
    {
      "epoch": 0.5700225879315909,
      "grad_norm": 1.1868877410888672,
      "learning_rate": 0.0005,
      "loss": 2.6830992698669434,
      "step": 3533
    },
    {
      "epoch": 0.5701839303000968,
      "grad_norm": 1.052332878112793,
      "learning_rate": 0.0005,
      "loss": 2.4136486053466797,
      "step": 3534
    },
    {
      "epoch": 0.5703452726686028,
      "grad_norm": 1.040261149406433,
      "learning_rate": 0.0005,
      "loss": 2.8291878700256348,
      "step": 3535
    },
    {
      "epoch": 0.5705066150371088,
      "grad_norm": 1.5326213836669922,
      "learning_rate": 0.0005,
      "loss": 2.628079891204834,
      "step": 3536
    },
    {
      "epoch": 0.5706679574056147,
      "grad_norm": 1.5719335079193115,
      "learning_rate": 0.0005,
      "loss": 2.620732307434082,
      "step": 3537
    },
    {
      "epoch": 0.5708292997741207,
      "grad_norm": 1.24905526638031,
      "learning_rate": 0.0005,
      "loss": 2.3618412017822266,
      "step": 3538
    },
    {
      "epoch": 0.5709906421426266,
      "grad_norm": 1.262481689453125,
      "learning_rate": 0.0005,
      "loss": 3.0645394325256348,
      "step": 3539
    },
    {
      "epoch": 0.5711519845111326,
      "grad_norm": 1.1636545658111572,
      "learning_rate": 0.0005,
      "loss": 2.5785012245178223,
      "step": 3540
    },
    {
      "epoch": 0.5713133268796385,
      "grad_norm": 1.222531795501709,
      "learning_rate": 0.0005,
      "loss": 2.3471059799194336,
      "step": 3541
    },
    {
      "epoch": 0.5714746692481446,
      "grad_norm": 1.8598275184631348,
      "learning_rate": 0.0005,
      "loss": 3.100391149520874,
      "step": 3542
    },
    {
      "epoch": 0.5716360116166506,
      "grad_norm": 1.2521851062774658,
      "learning_rate": 0.0005,
      "loss": 2.5447511672973633,
      "step": 3543
    },
    {
      "epoch": 0.5717973539851565,
      "grad_norm": 0.9962290525436401,
      "learning_rate": 0.0005,
      "loss": 2.441643714904785,
      "step": 3544
    },
    {
      "epoch": 0.5719586963536625,
      "grad_norm": 1.41148841381073,
      "learning_rate": 0.0005,
      "loss": 2.5468931198120117,
      "step": 3545
    },
    {
      "epoch": 0.5721200387221684,
      "grad_norm": 1.025048017501831,
      "learning_rate": 0.0005,
      "loss": 2.5774288177490234,
      "step": 3546
    },
    {
      "epoch": 0.5722813810906744,
      "grad_norm": 1.3216990232467651,
      "learning_rate": 0.0005,
      "loss": 2.929751396179199,
      "step": 3547
    },
    {
      "epoch": 0.5724427234591803,
      "grad_norm": 1.488259196281433,
      "learning_rate": 0.0005,
      "loss": 3.0292394161224365,
      "step": 3548
    },
    {
      "epoch": 0.5726040658276863,
      "grad_norm": 1.020448088645935,
      "learning_rate": 0.0005,
      "loss": 2.6106812953948975,
      "step": 3549
    },
    {
      "epoch": 0.5727654081961924,
      "grad_norm": 1.550540804862976,
      "learning_rate": 0.0005,
      "loss": 2.939581871032715,
      "step": 3550
    },
    {
      "epoch": 0.5729267505646983,
      "grad_norm": 1.2186098098754883,
      "learning_rate": 0.0005,
      "loss": 2.665510416030884,
      "step": 3551
    },
    {
      "epoch": 0.5730880929332043,
      "grad_norm": 1.2119542360305786,
      "learning_rate": 0.0005,
      "loss": 2.8509209156036377,
      "step": 3552
    },
    {
      "epoch": 0.5732494353017102,
      "grad_norm": 1.3851667642593384,
      "learning_rate": 0.0005,
      "loss": 2.87137508392334,
      "step": 3553
    },
    {
      "epoch": 0.5734107776702162,
      "grad_norm": 1.3886845111846924,
      "learning_rate": 0.0005,
      "loss": 2.6853256225585938,
      "step": 3554
    },
    {
      "epoch": 0.5735721200387222,
      "grad_norm": 0.9003841876983643,
      "learning_rate": 0.0005,
      "loss": 2.9681100845336914,
      "step": 3555
    },
    {
      "epoch": 0.5737334624072281,
      "grad_norm": 1.3363133668899536,
      "learning_rate": 0.0005,
      "loss": 2.822692394256592,
      "step": 3556
    },
    {
      "epoch": 0.5738948047757341,
      "grad_norm": 1.3444432020187378,
      "learning_rate": 0.0005,
      "loss": 2.658942461013794,
      "step": 3557
    },
    {
      "epoch": 0.57405614714424,
      "grad_norm": 1.856709361076355,
      "learning_rate": 0.0005,
      "loss": 2.892707347869873,
      "step": 3558
    },
    {
      "epoch": 0.5742174895127461,
      "grad_norm": 1.5984790325164795,
      "learning_rate": 0.0005,
      "loss": 2.747832775115967,
      "step": 3559
    },
    {
      "epoch": 0.574378831881252,
      "grad_norm": 1.0555357933044434,
      "learning_rate": 0.0005,
      "loss": 2.534184217453003,
      "step": 3560
    },
    {
      "epoch": 0.574540174249758,
      "grad_norm": 1.5168964862823486,
      "learning_rate": 0.0005,
      "loss": 2.7290306091308594,
      "step": 3561
    },
    {
      "epoch": 0.574701516618264,
      "grad_norm": 1.4935424327850342,
      "learning_rate": 0.0005,
      "loss": 2.4507927894592285,
      "step": 3562
    },
    {
      "epoch": 0.5748628589867699,
      "grad_norm": 1.648580551147461,
      "learning_rate": 0.0005,
      "loss": 2.4217028617858887,
      "step": 3563
    },
    {
      "epoch": 0.5750242013552759,
      "grad_norm": 1.6171605587005615,
      "learning_rate": 0.0005,
      "loss": 2.7258236408233643,
      "step": 3564
    },
    {
      "epoch": 0.5751855437237818,
      "grad_norm": 1.107550024986267,
      "learning_rate": 0.0005,
      "loss": 2.9759674072265625,
      "step": 3565
    },
    {
      "epoch": 0.5753468860922878,
      "grad_norm": 0.9721173644065857,
      "learning_rate": 0.0005,
      "loss": 2.972555160522461,
      "step": 3566
    },
    {
      "epoch": 0.5755082284607939,
      "grad_norm": 1.1902117729187012,
      "learning_rate": 0.0005,
      "loss": 2.7330245971679688,
      "step": 3567
    },
    {
      "epoch": 0.5756695708292998,
      "grad_norm": 1.1342546939849854,
      "learning_rate": 0.0005,
      "loss": 2.0829386711120605,
      "step": 3568
    },
    {
      "epoch": 0.5758309131978058,
      "grad_norm": 1.0589702129364014,
      "learning_rate": 0.0005,
      "loss": 2.970834255218506,
      "step": 3569
    },
    {
      "epoch": 0.5759922555663117,
      "grad_norm": 0.9121820330619812,
      "learning_rate": 0.0005,
      "loss": 3.013326406478882,
      "step": 3570
    },
    {
      "epoch": 0.5761535979348177,
      "grad_norm": 1.6200777292251587,
      "learning_rate": 0.0005,
      "loss": 2.7335610389709473,
      "step": 3571
    },
    {
      "epoch": 0.5763149403033236,
      "grad_norm": 1.2806313037872314,
      "learning_rate": 0.0005,
      "loss": 3.002918004989624,
      "step": 3572
    },
    {
      "epoch": 0.5764762826718296,
      "grad_norm": 1.1974096298217773,
      "learning_rate": 0.0005,
      "loss": 2.572526454925537,
      "step": 3573
    },
    {
      "epoch": 0.5766376250403356,
      "grad_norm": 1.0369224548339844,
      "learning_rate": 0.0005,
      "loss": 3.099698543548584,
      "step": 3574
    },
    {
      "epoch": 0.5767989674088415,
      "grad_norm": 1.1601526737213135,
      "learning_rate": 0.0005,
      "loss": 2.8764729499816895,
      "step": 3575
    },
    {
      "epoch": 0.5769603097773476,
      "grad_norm": 1.4521318674087524,
      "learning_rate": 0.0005,
      "loss": 3.177119493484497,
      "step": 3576
    },
    {
      "epoch": 0.5771216521458535,
      "grad_norm": 0.9661845564842224,
      "learning_rate": 0.0005,
      "loss": 2.8467233180999756,
      "step": 3577
    },
    {
      "epoch": 0.5772829945143595,
      "grad_norm": 1.4115934371948242,
      "learning_rate": 0.0005,
      "loss": 2.528381824493408,
      "step": 3578
    },
    {
      "epoch": 0.5774443368828654,
      "grad_norm": 2.1815567016601562,
      "learning_rate": 0.0005,
      "loss": 2.6362037658691406,
      "step": 3579
    },
    {
      "epoch": 0.5776056792513714,
      "grad_norm": 1.1972275972366333,
      "learning_rate": 0.0005,
      "loss": 2.387450695037842,
      "step": 3580
    },
    {
      "epoch": 0.5777670216198774,
      "grad_norm": 1.2251462936401367,
      "learning_rate": 0.0005,
      "loss": 2.345679521560669,
      "step": 3581
    },
    {
      "epoch": 0.5779283639883833,
      "grad_norm": 2.1118459701538086,
      "learning_rate": 0.0005,
      "loss": 2.8025715351104736,
      "step": 3582
    },
    {
      "epoch": 0.5780897063568893,
      "grad_norm": 1.0926984548568726,
      "learning_rate": 0.0005,
      "loss": 2.7874526977539062,
      "step": 3583
    },
    {
      "epoch": 0.5782510487253952,
      "grad_norm": 1.2818689346313477,
      "learning_rate": 0.0005,
      "loss": 2.5321712493896484,
      "step": 3584
    },
    {
      "epoch": 0.5784123910939013,
      "grad_norm": 1.2303662300109863,
      "learning_rate": 0.0005,
      "loss": 2.965491771697998,
      "step": 3585
    },
    {
      "epoch": 0.5785737334624073,
      "grad_norm": 1.279686450958252,
      "learning_rate": 0.0005,
      "loss": 2.487339973449707,
      "step": 3586
    },
    {
      "epoch": 0.5787350758309132,
      "grad_norm": 1.1843879222869873,
      "learning_rate": 0.0005,
      "loss": 3.0372867584228516,
      "step": 3587
    },
    {
      "epoch": 0.5788964181994192,
      "grad_norm": 1.431519865989685,
      "learning_rate": 0.0005,
      "loss": 2.821218490600586,
      "step": 3588
    },
    {
      "epoch": 0.5790577605679251,
      "grad_norm": 2.1007261276245117,
      "learning_rate": 0.0005,
      "loss": 2.588155508041382,
      "step": 3589
    },
    {
      "epoch": 0.5792191029364311,
      "grad_norm": 1.7281017303466797,
      "learning_rate": 0.0005,
      "loss": 2.7282538414001465,
      "step": 3590
    },
    {
      "epoch": 0.579380445304937,
      "grad_norm": 1.0788958072662354,
      "learning_rate": 0.0005,
      "loss": 2.551236391067505,
      "step": 3591
    },
    {
      "epoch": 0.579541787673443,
      "grad_norm": 1.0609221458435059,
      "learning_rate": 0.0005,
      "loss": 2.3272886276245117,
      "step": 3592
    },
    {
      "epoch": 0.5797031300419491,
      "grad_norm": 0.9168956875801086,
      "learning_rate": 0.0005,
      "loss": 2.425762176513672,
      "step": 3593
    },
    {
      "epoch": 0.579864472410455,
      "grad_norm": 1.276932716369629,
      "learning_rate": 0.0005,
      "loss": 2.563194513320923,
      "step": 3594
    },
    {
      "epoch": 0.580025814778961,
      "grad_norm": 2.0237715244293213,
      "learning_rate": 0.0005,
      "loss": 2.5525572299957275,
      "step": 3595
    },
    {
      "epoch": 0.5801871571474669,
      "grad_norm": 1.3833273649215698,
      "learning_rate": 0.0005,
      "loss": 2.7323389053344727,
      "step": 3596
    },
    {
      "epoch": 0.5803484995159729,
      "grad_norm": 1.482284665107727,
      "learning_rate": 0.0005,
      "loss": 2.501490354537964,
      "step": 3597
    },
    {
      "epoch": 0.5805098418844788,
      "grad_norm": 1.407151460647583,
      "learning_rate": 0.0005,
      "loss": 2.8182737827301025,
      "step": 3598
    },
    {
      "epoch": 0.5806711842529848,
      "grad_norm": 1.9164416790008545,
      "learning_rate": 0.0005,
      "loss": 2.3863682746887207,
      "step": 3599
    },
    {
      "epoch": 0.5808325266214908,
      "grad_norm": 1.8800352811813354,
      "learning_rate": 0.0005,
      "loss": 3.044894218444824,
      "step": 3600
    },
    {
      "epoch": 0.5809938689899967,
      "grad_norm": 1.6566811800003052,
      "learning_rate": 0.0005,
      "loss": 2.3666326999664307,
      "step": 3601
    },
    {
      "epoch": 0.5811552113585028,
      "grad_norm": 1.4569759368896484,
      "learning_rate": 0.0005,
      "loss": 2.6843972206115723,
      "step": 3602
    },
    {
      "epoch": 0.5813165537270087,
      "grad_norm": 1.008126974105835,
      "learning_rate": 0.0005,
      "loss": 2.694136142730713,
      "step": 3603
    },
    {
      "epoch": 0.5814778960955147,
      "grad_norm": 1.1992241144180298,
      "learning_rate": 0.0005,
      "loss": 2.6627533435821533,
      "step": 3604
    },
    {
      "epoch": 0.5816392384640207,
      "grad_norm": 1.1881715059280396,
      "learning_rate": 0.0005,
      "loss": 2.150320053100586,
      "step": 3605
    },
    {
      "epoch": 0.5818005808325266,
      "grad_norm": 1.2852734327316284,
      "learning_rate": 0.0005,
      "loss": 2.9569973945617676,
      "step": 3606
    },
    {
      "epoch": 0.5819619232010326,
      "grad_norm": 1.02700936794281,
      "learning_rate": 0.0005,
      "loss": 2.8828227519989014,
      "step": 3607
    },
    {
      "epoch": 0.5821232655695385,
      "grad_norm": 1.3314810991287231,
      "learning_rate": 0.0005,
      "loss": 2.7686939239501953,
      "step": 3608
    },
    {
      "epoch": 0.5822846079380445,
      "grad_norm": 1.1378633975982666,
      "learning_rate": 0.0005,
      "loss": 2.593879222869873,
      "step": 3609
    },
    {
      "epoch": 0.5824459503065506,
      "grad_norm": 1.2011371850967407,
      "learning_rate": 0.0005,
      "loss": 2.5077786445617676,
      "step": 3610
    },
    {
      "epoch": 0.5826072926750565,
      "grad_norm": 1.5220723152160645,
      "learning_rate": 0.0005,
      "loss": 2.5712432861328125,
      "step": 3611
    },
    {
      "epoch": 0.5827686350435625,
      "grad_norm": 1.1511914730072021,
      "learning_rate": 0.0005,
      "loss": 2.8139171600341797,
      "step": 3612
    },
    {
      "epoch": 0.5829299774120684,
      "grad_norm": 1.9847872257232666,
      "learning_rate": 0.0005,
      "loss": 2.7184996604919434,
      "step": 3613
    },
    {
      "epoch": 0.5830913197805744,
      "grad_norm": 1.0460153818130493,
      "learning_rate": 0.0005,
      "loss": 2.7577946186065674,
      "step": 3614
    },
    {
      "epoch": 0.5832526621490803,
      "grad_norm": 0.9699302315711975,
      "learning_rate": 0.0005,
      "loss": 2.94683837890625,
      "step": 3615
    },
    {
      "epoch": 0.5834140045175863,
      "grad_norm": 1.0685023069381714,
      "learning_rate": 0.0005,
      "loss": 2.5484204292297363,
      "step": 3616
    },
    {
      "epoch": 0.5835753468860922,
      "grad_norm": 1.1516271829605103,
      "learning_rate": 0.0005,
      "loss": 2.277639150619507,
      "step": 3617
    },
    {
      "epoch": 0.5837366892545982,
      "grad_norm": 1.0677083730697632,
      "learning_rate": 0.0005,
      "loss": 2.924020767211914,
      "step": 3618
    },
    {
      "epoch": 0.5838980316231043,
      "grad_norm": 1.0451873540878296,
      "learning_rate": 0.0005,
      "loss": 2.364439010620117,
      "step": 3619
    },
    {
      "epoch": 0.5840593739916102,
      "grad_norm": 1.5009993314743042,
      "learning_rate": 0.0005,
      "loss": 2.644686222076416,
      "step": 3620
    },
    {
      "epoch": 0.5842207163601162,
      "grad_norm": 1.4263960123062134,
      "learning_rate": 0.0005,
      "loss": 2.457570791244507,
      "step": 3621
    },
    {
      "epoch": 0.5843820587286221,
      "grad_norm": 2.380108118057251,
      "learning_rate": 0.0005,
      "loss": 2.814662456512451,
      "step": 3622
    },
    {
      "epoch": 0.5845434010971281,
      "grad_norm": 1.1517406702041626,
      "learning_rate": 0.0005,
      "loss": 2.749497413635254,
      "step": 3623
    },
    {
      "epoch": 0.5847047434656341,
      "grad_norm": 1.8945103883743286,
      "learning_rate": 0.0005,
      "loss": 2.5624663829803467,
      "step": 3624
    },
    {
      "epoch": 0.58486608583414,
      "grad_norm": 1.4783114194869995,
      "learning_rate": 0.0005,
      "loss": 2.733915328979492,
      "step": 3625
    },
    {
      "epoch": 0.585027428202646,
      "grad_norm": 1.3926494121551514,
      "learning_rate": 0.0005,
      "loss": 2.805239200592041,
      "step": 3626
    },
    {
      "epoch": 0.585188770571152,
      "grad_norm": 1.505286455154419,
      "learning_rate": 0.0005,
      "loss": 3.2643356323242188,
      "step": 3627
    },
    {
      "epoch": 0.585350112939658,
      "grad_norm": 1.2738842964172363,
      "learning_rate": 0.0005,
      "loss": 2.331317901611328,
      "step": 3628
    },
    {
      "epoch": 0.585511455308164,
      "grad_norm": 1.6075923442840576,
      "learning_rate": 0.0005,
      "loss": 3.069005012512207,
      "step": 3629
    },
    {
      "epoch": 0.5856727976766699,
      "grad_norm": 1.462116003036499,
      "learning_rate": 0.0005,
      "loss": 2.6326870918273926,
      "step": 3630
    },
    {
      "epoch": 0.5858341400451759,
      "grad_norm": 1.4412206411361694,
      "learning_rate": 0.0005,
      "loss": 2.941385269165039,
      "step": 3631
    },
    {
      "epoch": 0.5859954824136818,
      "grad_norm": 1.6434431076049805,
      "learning_rate": 0.0005,
      "loss": 2.6587038040161133,
      "step": 3632
    },
    {
      "epoch": 0.5861568247821878,
      "grad_norm": 1.0826776027679443,
      "learning_rate": 0.0005,
      "loss": 2.52131724357605,
      "step": 3633
    },
    {
      "epoch": 0.5863181671506937,
      "grad_norm": 1.1563093662261963,
      "learning_rate": 0.0005,
      "loss": 2.8229336738586426,
      "step": 3634
    },
    {
      "epoch": 0.5864795095191997,
      "grad_norm": 1.5019164085388184,
      "learning_rate": 0.0005,
      "loss": 2.9888060092926025,
      "step": 3635
    },
    {
      "epoch": 0.5866408518877058,
      "grad_norm": 1.2266991138458252,
      "learning_rate": 0.0005,
      "loss": 2.671804904937744,
      "step": 3636
    },
    {
      "epoch": 0.5868021942562117,
      "grad_norm": 1.4304903745651245,
      "learning_rate": 0.0005,
      "loss": 2.8150076866149902,
      "step": 3637
    },
    {
      "epoch": 0.5869635366247177,
      "grad_norm": 1.0687408447265625,
      "learning_rate": 0.0005,
      "loss": 2.666001081466675,
      "step": 3638
    },
    {
      "epoch": 0.5871248789932236,
      "grad_norm": 1.561372995376587,
      "learning_rate": 0.0005,
      "loss": 2.4786834716796875,
      "step": 3639
    },
    {
      "epoch": 0.5872862213617296,
      "grad_norm": 1.2504545450210571,
      "learning_rate": 0.0005,
      "loss": 3.0336408615112305,
      "step": 3640
    },
    {
      "epoch": 0.5874475637302355,
      "grad_norm": 1.3830351829528809,
      "learning_rate": 0.0005,
      "loss": 2.6998543739318848,
      "step": 3641
    },
    {
      "epoch": 0.5876089060987415,
      "grad_norm": 1.125779628753662,
      "learning_rate": 0.0005,
      "loss": 2.5266313552856445,
      "step": 3642
    },
    {
      "epoch": 0.5877702484672475,
      "grad_norm": 1.2087504863739014,
      "learning_rate": 0.0005,
      "loss": 3.066171169281006,
      "step": 3643
    },
    {
      "epoch": 0.5879315908357534,
      "grad_norm": 1.404778242111206,
      "learning_rate": 0.0005,
      "loss": 2.603219509124756,
      "step": 3644
    },
    {
      "epoch": 0.5880929332042595,
      "grad_norm": 1.5589900016784668,
      "learning_rate": 0.0005,
      "loss": 2.9709391593933105,
      "step": 3645
    },
    {
      "epoch": 0.5882542755727654,
      "grad_norm": 1.2017650604248047,
      "learning_rate": 0.0005,
      "loss": 2.6650705337524414,
      "step": 3646
    },
    {
      "epoch": 0.5884156179412714,
      "grad_norm": 1.2813892364501953,
      "learning_rate": 0.0005,
      "loss": 2.4814653396606445,
      "step": 3647
    },
    {
      "epoch": 0.5885769603097774,
      "grad_norm": 1.082553505897522,
      "learning_rate": 0.0005,
      "loss": 2.846703052520752,
      "step": 3648
    },
    {
      "epoch": 0.5887383026782833,
      "grad_norm": 1.0619248151779175,
      "learning_rate": 0.0005,
      "loss": 3.0813026428222656,
      "step": 3649
    },
    {
      "epoch": 0.5888996450467893,
      "grad_norm": 1.3823543787002563,
      "learning_rate": 0.0005,
      "loss": 2.438260078430176,
      "step": 3650
    },
    {
      "epoch": 0.5890609874152952,
      "grad_norm": 1.045714020729065,
      "learning_rate": 0.0005,
      "loss": 2.4687161445617676,
      "step": 3651
    },
    {
      "epoch": 0.5892223297838012,
      "grad_norm": 1.3530722856521606,
      "learning_rate": 0.0005,
      "loss": 2.5036041736602783,
      "step": 3652
    },
    {
      "epoch": 0.5893836721523072,
      "grad_norm": 1.7117170095443726,
      "learning_rate": 0.0005,
      "loss": 2.7731127738952637,
      "step": 3653
    },
    {
      "epoch": 0.5895450145208132,
      "grad_norm": 1.3954240083694458,
      "learning_rate": 0.0005,
      "loss": 3.1035494804382324,
      "step": 3654
    },
    {
      "epoch": 0.5897063568893192,
      "grad_norm": 1.153395414352417,
      "learning_rate": 0.0005,
      "loss": 2.565697193145752,
      "step": 3655
    },
    {
      "epoch": 0.5898676992578251,
      "grad_norm": 1.5035539865493774,
      "learning_rate": 0.0005,
      "loss": 2.9483847618103027,
      "step": 3656
    },
    {
      "epoch": 0.5900290416263311,
      "grad_norm": 1.1206144094467163,
      "learning_rate": 0.0005,
      "loss": 2.609384536743164,
      "step": 3657
    },
    {
      "epoch": 0.590190383994837,
      "grad_norm": 1.2951180934906006,
      "learning_rate": 0.0005,
      "loss": 2.743457317352295,
      "step": 3658
    },
    {
      "epoch": 0.590351726363343,
      "grad_norm": 1.276469111442566,
      "learning_rate": 0.0005,
      "loss": 2.978257179260254,
      "step": 3659
    },
    {
      "epoch": 0.590513068731849,
      "grad_norm": 0.9803788661956787,
      "learning_rate": 0.0005,
      "loss": 2.651036262512207,
      "step": 3660
    },
    {
      "epoch": 0.5906744111003549,
      "grad_norm": 1.245599627494812,
      "learning_rate": 0.0005,
      "loss": 2.5316011905670166,
      "step": 3661
    },
    {
      "epoch": 0.590835753468861,
      "grad_norm": 1.0239133834838867,
      "learning_rate": 0.0005,
      "loss": 2.78739070892334,
      "step": 3662
    },
    {
      "epoch": 0.5909970958373669,
      "grad_norm": 1.4198709726333618,
      "learning_rate": 0.0005,
      "loss": 2.8994216918945312,
      "step": 3663
    },
    {
      "epoch": 0.5911584382058729,
      "grad_norm": 1.4448329210281372,
      "learning_rate": 0.0005,
      "loss": 3.1119368076324463,
      "step": 3664
    },
    {
      "epoch": 0.5913197805743788,
      "grad_norm": 1.5730026960372925,
      "learning_rate": 0.0005,
      "loss": 2.856663227081299,
      "step": 3665
    },
    {
      "epoch": 0.5914811229428848,
      "grad_norm": 1.1531476974487305,
      "learning_rate": 0.0005,
      "loss": 2.8348770141601562,
      "step": 3666
    },
    {
      "epoch": 0.5916424653113908,
      "grad_norm": 1.2997983694076538,
      "learning_rate": 0.0005,
      "loss": 2.491882085800171,
      "step": 3667
    },
    {
      "epoch": 0.5918038076798967,
      "grad_norm": 1.1217118501663208,
      "learning_rate": 0.0005,
      "loss": 2.9194462299346924,
      "step": 3668
    },
    {
      "epoch": 0.5919651500484027,
      "grad_norm": 1.1871799230575562,
      "learning_rate": 0.0005,
      "loss": 2.734447479248047,
      "step": 3669
    },
    {
      "epoch": 0.5921264924169087,
      "grad_norm": 1.1250338554382324,
      "learning_rate": 0.0005,
      "loss": 2.660898208618164,
      "step": 3670
    },
    {
      "epoch": 0.5922878347854147,
      "grad_norm": 1.298371434211731,
      "learning_rate": 0.0005,
      "loss": 2.2524147033691406,
      "step": 3671
    },
    {
      "epoch": 0.5924491771539206,
      "grad_norm": 1.2489581108093262,
      "learning_rate": 0.0005,
      "loss": 2.5912835597991943,
      "step": 3672
    },
    {
      "epoch": 0.5926105195224266,
      "grad_norm": 1.1640453338623047,
      "learning_rate": 0.0005,
      "loss": 2.960122585296631,
      "step": 3673
    },
    {
      "epoch": 0.5927718618909326,
      "grad_norm": 1.3343456983566284,
      "learning_rate": 0.0005,
      "loss": 2.7103636264801025,
      "step": 3674
    },
    {
      "epoch": 0.5929332042594385,
      "grad_norm": 1.026445746421814,
      "learning_rate": 0.0005,
      "loss": 2.6302311420440674,
      "step": 3675
    },
    {
      "epoch": 0.5930945466279445,
      "grad_norm": 0.9873700737953186,
      "learning_rate": 0.0005,
      "loss": 2.5431363582611084,
      "step": 3676
    },
    {
      "epoch": 0.5932558889964504,
      "grad_norm": 1.4178926944732666,
      "learning_rate": 0.0005,
      "loss": 2.5241541862487793,
      "step": 3677
    },
    {
      "epoch": 0.5934172313649564,
      "grad_norm": 1.467764973640442,
      "learning_rate": 0.0005,
      "loss": 2.8949203491210938,
      "step": 3678
    },
    {
      "epoch": 0.5935785737334625,
      "grad_norm": 1.0983672142028809,
      "learning_rate": 0.0005,
      "loss": 2.661008834838867,
      "step": 3679
    },
    {
      "epoch": 0.5937399161019684,
      "grad_norm": 1.4196531772613525,
      "learning_rate": 0.0005,
      "loss": 2.667189836502075,
      "step": 3680
    },
    {
      "epoch": 0.5939012584704744,
      "grad_norm": 1.3880221843719482,
      "learning_rate": 0.0005,
      "loss": 2.752558946609497,
      "step": 3681
    },
    {
      "epoch": 0.5940626008389803,
      "grad_norm": 2.220085382461548,
      "learning_rate": 0.0005,
      "loss": 2.8205432891845703,
      "step": 3682
    },
    {
      "epoch": 0.5942239432074863,
      "grad_norm": 1.4109833240509033,
      "learning_rate": 0.0005,
      "loss": 2.799222946166992,
      "step": 3683
    },
    {
      "epoch": 0.5943852855759922,
      "grad_norm": 2.0130677223205566,
      "learning_rate": 0.0005,
      "loss": 2.9813179969787598,
      "step": 3684
    },
    {
      "epoch": 0.5945466279444982,
      "grad_norm": 2.0859220027923584,
      "learning_rate": 0.0005,
      "loss": 2.785860061645508,
      "step": 3685
    },
    {
      "epoch": 0.5947079703130042,
      "grad_norm": 1.1427115201950073,
      "learning_rate": 0.0005,
      "loss": 2.6722769737243652,
      "step": 3686
    },
    {
      "epoch": 0.5948693126815101,
      "grad_norm": 1.1532514095306396,
      "learning_rate": 0.0005,
      "loss": 2.589545249938965,
      "step": 3687
    },
    {
      "epoch": 0.5950306550500162,
      "grad_norm": 1.385259747505188,
      "learning_rate": 0.0005,
      "loss": 2.2016761302948,
      "step": 3688
    },
    {
      "epoch": 0.5951919974185221,
      "grad_norm": 1.3239706754684448,
      "learning_rate": 0.0005,
      "loss": 2.7877309322357178,
      "step": 3689
    },
    {
      "epoch": 0.5953533397870281,
      "grad_norm": 1.3108936548233032,
      "learning_rate": 0.0005,
      "loss": 2.4758033752441406,
      "step": 3690
    },
    {
      "epoch": 0.595514682155534,
      "grad_norm": 1.2842862606048584,
      "learning_rate": 0.0005,
      "loss": 2.8020412921905518,
      "step": 3691
    },
    {
      "epoch": 0.59567602452404,
      "grad_norm": 0.9767389297485352,
      "learning_rate": 0.0005,
      "loss": 3.055651903152466,
      "step": 3692
    },
    {
      "epoch": 0.595837366892546,
      "grad_norm": 1.209513545036316,
      "learning_rate": 0.0005,
      "loss": 2.900484323501587,
      "step": 3693
    },
    {
      "epoch": 0.5959987092610519,
      "grad_norm": 2.064495801925659,
      "learning_rate": 0.0005,
      "loss": 2.7658162117004395,
      "step": 3694
    },
    {
      "epoch": 0.5961600516295579,
      "grad_norm": 1.1408581733703613,
      "learning_rate": 0.0005,
      "loss": 2.9247915744781494,
      "step": 3695
    },
    {
      "epoch": 0.5963213939980639,
      "grad_norm": 30.629337310791016,
      "learning_rate": 0.0005,
      "loss": 2.686084747314453,
      "step": 3696
    },
    {
      "epoch": 0.5964827363665699,
      "grad_norm": 1.1601418256759644,
      "learning_rate": 0.0005,
      "loss": 2.8357410430908203,
      "step": 3697
    },
    {
      "epoch": 0.5966440787350759,
      "grad_norm": 1.1051270961761475,
      "learning_rate": 0.0005,
      "loss": 2.446290969848633,
      "step": 3698
    },
    {
      "epoch": 0.5968054211035818,
      "grad_norm": 1.186638355255127,
      "learning_rate": 0.0005,
      "loss": 2.7221484184265137,
      "step": 3699
    },
    {
      "epoch": 0.5969667634720878,
      "grad_norm": 1.125732660293579,
      "learning_rate": 0.0005,
      "loss": 2.5006794929504395,
      "step": 3700
    },
    {
      "epoch": 0.5971281058405937,
      "grad_norm": 1.2024061679840088,
      "learning_rate": 0.0005,
      "loss": 2.442680835723877,
      "step": 3701
    },
    {
      "epoch": 0.5972894482090997,
      "grad_norm": 1.2315881252288818,
      "learning_rate": 0.0005,
      "loss": 2.3632593154907227,
      "step": 3702
    },
    {
      "epoch": 0.5974507905776056,
      "grad_norm": 1.4602084159851074,
      "learning_rate": 0.0005,
      "loss": 2.663987159729004,
      "step": 3703
    },
    {
      "epoch": 0.5976121329461116,
      "grad_norm": 1.229608416557312,
      "learning_rate": 0.0005,
      "loss": 2.35079288482666,
      "step": 3704
    },
    {
      "epoch": 0.5977734753146177,
      "grad_norm": 2.177072048187256,
      "learning_rate": 0.0005,
      "loss": 2.7465550899505615,
      "step": 3705
    },
    {
      "epoch": 0.5979348176831236,
      "grad_norm": 1.0926330089569092,
      "learning_rate": 0.0005,
      "loss": 2.4796462059020996,
      "step": 3706
    },
    {
      "epoch": 0.5980961600516296,
      "grad_norm": 1.2240058183670044,
      "learning_rate": 0.0005,
      "loss": 2.8063597679138184,
      "step": 3707
    },
    {
      "epoch": 0.5982575024201355,
      "grad_norm": 1.1195679903030396,
      "learning_rate": 0.0005,
      "loss": 2.815586566925049,
      "step": 3708
    },
    {
      "epoch": 0.5984188447886415,
      "grad_norm": 1.2288588285446167,
      "learning_rate": 0.0005,
      "loss": 2.7165470123291016,
      "step": 3709
    },
    {
      "epoch": 0.5985801871571474,
      "grad_norm": 1.1862882375717163,
      "learning_rate": 0.0005,
      "loss": 3.2592415809631348,
      "step": 3710
    },
    {
      "epoch": 0.5987415295256534,
      "grad_norm": 1.1605567932128906,
      "learning_rate": 0.0005,
      "loss": 2.2337899208068848,
      "step": 3711
    },
    {
      "epoch": 0.5989028718941594,
      "grad_norm": 1.2000473737716675,
      "learning_rate": 0.0005,
      "loss": 2.3837742805480957,
      "step": 3712
    },
    {
      "epoch": 0.5990642142626654,
      "grad_norm": 1.1706308126449585,
      "learning_rate": 0.0005,
      "loss": 3.0371453762054443,
      "step": 3713
    },
    {
      "epoch": 0.5992255566311714,
      "grad_norm": 1.4942543506622314,
      "learning_rate": 0.0005,
      "loss": 2.8248980045318604,
      "step": 3714
    },
    {
      "epoch": 0.5993868989996773,
      "grad_norm": 2.1054511070251465,
      "learning_rate": 0.0005,
      "loss": 3.120208263397217,
      "step": 3715
    },
    {
      "epoch": 0.5995482413681833,
      "grad_norm": 1.0965778827667236,
      "learning_rate": 0.0005,
      "loss": 2.7289328575134277,
      "step": 3716
    },
    {
      "epoch": 0.5997095837366893,
      "grad_norm": 1.2285807132720947,
      "learning_rate": 0.0005,
      "loss": 2.629380464553833,
      "step": 3717
    },
    {
      "epoch": 0.5998709261051952,
      "grad_norm": 1.1385644674301147,
      "learning_rate": 0.0005,
      "loss": 2.6008121967315674,
      "step": 3718
    },
    {
      "epoch": 0.6000322684737012,
      "grad_norm": 1.1822843551635742,
      "learning_rate": 0.0005,
      "loss": 2.7841830253601074,
      "step": 3719
    },
    {
      "epoch": 0.6001936108422071,
      "grad_norm": 0.9519827365875244,
      "learning_rate": 0.0005,
      "loss": 2.9315028190612793,
      "step": 3720
    },
    {
      "epoch": 0.6003549532107131,
      "grad_norm": 1.4217631816864014,
      "learning_rate": 0.0005,
      "loss": 2.655076026916504,
      "step": 3721
    },
    {
      "epoch": 0.6005162955792192,
      "grad_norm": 1.0120328664779663,
      "learning_rate": 0.0005,
      "loss": 2.677894115447998,
      "step": 3722
    },
    {
      "epoch": 0.6006776379477251,
      "grad_norm": 1.1222628355026245,
      "learning_rate": 0.0005,
      "loss": 2.7657718658447266,
      "step": 3723
    },
    {
      "epoch": 0.6008389803162311,
      "grad_norm": 1.2610657215118408,
      "learning_rate": 0.0005,
      "loss": 2.5227367877960205,
      "step": 3724
    },
    {
      "epoch": 0.601000322684737,
      "grad_norm": 1.2196580171585083,
      "learning_rate": 0.0005,
      "loss": 2.7086668014526367,
      "step": 3725
    },
    {
      "epoch": 0.601161665053243,
      "grad_norm": 1.0112900733947754,
      "learning_rate": 0.0005,
      "loss": 2.5902130603790283,
      "step": 3726
    },
    {
      "epoch": 0.6013230074217489,
      "grad_norm": 1.0788341760635376,
      "learning_rate": 0.0005,
      "loss": 2.863689422607422,
      "step": 3727
    },
    {
      "epoch": 0.6014843497902549,
      "grad_norm": 1.4695569276809692,
      "learning_rate": 0.0005,
      "loss": 2.7338035106658936,
      "step": 3728
    },
    {
      "epoch": 0.6016456921587608,
      "grad_norm": 1.0083938837051392,
      "learning_rate": 0.0005,
      "loss": 2.6891703605651855,
      "step": 3729
    },
    {
      "epoch": 0.6018070345272669,
      "grad_norm": 1.2636561393737793,
      "learning_rate": 0.0005,
      "loss": 3.152759552001953,
      "step": 3730
    },
    {
      "epoch": 0.6019683768957729,
      "grad_norm": 2.030977964401245,
      "learning_rate": 0.0005,
      "loss": 3.1321423053741455,
      "step": 3731
    },
    {
      "epoch": 0.6021297192642788,
      "grad_norm": 1.3821982145309448,
      "learning_rate": 0.0005,
      "loss": 2.309511423110962,
      "step": 3732
    },
    {
      "epoch": 0.6022910616327848,
      "grad_norm": 1.438372015953064,
      "learning_rate": 0.0005,
      "loss": 2.5263514518737793,
      "step": 3733
    },
    {
      "epoch": 0.6024524040012907,
      "grad_norm": 1.1919193267822266,
      "learning_rate": 0.0005,
      "loss": 2.729119300842285,
      "step": 3734
    },
    {
      "epoch": 0.6026137463697967,
      "grad_norm": 1.5766830444335938,
      "learning_rate": 0.0005,
      "loss": 3.149024486541748,
      "step": 3735
    },
    {
      "epoch": 0.6027750887383027,
      "grad_norm": 1.6341010332107544,
      "learning_rate": 0.0005,
      "loss": 2.5020313262939453,
      "step": 3736
    },
    {
      "epoch": 0.6029364311068086,
      "grad_norm": 1.3111978769302368,
      "learning_rate": 0.0005,
      "loss": 2.928859233856201,
      "step": 3737
    },
    {
      "epoch": 0.6030977734753146,
      "grad_norm": 1.17197585105896,
      "learning_rate": 0.0005,
      "loss": 2.4494190216064453,
      "step": 3738
    },
    {
      "epoch": 0.6032591158438206,
      "grad_norm": 1.3321765661239624,
      "learning_rate": 0.0005,
      "loss": 2.447707176208496,
      "step": 3739
    },
    {
      "epoch": 0.6034204582123266,
      "grad_norm": 1.2440131902694702,
      "learning_rate": 0.0005,
      "loss": 2.6303036212921143,
      "step": 3740
    },
    {
      "epoch": 0.6035818005808326,
      "grad_norm": 0.8621412515640259,
      "learning_rate": 0.0005,
      "loss": 2.485356330871582,
      "step": 3741
    },
    {
      "epoch": 0.6037431429493385,
      "grad_norm": 1.122570514678955,
      "learning_rate": 0.0005,
      "loss": 2.7302231788635254,
      "step": 3742
    },
    {
      "epoch": 0.6039044853178445,
      "grad_norm": 1.1475566625595093,
      "learning_rate": 0.0005,
      "loss": 2.288069486618042,
      "step": 3743
    },
    {
      "epoch": 0.6040658276863504,
      "grad_norm": 1.0733305215835571,
      "learning_rate": 0.0005,
      "loss": 2.5573019981384277,
      "step": 3744
    },
    {
      "epoch": 0.6042271700548564,
      "grad_norm": 1.5402977466583252,
      "learning_rate": 0.0005,
      "loss": 3.258601427078247,
      "step": 3745
    },
    {
      "epoch": 0.6043885124233623,
      "grad_norm": 1.2511507272720337,
      "learning_rate": 0.0005,
      "loss": 2.568826198577881,
      "step": 3746
    },
    {
      "epoch": 0.6045498547918683,
      "grad_norm": 1.5361621379852295,
      "learning_rate": 0.0005,
      "loss": 2.389094829559326,
      "step": 3747
    },
    {
      "epoch": 0.6047111971603744,
      "grad_norm": 0.9328612089157104,
      "learning_rate": 0.0005,
      "loss": 2.541323661804199,
      "step": 3748
    },
    {
      "epoch": 0.6048725395288803,
      "grad_norm": 1.4231780767440796,
      "learning_rate": 0.0005,
      "loss": 2.908295154571533,
      "step": 3749
    },
    {
      "epoch": 0.6050338818973863,
      "grad_norm": 1.411826729774475,
      "learning_rate": 0.0005,
      "loss": 2.5019216537475586,
      "step": 3750
    },
    {
      "epoch": 0.6051952242658922,
      "grad_norm": 1.910658597946167,
      "learning_rate": 0.0005,
      "loss": 2.6662442684173584,
      "step": 3751
    },
    {
      "epoch": 0.6053565666343982,
      "grad_norm": 1.1262799501419067,
      "learning_rate": 0.0005,
      "loss": 2.5098907947540283,
      "step": 3752
    },
    {
      "epoch": 0.6055179090029041,
      "grad_norm": 1.3484734296798706,
      "learning_rate": 0.0005,
      "loss": 2.4361438751220703,
      "step": 3753
    },
    {
      "epoch": 0.6056792513714101,
      "grad_norm": 1.151140809059143,
      "learning_rate": 0.0005,
      "loss": 3.2066454887390137,
      "step": 3754
    },
    {
      "epoch": 0.6058405937399161,
      "grad_norm": 1.0774719715118408,
      "learning_rate": 0.0005,
      "loss": 2.7458629608154297,
      "step": 3755
    },
    {
      "epoch": 0.6060019361084221,
      "grad_norm": 1.3378875255584717,
      "learning_rate": 0.0005,
      "loss": 3.068129062652588,
      "step": 3756
    },
    {
      "epoch": 0.6061632784769281,
      "grad_norm": 1.4780601263046265,
      "learning_rate": 0.0005,
      "loss": 2.7099218368530273,
      "step": 3757
    },
    {
      "epoch": 0.606324620845434,
      "grad_norm": 1.1345949172973633,
      "learning_rate": 0.0005,
      "loss": 2.638115406036377,
      "step": 3758
    },
    {
      "epoch": 0.60648596321394,
      "grad_norm": 1.1853855848312378,
      "learning_rate": 0.0005,
      "loss": 2.567164421081543,
      "step": 3759
    },
    {
      "epoch": 0.606647305582446,
      "grad_norm": 1.0248960256576538,
      "learning_rate": 0.0005,
      "loss": 3.0483791828155518,
      "step": 3760
    },
    {
      "epoch": 0.6068086479509519,
      "grad_norm": 1.3073768615722656,
      "learning_rate": 0.0005,
      "loss": 2.8231630325317383,
      "step": 3761
    },
    {
      "epoch": 0.6069699903194579,
      "grad_norm": 1.1957067251205444,
      "learning_rate": 0.0005,
      "loss": 2.4760775566101074,
      "step": 3762
    },
    {
      "epoch": 0.6071313326879638,
      "grad_norm": 1.0793569087982178,
      "learning_rate": 0.0005,
      "loss": 2.371321678161621,
      "step": 3763
    },
    {
      "epoch": 0.6072926750564698,
      "grad_norm": 1.2482690811157227,
      "learning_rate": 0.0005,
      "loss": 2.866359233856201,
      "step": 3764
    },
    {
      "epoch": 0.6074540174249758,
      "grad_norm": 1.7143714427947998,
      "learning_rate": 0.0005,
      "loss": 2.6607627868652344,
      "step": 3765
    },
    {
      "epoch": 0.6076153597934818,
      "grad_norm": 1.2023820877075195,
      "learning_rate": 0.0005,
      "loss": 2.797150135040283,
      "step": 3766
    },
    {
      "epoch": 0.6077767021619878,
      "grad_norm": 1.0133564472198486,
      "learning_rate": 0.0005,
      "loss": 2.5277533531188965,
      "step": 3767
    },
    {
      "epoch": 0.6079380445304937,
      "grad_norm": 1.1326587200164795,
      "learning_rate": 0.0005,
      "loss": 2.6390719413757324,
      "step": 3768
    },
    {
      "epoch": 0.6080993868989997,
      "grad_norm": 1.0364043712615967,
      "learning_rate": 0.0005,
      "loss": 3.158231258392334,
      "step": 3769
    },
    {
      "epoch": 0.6082607292675056,
      "grad_norm": 0.9770113229751587,
      "learning_rate": 0.0005,
      "loss": 2.8794522285461426,
      "step": 3770
    },
    {
      "epoch": 0.6084220716360116,
      "grad_norm": 1.264951467514038,
      "learning_rate": 0.0005,
      "loss": 3.2903361320495605,
      "step": 3771
    },
    {
      "epoch": 0.6085834140045175,
      "grad_norm": 1.2422763109207153,
      "learning_rate": 0.0005,
      "loss": 2.5705137252807617,
      "step": 3772
    },
    {
      "epoch": 0.6087447563730236,
      "grad_norm": 1.3008959293365479,
      "learning_rate": 0.0005,
      "loss": 2.7824559211730957,
      "step": 3773
    },
    {
      "epoch": 0.6089060987415296,
      "grad_norm": 1.8886756896972656,
      "learning_rate": 0.0005,
      "loss": 2.9607794284820557,
      "step": 3774
    },
    {
      "epoch": 0.6090674411100355,
      "grad_norm": 1.4582960605621338,
      "learning_rate": 0.0005,
      "loss": 3.094630718231201,
      "step": 3775
    },
    {
      "epoch": 0.6092287834785415,
      "grad_norm": 1.1212199926376343,
      "learning_rate": 0.0005,
      "loss": 2.6063408851623535,
      "step": 3776
    },
    {
      "epoch": 0.6093901258470474,
      "grad_norm": 1.6267971992492676,
      "learning_rate": 0.0005,
      "loss": 2.993720054626465,
      "step": 3777
    },
    {
      "epoch": 0.6095514682155534,
      "grad_norm": 1.351336121559143,
      "learning_rate": 0.0005,
      "loss": 2.567901611328125,
      "step": 3778
    },
    {
      "epoch": 0.6097128105840594,
      "grad_norm": 1.0658665895462036,
      "learning_rate": 0.0005,
      "loss": 2.8615686893463135,
      "step": 3779
    },
    {
      "epoch": 0.6098741529525653,
      "grad_norm": 1.7254798412322998,
      "learning_rate": 0.0005,
      "loss": 2.397345781326294,
      "step": 3780
    },
    {
      "epoch": 0.6100354953210713,
      "grad_norm": 1.8896609544754028,
      "learning_rate": 0.0005,
      "loss": 3.1694321632385254,
      "step": 3781
    },
    {
      "epoch": 0.6101968376895773,
      "grad_norm": 1.1394188404083252,
      "learning_rate": 0.0005,
      "loss": 2.3897576332092285,
      "step": 3782
    },
    {
      "epoch": 0.6103581800580833,
      "grad_norm": 1.1209226846694946,
      "learning_rate": 0.0005,
      "loss": 2.1788787841796875,
      "step": 3783
    },
    {
      "epoch": 0.6105195224265892,
      "grad_norm": 1.6085447072982788,
      "learning_rate": 0.0005,
      "loss": 2.7529075145721436,
      "step": 3784
    },
    {
      "epoch": 0.6106808647950952,
      "grad_norm": 1.1969352960586548,
      "learning_rate": 0.0005,
      "loss": 2.588510036468506,
      "step": 3785
    },
    {
      "epoch": 0.6108422071636012,
      "grad_norm": 1.2287949323654175,
      "learning_rate": 0.0005,
      "loss": 3.1716480255126953,
      "step": 3786
    },
    {
      "epoch": 0.6110035495321071,
      "grad_norm": 1.3462907075881958,
      "learning_rate": 0.0005,
      "loss": 2.8195555210113525,
      "step": 3787
    },
    {
      "epoch": 0.6111648919006131,
      "grad_norm": 1.5242223739624023,
      "learning_rate": 0.0005,
      "loss": 2.597893714904785,
      "step": 3788
    },
    {
      "epoch": 0.611326234269119,
      "grad_norm": 1.473102331161499,
      "learning_rate": 0.0005,
      "loss": 2.544617176055908,
      "step": 3789
    },
    {
      "epoch": 0.611487576637625,
      "grad_norm": 1.1062753200531006,
      "learning_rate": 0.0005,
      "loss": 2.9961507320404053,
      "step": 3790
    },
    {
      "epoch": 0.611648919006131,
      "grad_norm": 1.1857904195785522,
      "learning_rate": 0.0005,
      "loss": 2.274324417114258,
      "step": 3791
    },
    {
      "epoch": 0.611810261374637,
      "grad_norm": 1.4248452186584473,
      "learning_rate": 0.0005,
      "loss": 2.6973438262939453,
      "step": 3792
    },
    {
      "epoch": 0.611971603743143,
      "grad_norm": 1.232481598854065,
      "learning_rate": 0.0005,
      "loss": 2.5416955947875977,
      "step": 3793
    },
    {
      "epoch": 0.6121329461116489,
      "grad_norm": 1.4460363388061523,
      "learning_rate": 0.0005,
      "loss": 3.0042715072631836,
      "step": 3794
    },
    {
      "epoch": 0.6122942884801549,
      "grad_norm": 1.6305677890777588,
      "learning_rate": 0.0005,
      "loss": 2.9426608085632324,
      "step": 3795
    },
    {
      "epoch": 0.6124556308486608,
      "grad_norm": 1.491310715675354,
      "learning_rate": 0.0005,
      "loss": 3.032541275024414,
      "step": 3796
    },
    {
      "epoch": 0.6126169732171668,
      "grad_norm": 1.0295337438583374,
      "learning_rate": 0.0005,
      "loss": 2.3813095092773438,
      "step": 3797
    },
    {
      "epoch": 0.6127783155856728,
      "grad_norm": 1.5843567848205566,
      "learning_rate": 0.0005,
      "loss": 2.4418749809265137,
      "step": 3798
    },
    {
      "epoch": 0.6129396579541788,
      "grad_norm": 1.305608868598938,
      "learning_rate": 0.0005,
      "loss": 2.8817336559295654,
      "step": 3799
    },
    {
      "epoch": 0.6131010003226848,
      "grad_norm": 4.398832321166992,
      "learning_rate": 0.0005,
      "loss": 2.7574329376220703,
      "step": 3800
    },
    {
      "epoch": 0.6132623426911907,
      "grad_norm": 1.4183349609375,
      "learning_rate": 0.0005,
      "loss": 2.6459319591522217,
      "step": 3801
    },
    {
      "epoch": 0.6134236850596967,
      "grad_norm": 2.9032583236694336,
      "learning_rate": 0.0005,
      "loss": 2.5904834270477295,
      "step": 3802
    },
    {
      "epoch": 0.6135850274282026,
      "grad_norm": 1.54672372341156,
      "learning_rate": 0.0005,
      "loss": 2.7536635398864746,
      "step": 3803
    },
    {
      "epoch": 0.6137463697967086,
      "grad_norm": 1.328546166419983,
      "learning_rate": 0.0005,
      "loss": 2.5720529556274414,
      "step": 3804
    },
    {
      "epoch": 0.6139077121652146,
      "grad_norm": 1.2707008123397827,
      "learning_rate": 0.0005,
      "loss": 2.8770265579223633,
      "step": 3805
    },
    {
      "epoch": 0.6140690545337205,
      "grad_norm": 1.544185757637024,
      "learning_rate": 0.0005,
      "loss": 3.0895133018493652,
      "step": 3806
    },
    {
      "epoch": 0.6142303969022265,
      "grad_norm": 4.648969650268555,
      "learning_rate": 0.0005,
      "loss": 2.779472827911377,
      "step": 3807
    },
    {
      "epoch": 0.6143917392707325,
      "grad_norm": 2.42482328414917,
      "learning_rate": 0.0005,
      "loss": 2.5459256172180176,
      "step": 3808
    },
    {
      "epoch": 0.6145530816392385,
      "grad_norm": 1.5612397193908691,
      "learning_rate": 0.0005,
      "loss": 2.607630968093872,
      "step": 3809
    },
    {
      "epoch": 0.6147144240077445,
      "grad_norm": 1.8349354267120361,
      "learning_rate": 0.0005,
      "loss": 2.9628288745880127,
      "step": 3810
    },
    {
      "epoch": 0.6148757663762504,
      "grad_norm": 1.5031217336654663,
      "learning_rate": 0.0005,
      "loss": 2.960308790206909,
      "step": 3811
    },
    {
      "epoch": 0.6150371087447564,
      "grad_norm": 1.3040437698364258,
      "learning_rate": 0.0005,
      "loss": 2.718681812286377,
      "step": 3812
    },
    {
      "epoch": 0.6151984511132623,
      "grad_norm": 1.327523946762085,
      "learning_rate": 0.0005,
      "loss": 2.656959056854248,
      "step": 3813
    },
    {
      "epoch": 0.6153597934817683,
      "grad_norm": 1.1053439378738403,
      "learning_rate": 0.0005,
      "loss": 2.620633125305176,
      "step": 3814
    },
    {
      "epoch": 0.6155211358502742,
      "grad_norm": 1.276076316833496,
      "learning_rate": 0.0005,
      "loss": 2.205544948577881,
      "step": 3815
    },
    {
      "epoch": 0.6156824782187803,
      "grad_norm": 0.9578116536140442,
      "learning_rate": 0.0005,
      "loss": 2.4602136611938477,
      "step": 3816
    },
    {
      "epoch": 0.6158438205872863,
      "grad_norm": 1.4942481517791748,
      "learning_rate": 0.0005,
      "loss": 2.8070878982543945,
      "step": 3817
    },
    {
      "epoch": 0.6160051629557922,
      "grad_norm": 1.216132402420044,
      "learning_rate": 0.0005,
      "loss": 2.7347795963287354,
      "step": 3818
    },
    {
      "epoch": 0.6161665053242982,
      "grad_norm": 1.4224166870117188,
      "learning_rate": 0.0005,
      "loss": 2.5045886039733887,
      "step": 3819
    },
    {
      "epoch": 0.6163278476928041,
      "grad_norm": 1.242068886756897,
      "learning_rate": 0.0005,
      "loss": 2.8136649131774902,
      "step": 3820
    },
    {
      "epoch": 0.6164891900613101,
      "grad_norm": 1.2453372478485107,
      "learning_rate": 0.0005,
      "loss": 2.9026906490325928,
      "step": 3821
    },
    {
      "epoch": 0.616650532429816,
      "grad_norm": 1.8395112752914429,
      "learning_rate": 0.0005,
      "loss": 2.7117490768432617,
      "step": 3822
    },
    {
      "epoch": 0.616811874798322,
      "grad_norm": 1.472429871559143,
      "learning_rate": 0.0005,
      "loss": 2.4103341102600098,
      "step": 3823
    },
    {
      "epoch": 0.616973217166828,
      "grad_norm": 1.0578513145446777,
      "learning_rate": 0.0005,
      "loss": 2.343116044998169,
      "step": 3824
    },
    {
      "epoch": 0.617134559535334,
      "grad_norm": 1.2291617393493652,
      "learning_rate": 0.0005,
      "loss": 2.736586570739746,
      "step": 3825
    },
    {
      "epoch": 0.61729590190384,
      "grad_norm": 1.208422064781189,
      "learning_rate": 0.0005,
      "loss": 3.0347182750701904,
      "step": 3826
    },
    {
      "epoch": 0.6174572442723459,
      "grad_norm": 1.1090425252914429,
      "learning_rate": 0.0005,
      "loss": 3.0601463317871094,
      "step": 3827
    },
    {
      "epoch": 0.6176185866408519,
      "grad_norm": 1.7315694093704224,
      "learning_rate": 0.0005,
      "loss": 2.396897792816162,
      "step": 3828
    },
    {
      "epoch": 0.6177799290093579,
      "grad_norm": 1.264380693435669,
      "learning_rate": 0.0005,
      "loss": 2.5045580863952637,
      "step": 3829
    },
    {
      "epoch": 0.6179412713778638,
      "grad_norm": 1.3713082075119019,
      "learning_rate": 0.0005,
      "loss": 3.242152214050293,
      "step": 3830
    },
    {
      "epoch": 0.6181026137463698,
      "grad_norm": 1.1254383325576782,
      "learning_rate": 0.0005,
      "loss": 2.2662861347198486,
      "step": 3831
    },
    {
      "epoch": 0.6182639561148757,
      "grad_norm": 2.3678815364837646,
      "learning_rate": 0.0005,
      "loss": 2.428558349609375,
      "step": 3832
    },
    {
      "epoch": 0.6184252984833818,
      "grad_norm": 1.5298588275909424,
      "learning_rate": 0.0005,
      "loss": 2.7377841472625732,
      "step": 3833
    },
    {
      "epoch": 0.6185866408518877,
      "grad_norm": 1.4367578029632568,
      "learning_rate": 0.0005,
      "loss": 2.644541025161743,
      "step": 3834
    },
    {
      "epoch": 0.6187479832203937,
      "grad_norm": 1.5423458814620972,
      "learning_rate": 0.0005,
      "loss": 3.093238353729248,
      "step": 3835
    },
    {
      "epoch": 0.6189093255888997,
      "grad_norm": 1.325042724609375,
      "learning_rate": 0.0005,
      "loss": 3.0917325019836426,
      "step": 3836
    },
    {
      "epoch": 0.6190706679574056,
      "grad_norm": 1.128402590751648,
      "learning_rate": 0.0005,
      "loss": 2.8015828132629395,
      "step": 3837
    },
    {
      "epoch": 0.6192320103259116,
      "grad_norm": 1.708260416984558,
      "learning_rate": 0.0005,
      "loss": 2.9183695316314697,
      "step": 3838
    },
    {
      "epoch": 0.6193933526944175,
      "grad_norm": 1.6278549432754517,
      "learning_rate": 0.0005,
      "loss": 2.580235242843628,
      "step": 3839
    },
    {
      "epoch": 0.6195546950629235,
      "grad_norm": 1.2783123254776,
      "learning_rate": 0.0005,
      "loss": 2.8567349910736084,
      "step": 3840
    },
    {
      "epoch": 0.6197160374314294,
      "grad_norm": 1.9687976837158203,
      "learning_rate": 0.0005,
      "loss": 3.0603485107421875,
      "step": 3841
    },
    {
      "epoch": 0.6198773797999355,
      "grad_norm": 1.2173128128051758,
      "learning_rate": 0.0005,
      "loss": 2.315993070602417,
      "step": 3842
    },
    {
      "epoch": 0.6200387221684415,
      "grad_norm": 2.2828304767608643,
      "learning_rate": 0.0005,
      "loss": 2.6829721927642822,
      "step": 3843
    },
    {
      "epoch": 0.6202000645369474,
      "grad_norm": 0.8358250260353088,
      "learning_rate": 0.0005,
      "loss": 2.471184730529785,
      "step": 3844
    },
    {
      "epoch": 0.6203614069054534,
      "grad_norm": 1.1829602718353271,
      "learning_rate": 0.0005,
      "loss": 2.6533455848693848,
      "step": 3845
    },
    {
      "epoch": 0.6205227492739593,
      "grad_norm": 1.0445291996002197,
      "learning_rate": 0.0005,
      "loss": 2.5788729190826416,
      "step": 3846
    },
    {
      "epoch": 0.6206840916424653,
      "grad_norm": 1.3116199970245361,
      "learning_rate": 0.0005,
      "loss": 2.4064671993255615,
      "step": 3847
    },
    {
      "epoch": 0.6208454340109713,
      "grad_norm": 0.9407937526702881,
      "learning_rate": 0.0005,
      "loss": 2.3763933181762695,
      "step": 3848
    },
    {
      "epoch": 0.6210067763794772,
      "grad_norm": 1.3602744340896606,
      "learning_rate": 0.0005,
      "loss": 2.6052541732788086,
      "step": 3849
    },
    {
      "epoch": 0.6211681187479832,
      "grad_norm": 1.463915228843689,
      "learning_rate": 0.0005,
      "loss": 3.402642250061035,
      "step": 3850
    },
    {
      "epoch": 0.6213294611164892,
      "grad_norm": 1.3393937349319458,
      "learning_rate": 0.0005,
      "loss": 2.8171520233154297,
      "step": 3851
    },
    {
      "epoch": 0.6214908034849952,
      "grad_norm": 1.0083221197128296,
      "learning_rate": 0.0005,
      "loss": 2.85867977142334,
      "step": 3852
    },
    {
      "epoch": 0.6216521458535011,
      "grad_norm": 1.0439680814743042,
      "learning_rate": 0.0005,
      "loss": 2.6517367362976074,
      "step": 3853
    },
    {
      "epoch": 0.6218134882220071,
      "grad_norm": 1.6286085844039917,
      "learning_rate": 0.0005,
      "loss": 2.308872699737549,
      "step": 3854
    },
    {
      "epoch": 0.6219748305905131,
      "grad_norm": 1.1205604076385498,
      "learning_rate": 0.0005,
      "loss": 2.821061849594116,
      "step": 3855
    },
    {
      "epoch": 0.622136172959019,
      "grad_norm": 0.9481524229049683,
      "learning_rate": 0.0005,
      "loss": 3.0117406845092773,
      "step": 3856
    },
    {
      "epoch": 0.622297515327525,
      "grad_norm": 1.3124265670776367,
      "learning_rate": 0.0005,
      "loss": 3.293337821960449,
      "step": 3857
    },
    {
      "epoch": 0.6224588576960309,
      "grad_norm": 1.079146146774292,
      "learning_rate": 0.0005,
      "loss": 2.702435255050659,
      "step": 3858
    },
    {
      "epoch": 0.622620200064537,
      "grad_norm": 1.1035889387130737,
      "learning_rate": 0.0005,
      "loss": 2.8404269218444824,
      "step": 3859
    },
    {
      "epoch": 0.622781542433043,
      "grad_norm": 1.6804031133651733,
      "learning_rate": 0.0005,
      "loss": 2.6705310344696045,
      "step": 3860
    },
    {
      "epoch": 0.6229428848015489,
      "grad_norm": 1.0706393718719482,
      "learning_rate": 0.0005,
      "loss": 2.6436808109283447,
      "step": 3861
    },
    {
      "epoch": 0.6231042271700549,
      "grad_norm": 1.062954068183899,
      "learning_rate": 0.0005,
      "loss": 2.357104778289795,
      "step": 3862
    },
    {
      "epoch": 0.6232655695385608,
      "grad_norm": 1.2119646072387695,
      "learning_rate": 0.0005,
      "loss": 2.9820618629455566,
      "step": 3863
    },
    {
      "epoch": 0.6234269119070668,
      "grad_norm": 0.9186913967132568,
      "learning_rate": 0.0005,
      "loss": 2.530503511428833,
      "step": 3864
    },
    {
      "epoch": 0.6235882542755727,
      "grad_norm": 5.418235778808594,
      "learning_rate": 0.0005,
      "loss": 3.365245819091797,
      "step": 3865
    },
    {
      "epoch": 0.6237495966440787,
      "grad_norm": 1.2893511056900024,
      "learning_rate": 0.0005,
      "loss": 2.5171616077423096,
      "step": 3866
    },
    {
      "epoch": 0.6239109390125847,
      "grad_norm": 1.7452600002288818,
      "learning_rate": 0.0005,
      "loss": 3.063948631286621,
      "step": 3867
    },
    {
      "epoch": 0.6240722813810907,
      "grad_norm": 1.174473762512207,
      "learning_rate": 0.0005,
      "loss": 2.824061393737793,
      "step": 3868
    },
    {
      "epoch": 0.6242336237495967,
      "grad_norm": 1.3066219091415405,
      "learning_rate": 0.0005,
      "loss": 2.886538505554199,
      "step": 3869
    },
    {
      "epoch": 0.6243949661181026,
      "grad_norm": 2.212090015411377,
      "learning_rate": 0.0005,
      "loss": 2.6390318870544434,
      "step": 3870
    },
    {
      "epoch": 0.6245563084866086,
      "grad_norm": 1.1967424154281616,
      "learning_rate": 0.0005,
      "loss": 2.5357024669647217,
      "step": 3871
    },
    {
      "epoch": 0.6247176508551145,
      "grad_norm": 0.9500588774681091,
      "learning_rate": 0.0005,
      "loss": 2.751654863357544,
      "step": 3872
    },
    {
      "epoch": 0.6248789932236205,
      "grad_norm": 1.2494348287582397,
      "learning_rate": 0.0005,
      "loss": 2.6537742614746094,
      "step": 3873
    },
    {
      "epoch": 0.6250403355921265,
      "grad_norm": 1.4831163883209229,
      "learning_rate": 0.0005,
      "loss": 2.6129257678985596,
      "step": 3874
    },
    {
      "epoch": 0.6252016779606324,
      "grad_norm": 1.6028600931167603,
      "learning_rate": 0.0005,
      "loss": 3.0022404193878174,
      "step": 3875
    },
    {
      "epoch": 0.6253630203291385,
      "grad_norm": 1.2832163572311401,
      "learning_rate": 0.0005,
      "loss": 2.6958768367767334,
      "step": 3876
    },
    {
      "epoch": 0.6255243626976444,
      "grad_norm": 1.051782488822937,
      "learning_rate": 0.0005,
      "loss": 2.333888530731201,
      "step": 3877
    },
    {
      "epoch": 0.6256857050661504,
      "grad_norm": 1.251576542854309,
      "learning_rate": 0.0005,
      "loss": 2.9016427993774414,
      "step": 3878
    },
    {
      "epoch": 0.6258470474346564,
      "grad_norm": 1.1258212327957153,
      "learning_rate": 0.0005,
      "loss": 2.4322891235351562,
      "step": 3879
    },
    {
      "epoch": 0.6260083898031623,
      "grad_norm": 1.461465835571289,
      "learning_rate": 0.0005,
      "loss": 2.780254364013672,
      "step": 3880
    },
    {
      "epoch": 0.6261697321716683,
      "grad_norm": 1.2065318822860718,
      "learning_rate": 0.0005,
      "loss": 3.0408761501312256,
      "step": 3881
    },
    {
      "epoch": 0.6263310745401742,
      "grad_norm": 1.0922638177871704,
      "learning_rate": 0.0005,
      "loss": 2.8309574127197266,
      "step": 3882
    },
    {
      "epoch": 0.6264924169086802,
      "grad_norm": 1.6545120477676392,
      "learning_rate": 0.0005,
      "loss": 2.803237199783325,
      "step": 3883
    },
    {
      "epoch": 0.6266537592771861,
      "grad_norm": 0.9342891573905945,
      "learning_rate": 0.0005,
      "loss": 2.7787513732910156,
      "step": 3884
    },
    {
      "epoch": 0.6268151016456922,
      "grad_norm": 1.2217743396759033,
      "learning_rate": 0.0005,
      "loss": 2.687385082244873,
      "step": 3885
    },
    {
      "epoch": 0.6269764440141982,
      "grad_norm": 1.3454484939575195,
      "learning_rate": 0.0005,
      "loss": 2.5512735843658447,
      "step": 3886
    },
    {
      "epoch": 0.6271377863827041,
      "grad_norm": 1.114524245262146,
      "learning_rate": 0.0005,
      "loss": 2.7252888679504395,
      "step": 3887
    },
    {
      "epoch": 0.6272991287512101,
      "grad_norm": 1.1793415546417236,
      "learning_rate": 0.0005,
      "loss": 2.499298572540283,
      "step": 3888
    },
    {
      "epoch": 0.627460471119716,
      "grad_norm": 1.162268877029419,
      "learning_rate": 0.0005,
      "loss": 2.704270362854004,
      "step": 3889
    },
    {
      "epoch": 0.627621813488222,
      "grad_norm": 1.1370259523391724,
      "learning_rate": 0.0005,
      "loss": 2.1750552654266357,
      "step": 3890
    },
    {
      "epoch": 0.627783155856728,
      "grad_norm": 1.1752561330795288,
      "learning_rate": 0.0005,
      "loss": 2.5238394737243652,
      "step": 3891
    },
    {
      "epoch": 0.6279444982252339,
      "grad_norm": 1.1171894073486328,
      "learning_rate": 0.0005,
      "loss": 2.865957736968994,
      "step": 3892
    },
    {
      "epoch": 0.6281058405937399,
      "grad_norm": 1.5450875759124756,
      "learning_rate": 0.0005,
      "loss": 2.8156466484069824,
      "step": 3893
    },
    {
      "epoch": 0.6282671829622459,
      "grad_norm": 1.330093502998352,
      "learning_rate": 0.0005,
      "loss": 2.659228801727295,
      "step": 3894
    },
    {
      "epoch": 0.6284285253307519,
      "grad_norm": 1.4905259609222412,
      "learning_rate": 0.0005,
      "loss": 2.6968278884887695,
      "step": 3895
    },
    {
      "epoch": 0.6285898676992578,
      "grad_norm": 1.2111082077026367,
      "learning_rate": 0.0005,
      "loss": 2.674807548522949,
      "step": 3896
    },
    {
      "epoch": 0.6287512100677638,
      "grad_norm": 1.6184520721435547,
      "learning_rate": 0.0005,
      "loss": 2.1261556148529053,
      "step": 3897
    },
    {
      "epoch": 0.6289125524362698,
      "grad_norm": 1.5356146097183228,
      "learning_rate": 0.0005,
      "loss": 2.759535074234009,
      "step": 3898
    },
    {
      "epoch": 0.6290738948047757,
      "grad_norm": 1.1981028318405151,
      "learning_rate": 0.0005,
      "loss": 2.6219446659088135,
      "step": 3899
    },
    {
      "epoch": 0.6292352371732817,
      "grad_norm": 1.6060587167739868,
      "learning_rate": 0.0005,
      "loss": 2.550849199295044,
      "step": 3900
    },
    {
      "epoch": 0.6293965795417876,
      "grad_norm": 4.573112964630127,
      "learning_rate": 0.0005,
      "loss": 2.716909885406494,
      "step": 3901
    },
    {
      "epoch": 0.6295579219102937,
      "grad_norm": 1.169875979423523,
      "learning_rate": 0.0005,
      "loss": 2.495192050933838,
      "step": 3902
    },
    {
      "epoch": 0.6297192642787997,
      "grad_norm": 1.1532865762710571,
      "learning_rate": 0.0005,
      "loss": 2.879234790802002,
      "step": 3903
    },
    {
      "epoch": 0.6298806066473056,
      "grad_norm": 1.2678385972976685,
      "learning_rate": 0.0005,
      "loss": 2.8916831016540527,
      "step": 3904
    },
    {
      "epoch": 0.6300419490158116,
      "grad_norm": 1.7773271799087524,
      "learning_rate": 0.0005,
      "loss": 2.5500497817993164,
      "step": 3905
    },
    {
      "epoch": 0.6302032913843175,
      "grad_norm": 1.5719558000564575,
      "learning_rate": 0.0005,
      "loss": 2.829540729522705,
      "step": 3906
    },
    {
      "epoch": 0.6303646337528235,
      "grad_norm": 1.2540216445922852,
      "learning_rate": 0.0005,
      "loss": 2.476156234741211,
      "step": 3907
    },
    {
      "epoch": 0.6305259761213294,
      "grad_norm": 1.0922198295593262,
      "learning_rate": 0.0005,
      "loss": 2.524310350418091,
      "step": 3908
    },
    {
      "epoch": 0.6306873184898354,
      "grad_norm": 0.9425232410430908,
      "learning_rate": 0.0005,
      "loss": 2.423518419265747,
      "step": 3909
    },
    {
      "epoch": 0.6308486608583413,
      "grad_norm": 1.3054969310760498,
      "learning_rate": 0.0005,
      "loss": 2.8328258991241455,
      "step": 3910
    },
    {
      "epoch": 0.6310100032268474,
      "grad_norm": 1.0810949802398682,
      "learning_rate": 0.0005,
      "loss": 3.189363479614258,
      "step": 3911
    },
    {
      "epoch": 0.6311713455953534,
      "grad_norm": 1.2733210325241089,
      "learning_rate": 0.0005,
      "loss": 3.0727405548095703,
      "step": 3912
    },
    {
      "epoch": 0.6313326879638593,
      "grad_norm": 1.0604532957077026,
      "learning_rate": 0.0005,
      "loss": 2.732649803161621,
      "step": 3913
    },
    {
      "epoch": 0.6314940303323653,
      "grad_norm": 1.2571182250976562,
      "learning_rate": 0.0005,
      "loss": 2.607856035232544,
      "step": 3914
    },
    {
      "epoch": 0.6316553727008712,
      "grad_norm": 1.0612518787384033,
      "learning_rate": 0.0005,
      "loss": 2.958252429962158,
      "step": 3915
    },
    {
      "epoch": 0.6318167150693772,
      "grad_norm": 1.5787346363067627,
      "learning_rate": 0.0005,
      "loss": 2.8779735565185547,
      "step": 3916
    },
    {
      "epoch": 0.6319780574378832,
      "grad_norm": 1.0332764387130737,
      "learning_rate": 0.0005,
      "loss": 2.672215461730957,
      "step": 3917
    },
    {
      "epoch": 0.6321393998063891,
      "grad_norm": 1.102351188659668,
      "learning_rate": 0.0005,
      "loss": 3.1327242851257324,
      "step": 3918
    },
    {
      "epoch": 0.6323007421748952,
      "grad_norm": 1.2621744871139526,
      "learning_rate": 0.0005,
      "loss": 2.1763370037078857,
      "step": 3919
    },
    {
      "epoch": 0.6324620845434011,
      "grad_norm": 1.2977044582366943,
      "learning_rate": 0.0005,
      "loss": 2.8058860301971436,
      "step": 3920
    },
    {
      "epoch": 0.6326234269119071,
      "grad_norm": 1.3424570560455322,
      "learning_rate": 0.0005,
      "loss": 2.6430857181549072,
      "step": 3921
    },
    {
      "epoch": 0.632784769280413,
      "grad_norm": 1.2318108081817627,
      "learning_rate": 0.0005,
      "loss": 3.1858136653900146,
      "step": 3922
    },
    {
      "epoch": 0.632946111648919,
      "grad_norm": 1.2484320402145386,
      "learning_rate": 0.0005,
      "loss": 2.527602434158325,
      "step": 3923
    },
    {
      "epoch": 0.633107454017425,
      "grad_norm": 1.1273140907287598,
      "learning_rate": 0.0005,
      "loss": 2.824270725250244,
      "step": 3924
    },
    {
      "epoch": 0.6332687963859309,
      "grad_norm": 1.4412940740585327,
      "learning_rate": 0.0005,
      "loss": 2.913872003555298,
      "step": 3925
    },
    {
      "epoch": 0.6334301387544369,
      "grad_norm": 1.640679955482483,
      "learning_rate": 0.0005,
      "loss": 2.6698098182678223,
      "step": 3926
    },
    {
      "epoch": 0.6335914811229428,
      "grad_norm": 1.022852897644043,
      "learning_rate": 0.0005,
      "loss": 2.9960508346557617,
      "step": 3927
    },
    {
      "epoch": 0.6337528234914489,
      "grad_norm": 1.0754011869430542,
      "learning_rate": 0.0005,
      "loss": 2.6728124618530273,
      "step": 3928
    },
    {
      "epoch": 0.6339141658599549,
      "grad_norm": 1.4841444492340088,
      "learning_rate": 0.0005,
      "loss": 2.7699058055877686,
      "step": 3929
    },
    {
      "epoch": 0.6340755082284608,
      "grad_norm": 1.3382002115249634,
      "learning_rate": 0.0005,
      "loss": 2.5219550132751465,
      "step": 3930
    },
    {
      "epoch": 0.6342368505969668,
      "grad_norm": 1.1762542724609375,
      "learning_rate": 0.0005,
      "loss": 1.8874003887176514,
      "step": 3931
    },
    {
      "epoch": 0.6343981929654727,
      "grad_norm": 1.1953449249267578,
      "learning_rate": 0.0005,
      "loss": 2.4762024879455566,
      "step": 3932
    },
    {
      "epoch": 0.6345595353339787,
      "grad_norm": 1.2811206579208374,
      "learning_rate": 0.0005,
      "loss": 2.789799690246582,
      "step": 3933
    },
    {
      "epoch": 0.6347208777024846,
      "grad_norm": 1.1472558975219727,
      "learning_rate": 0.0005,
      "loss": 3.0266098976135254,
      "step": 3934
    },
    {
      "epoch": 0.6348822200709906,
      "grad_norm": 1.2934088706970215,
      "learning_rate": 0.0005,
      "loss": 2.817931652069092,
      "step": 3935
    },
    {
      "epoch": 0.6350435624394967,
      "grad_norm": 1.1104950904846191,
      "learning_rate": 0.0005,
      "loss": 2.9918482303619385,
      "step": 3936
    },
    {
      "epoch": 0.6352049048080026,
      "grad_norm": 1.1680158376693726,
      "learning_rate": 0.0005,
      "loss": 2.828479290008545,
      "step": 3937
    },
    {
      "epoch": 0.6353662471765086,
      "grad_norm": 1.267879843711853,
      "learning_rate": 0.0005,
      "loss": 2.7970104217529297,
      "step": 3938
    },
    {
      "epoch": 0.6355275895450145,
      "grad_norm": 1.7266241312026978,
      "learning_rate": 0.0005,
      "loss": 2.2174439430236816,
      "step": 3939
    },
    {
      "epoch": 0.6356889319135205,
      "grad_norm": 1.3335554599761963,
      "learning_rate": 0.0005,
      "loss": 2.287229537963867,
      "step": 3940
    },
    {
      "epoch": 0.6358502742820265,
      "grad_norm": 1.2960786819458008,
      "learning_rate": 0.0005,
      "loss": 2.571800708770752,
      "step": 3941
    },
    {
      "epoch": 0.6360116166505324,
      "grad_norm": 1.4504300355911255,
      "learning_rate": 0.0005,
      "loss": 2.5457329750061035,
      "step": 3942
    },
    {
      "epoch": 0.6361729590190384,
      "grad_norm": 1.016492486000061,
      "learning_rate": 0.0005,
      "loss": 2.628045082092285,
      "step": 3943
    },
    {
      "epoch": 0.6363343013875443,
      "grad_norm": 1.3712066411972046,
      "learning_rate": 0.0005,
      "loss": 2.4631381034851074,
      "step": 3944
    },
    {
      "epoch": 0.6364956437560504,
      "grad_norm": 1.8321664333343506,
      "learning_rate": 0.0005,
      "loss": 3.0130605697631836,
      "step": 3945
    },
    {
      "epoch": 0.6366569861245563,
      "grad_norm": 1.1695950031280518,
      "learning_rate": 0.0005,
      "loss": 2.6207051277160645,
      "step": 3946
    },
    {
      "epoch": 0.6368183284930623,
      "grad_norm": 1.1561042070388794,
      "learning_rate": 0.0005,
      "loss": 2.749525547027588,
      "step": 3947
    },
    {
      "epoch": 0.6369796708615683,
      "grad_norm": 1.1739581823349,
      "learning_rate": 0.0005,
      "loss": 3.171739101409912,
      "step": 3948
    },
    {
      "epoch": 0.6371410132300742,
      "grad_norm": 1.0217173099517822,
      "learning_rate": 0.0005,
      "loss": 3.0495777130126953,
      "step": 3949
    },
    {
      "epoch": 0.6373023555985802,
      "grad_norm": 1.0483496189117432,
      "learning_rate": 0.0005,
      "loss": 2.596416473388672,
      "step": 3950
    },
    {
      "epoch": 0.6374636979670861,
      "grad_norm": 1.4834349155426025,
      "learning_rate": 0.0005,
      "loss": 2.683500289916992,
      "step": 3951
    },
    {
      "epoch": 0.6376250403355921,
      "grad_norm": 1.5140570402145386,
      "learning_rate": 0.0005,
      "loss": 2.778502941131592,
      "step": 3952
    },
    {
      "epoch": 0.637786382704098,
      "grad_norm": 1.0465421676635742,
      "learning_rate": 0.0005,
      "loss": 2.8467109203338623,
      "step": 3953
    },
    {
      "epoch": 0.6379477250726041,
      "grad_norm": 1.082513451576233,
      "learning_rate": 0.0005,
      "loss": 2.7952828407287598,
      "step": 3954
    },
    {
      "epoch": 0.6381090674411101,
      "grad_norm": 1.14582359790802,
      "learning_rate": 0.0005,
      "loss": 2.7832093238830566,
      "step": 3955
    },
    {
      "epoch": 0.638270409809616,
      "grad_norm": 0.9949118494987488,
      "learning_rate": 0.0005,
      "loss": 2.4828944206237793,
      "step": 3956
    },
    {
      "epoch": 0.638431752178122,
      "grad_norm": 0.9822730422019958,
      "learning_rate": 0.0005,
      "loss": 2.3080153465270996,
      "step": 3957
    },
    {
      "epoch": 0.6385930945466279,
      "grad_norm": 1.4305227994918823,
      "learning_rate": 0.0005,
      "loss": 2.504964590072632,
      "step": 3958
    },
    {
      "epoch": 0.6387544369151339,
      "grad_norm": 1.1987148523330688,
      "learning_rate": 0.0005,
      "loss": 2.6018800735473633,
      "step": 3959
    },
    {
      "epoch": 0.6389157792836399,
      "grad_norm": 28.82143211364746,
      "learning_rate": 0.0005,
      "loss": 2.671370506286621,
      "step": 3960
    },
    {
      "epoch": 0.6390771216521458,
      "grad_norm": 1.2709940671920776,
      "learning_rate": 0.0005,
      "loss": 2.632673740386963,
      "step": 3961
    },
    {
      "epoch": 0.6392384640206519,
      "grad_norm": 1.43191659450531,
      "learning_rate": 0.0005,
      "loss": 2.885838031768799,
      "step": 3962
    },
    {
      "epoch": 0.6393998063891578,
      "grad_norm": 1.6080474853515625,
      "learning_rate": 0.0005,
      "loss": 2.664151668548584,
      "step": 3963
    },
    {
      "epoch": 0.6395611487576638,
      "grad_norm": 1.4959437847137451,
      "learning_rate": 0.0005,
      "loss": 2.746291160583496,
      "step": 3964
    },
    {
      "epoch": 0.6397224911261697,
      "grad_norm": 1.0762057304382324,
      "learning_rate": 0.0005,
      "loss": 2.4845504760742188,
      "step": 3965
    },
    {
      "epoch": 0.6398838334946757,
      "grad_norm": 1.0586918592453003,
      "learning_rate": 0.0005,
      "loss": 2.9036402702331543,
      "step": 3966
    },
    {
      "epoch": 0.6400451758631817,
      "grad_norm": 1.445991039276123,
      "learning_rate": 0.0005,
      "loss": 3.1223976612091064,
      "step": 3967
    },
    {
      "epoch": 0.6402065182316876,
      "grad_norm": 1.5690686702728271,
      "learning_rate": 0.0005,
      "loss": 2.7851431369781494,
      "step": 3968
    },
    {
      "epoch": 0.6403678606001936,
      "grad_norm": 1.1712863445281982,
      "learning_rate": 0.0005,
      "loss": 2.3192079067230225,
      "step": 3969
    },
    {
      "epoch": 0.6405292029686995,
      "grad_norm": 1.1339051723480225,
      "learning_rate": 0.0005,
      "loss": 2.5027122497558594,
      "step": 3970
    },
    {
      "epoch": 0.6406905453372056,
      "grad_norm": 1.7894656658172607,
      "learning_rate": 0.0005,
      "loss": 2.9320201873779297,
      "step": 3971
    },
    {
      "epoch": 0.6408518877057116,
      "grad_norm": 1.190532922744751,
      "learning_rate": 0.0005,
      "loss": 2.8215458393096924,
      "step": 3972
    },
    {
      "epoch": 0.6410132300742175,
      "grad_norm": 0.9681673645973206,
      "learning_rate": 0.0005,
      "loss": 2.7402596473693848,
      "step": 3973
    },
    {
      "epoch": 0.6411745724427235,
      "grad_norm": 1.3985785245895386,
      "learning_rate": 0.0005,
      "loss": 2.3476929664611816,
      "step": 3974
    },
    {
      "epoch": 0.6413359148112294,
      "grad_norm": 1.1558359861373901,
      "learning_rate": 0.0005,
      "loss": 2.596970558166504,
      "step": 3975
    },
    {
      "epoch": 0.6414972571797354,
      "grad_norm": 2.1122896671295166,
      "learning_rate": 0.0005,
      "loss": 2.562033176422119,
      "step": 3976
    },
    {
      "epoch": 0.6416585995482413,
      "grad_norm": 1.1201022863388062,
      "learning_rate": 0.0005,
      "loss": 2.746572494506836,
      "step": 3977
    },
    {
      "epoch": 0.6418199419167473,
      "grad_norm": 1.0109360218048096,
      "learning_rate": 0.0005,
      "loss": 2.7128376960754395,
      "step": 3978
    },
    {
      "epoch": 0.6419812842852534,
      "grad_norm": 1.182787537574768,
      "learning_rate": 0.0005,
      "loss": 2.500074863433838,
      "step": 3979
    },
    {
      "epoch": 0.6421426266537593,
      "grad_norm": 1.753149390220642,
      "learning_rate": 0.0005,
      "loss": 2.1988649368286133,
      "step": 3980
    },
    {
      "epoch": 0.6423039690222653,
      "grad_norm": 1.466903805732727,
      "learning_rate": 0.0005,
      "loss": 2.716285228729248,
      "step": 3981
    },
    {
      "epoch": 0.6424653113907712,
      "grad_norm": 1.2210525274276733,
      "learning_rate": 0.0005,
      "loss": 3.1663296222686768,
      "step": 3982
    },
    {
      "epoch": 0.6426266537592772,
      "grad_norm": 1.056472897529602,
      "learning_rate": 0.0005,
      "loss": 2.6921334266662598,
      "step": 3983
    },
    {
      "epoch": 0.6427879961277831,
      "grad_norm": 1.2117433547973633,
      "learning_rate": 0.0005,
      "loss": 2.665004253387451,
      "step": 3984
    },
    {
      "epoch": 0.6429493384962891,
      "grad_norm": 1.283811092376709,
      "learning_rate": 0.0005,
      "loss": 2.144214630126953,
      "step": 3985
    },
    {
      "epoch": 0.6431106808647951,
      "grad_norm": 1.2736815214157104,
      "learning_rate": 0.0005,
      "loss": 2.681967258453369,
      "step": 3986
    },
    {
      "epoch": 0.643272023233301,
      "grad_norm": 1.0119361877441406,
      "learning_rate": 0.0005,
      "loss": 2.478480100631714,
      "step": 3987
    },
    {
      "epoch": 0.6434333656018071,
      "grad_norm": 1.1290334463119507,
      "learning_rate": 0.0005,
      "loss": 3.0271573066711426,
      "step": 3988
    },
    {
      "epoch": 0.643594707970313,
      "grad_norm": 0.9827144145965576,
      "learning_rate": 0.0005,
      "loss": 2.6883678436279297,
      "step": 3989
    },
    {
      "epoch": 0.643756050338819,
      "grad_norm": 1.1590384244918823,
      "learning_rate": 0.0005,
      "loss": 2.57068133354187,
      "step": 3990
    },
    {
      "epoch": 0.643917392707325,
      "grad_norm": 1.0254535675048828,
      "learning_rate": 0.0005,
      "loss": 2.7082574367523193,
      "step": 3991
    },
    {
      "epoch": 0.6440787350758309,
      "grad_norm": 1.1424421072006226,
      "learning_rate": 0.0005,
      "loss": 2.773716926574707,
      "step": 3992
    },
    {
      "epoch": 0.6442400774443369,
      "grad_norm": 0.9838871955871582,
      "learning_rate": 0.0005,
      "loss": 2.7942214012145996,
      "step": 3993
    },
    {
      "epoch": 0.6444014198128428,
      "grad_norm": 1.2322096824645996,
      "learning_rate": 0.0005,
      "loss": 2.402456283569336,
      "step": 3994
    },
    {
      "epoch": 0.6445627621813488,
      "grad_norm": 0.9052395224571228,
      "learning_rate": 0.0005,
      "loss": 2.8404479026794434,
      "step": 3995
    },
    {
      "epoch": 0.6447241045498547,
      "grad_norm": 1.3188449144363403,
      "learning_rate": 0.0005,
      "loss": 3.1852893829345703,
      "step": 3996
    },
    {
      "epoch": 0.6448854469183608,
      "grad_norm": 1.9830257892608643,
      "learning_rate": 0.0005,
      "loss": 2.686485528945923,
      "step": 3997
    },
    {
      "epoch": 0.6450467892868668,
      "grad_norm": 1.3169153928756714,
      "learning_rate": 0.0005,
      "loss": 2.58115816116333,
      "step": 3998
    },
    {
      "epoch": 0.6452081316553727,
      "grad_norm": 1.3889901638031006,
      "learning_rate": 0.0005,
      "loss": 2.5069847106933594,
      "step": 3999
    },
    {
      "epoch": 0.6453694740238787,
      "grad_norm": 1.5369857549667358,
      "learning_rate": 0.0005,
      "loss": 2.7353391647338867,
      "step": 4000
    },
    {
      "epoch": 0.6455308163923846,
      "grad_norm": 1.3165791034698486,
      "learning_rate": 0.0005,
      "loss": 2.952918291091919,
      "step": 4001
    },
    {
      "epoch": 0.6456921587608906,
      "grad_norm": 1.2876062393188477,
      "learning_rate": 0.0005,
      "loss": 2.765719413757324,
      "step": 4002
    },
    {
      "epoch": 0.6458535011293965,
      "grad_norm": 1.3897653818130493,
      "learning_rate": 0.0005,
      "loss": 2.8265132904052734,
      "step": 4003
    },
    {
      "epoch": 0.6460148434979025,
      "grad_norm": 1.0709689855575562,
      "learning_rate": 0.0005,
      "loss": 3.455106735229492,
      "step": 4004
    },
    {
      "epoch": 0.6461761858664086,
      "grad_norm": 1.0703144073486328,
      "learning_rate": 0.0005,
      "loss": 2.6437833309173584,
      "step": 4005
    },
    {
      "epoch": 0.6463375282349145,
      "grad_norm": 0.936835527420044,
      "learning_rate": 0.0005,
      "loss": 2.658818244934082,
      "step": 4006
    },
    {
      "epoch": 0.6464988706034205,
      "grad_norm": 1.413028597831726,
      "learning_rate": 0.0005,
      "loss": 2.459768772125244,
      "step": 4007
    },
    {
      "epoch": 0.6466602129719264,
      "grad_norm": 1.0520323514938354,
      "learning_rate": 0.0005,
      "loss": 2.1293716430664062,
      "step": 4008
    },
    {
      "epoch": 0.6468215553404324,
      "grad_norm": 1.0828431844711304,
      "learning_rate": 0.0005,
      "loss": 2.5516602993011475,
      "step": 4009
    },
    {
      "epoch": 0.6469828977089384,
      "grad_norm": 1.2546091079711914,
      "learning_rate": 0.0005,
      "loss": 2.7193589210510254,
      "step": 4010
    },
    {
      "epoch": 0.6471442400774443,
      "grad_norm": 1.0844112634658813,
      "learning_rate": 0.0005,
      "loss": 2.2309188842773438,
      "step": 4011
    },
    {
      "epoch": 0.6473055824459503,
      "grad_norm": 0.9180402755737305,
      "learning_rate": 0.0005,
      "loss": 2.9197278022766113,
      "step": 4012
    },
    {
      "epoch": 0.6474669248144562,
      "grad_norm": 1.4600541591644287,
      "learning_rate": 0.0005,
      "loss": 2.756556510925293,
      "step": 4013
    },
    {
      "epoch": 0.6476282671829623,
      "grad_norm": 1.2569063901901245,
      "learning_rate": 0.0005,
      "loss": 2.5663952827453613,
      "step": 4014
    },
    {
      "epoch": 0.6477896095514682,
      "grad_norm": 1.4500261545181274,
      "learning_rate": 0.0005,
      "loss": 2.675997257232666,
      "step": 4015
    },
    {
      "epoch": 0.6479509519199742,
      "grad_norm": 1.2282744646072388,
      "learning_rate": 0.0005,
      "loss": 3.153134346008301,
      "step": 4016
    },
    {
      "epoch": 0.6481122942884802,
      "grad_norm": 1.0148582458496094,
      "learning_rate": 0.0005,
      "loss": 2.8679137229919434,
      "step": 4017
    },
    {
      "epoch": 0.6482736366569861,
      "grad_norm": 1.33318030834198,
      "learning_rate": 0.0005,
      "loss": 2.7551229000091553,
      "step": 4018
    },
    {
      "epoch": 0.6484349790254921,
      "grad_norm": 1.1527005434036255,
      "learning_rate": 0.0005,
      "loss": 2.4548544883728027,
      "step": 4019
    },
    {
      "epoch": 0.648596321393998,
      "grad_norm": 1.0778297185897827,
      "learning_rate": 0.0005,
      "loss": 2.3668601512908936,
      "step": 4020
    },
    {
      "epoch": 0.648757663762504,
      "grad_norm": 1.1949398517608643,
      "learning_rate": 0.0005,
      "loss": 2.9703989028930664,
      "step": 4021
    },
    {
      "epoch": 0.6489190061310101,
      "grad_norm": 1.4525411128997803,
      "learning_rate": 0.0005,
      "loss": 2.620344638824463,
      "step": 4022
    },
    {
      "epoch": 0.649080348499516,
      "grad_norm": 1.3396592140197754,
      "learning_rate": 0.0005,
      "loss": 2.711282730102539,
      "step": 4023
    },
    {
      "epoch": 0.649241690868022,
      "grad_norm": 1.3681833744049072,
      "learning_rate": 0.0005,
      "loss": 2.2543153762817383,
      "step": 4024
    },
    {
      "epoch": 0.6494030332365279,
      "grad_norm": 1.3254197835922241,
      "learning_rate": 0.0005,
      "loss": 2.6818933486938477,
      "step": 4025
    },
    {
      "epoch": 0.6495643756050339,
      "grad_norm": 1.0440584421157837,
      "learning_rate": 0.0005,
      "loss": 2.567671298980713,
      "step": 4026
    },
    {
      "epoch": 0.6497257179735398,
      "grad_norm": 1.2524795532226562,
      "learning_rate": 0.0005,
      "loss": 2.4269657135009766,
      "step": 4027
    },
    {
      "epoch": 0.6498870603420458,
      "grad_norm": 1.105833649635315,
      "learning_rate": 0.0005,
      "loss": 2.3469529151916504,
      "step": 4028
    },
    {
      "epoch": 0.6500484027105518,
      "grad_norm": 2.948122978210449,
      "learning_rate": 0.0005,
      "loss": 2.568077564239502,
      "step": 4029
    },
    {
      "epoch": 0.6502097450790577,
      "grad_norm": 1.59101140499115,
      "learning_rate": 0.0005,
      "loss": 2.759228229522705,
      "step": 4030
    },
    {
      "epoch": 0.6503710874475638,
      "grad_norm": 1.3102871179580688,
      "learning_rate": 0.0005,
      "loss": 2.5433812141418457,
      "step": 4031
    },
    {
      "epoch": 0.6505324298160697,
      "grad_norm": 1.5338361263275146,
      "learning_rate": 0.0005,
      "loss": 2.64259672164917,
      "step": 4032
    },
    {
      "epoch": 0.6506937721845757,
      "grad_norm": 1.2289390563964844,
      "learning_rate": 0.0005,
      "loss": 2.606936454772949,
      "step": 4033
    },
    {
      "epoch": 0.6508551145530816,
      "grad_norm": 1.2604992389678955,
      "learning_rate": 0.0005,
      "loss": 2.6116456985473633,
      "step": 4034
    },
    {
      "epoch": 0.6510164569215876,
      "grad_norm": 1.1849029064178467,
      "learning_rate": 0.0005,
      "loss": 2.402894973754883,
      "step": 4035
    },
    {
      "epoch": 0.6511777992900936,
      "grad_norm": 0.9881041049957275,
      "learning_rate": 0.0005,
      "loss": 2.60123610496521,
      "step": 4036
    },
    {
      "epoch": 0.6513391416585995,
      "grad_norm": 1.646716594696045,
      "learning_rate": 0.0005,
      "loss": 3.1269044876098633,
      "step": 4037
    },
    {
      "epoch": 0.6515004840271055,
      "grad_norm": 1.2920948266983032,
      "learning_rate": 0.0005,
      "loss": 2.9273757934570312,
      "step": 4038
    },
    {
      "epoch": 0.6516618263956114,
      "grad_norm": 1.2174230813980103,
      "learning_rate": 0.0005,
      "loss": 2.1489458084106445,
      "step": 4039
    },
    {
      "epoch": 0.6518231687641175,
      "grad_norm": 1.0420366525650024,
      "learning_rate": 0.0005,
      "loss": 2.3450193405151367,
      "step": 4040
    },
    {
      "epoch": 0.6519845111326235,
      "grad_norm": 1.1356799602508545,
      "learning_rate": 0.0005,
      "loss": 2.4929308891296387,
      "step": 4041
    },
    {
      "epoch": 0.6521458535011294,
      "grad_norm": 13.680785179138184,
      "learning_rate": 0.0005,
      "loss": 2.7961347103118896,
      "step": 4042
    },
    {
      "epoch": 0.6523071958696354,
      "grad_norm": 1.0188080072402954,
      "learning_rate": 0.0005,
      "loss": 2.394313097000122,
      "step": 4043
    },
    {
      "epoch": 0.6524685382381413,
      "grad_norm": 1.092549443244934,
      "learning_rate": 0.0005,
      "loss": 2.1617753505706787,
      "step": 4044
    },
    {
      "epoch": 0.6526298806066473,
      "grad_norm": 1.3707082271575928,
      "learning_rate": 0.0005,
      "loss": 2.9656710624694824,
      "step": 4045
    },
    {
      "epoch": 0.6527912229751532,
      "grad_norm": 2.328981876373291,
      "learning_rate": 0.0005,
      "loss": 2.5729353427886963,
      "step": 4046
    },
    {
      "epoch": 0.6529525653436592,
      "grad_norm": 1.0551124811172485,
      "learning_rate": 0.0005,
      "loss": 2.2605767250061035,
      "step": 4047
    },
    {
      "epoch": 0.6531139077121653,
      "grad_norm": 1.564972996711731,
      "learning_rate": 0.0005,
      "loss": 2.665151834487915,
      "step": 4048
    },
    {
      "epoch": 0.6532752500806712,
      "grad_norm": 1.1826281547546387,
      "learning_rate": 0.0005,
      "loss": 3.057119846343994,
      "step": 4049
    },
    {
      "epoch": 0.6534365924491772,
      "grad_norm": 1.4423277378082275,
      "learning_rate": 0.0005,
      "loss": 2.897063732147217,
      "step": 4050
    },
    {
      "epoch": 0.6535979348176831,
      "grad_norm": 1.1635490655899048,
      "learning_rate": 0.0005,
      "loss": 2.6034209728240967,
      "step": 4051
    },
    {
      "epoch": 0.6537592771861891,
      "grad_norm": 1.1555970907211304,
      "learning_rate": 0.0005,
      "loss": 2.583627223968506,
      "step": 4052
    },
    {
      "epoch": 0.653920619554695,
      "grad_norm": 1.4142415523529053,
      "learning_rate": 0.0005,
      "loss": 2.0839672088623047,
      "step": 4053
    },
    {
      "epoch": 0.654081961923201,
      "grad_norm": 1.3479437828063965,
      "learning_rate": 0.0005,
      "loss": 2.7788567543029785,
      "step": 4054
    },
    {
      "epoch": 0.654243304291707,
      "grad_norm": 1.3626657724380493,
      "learning_rate": 0.0005,
      "loss": 2.894909381866455,
      "step": 4055
    },
    {
      "epoch": 0.6544046466602129,
      "grad_norm": 1.1723215579986572,
      "learning_rate": 0.0005,
      "loss": 2.8165292739868164,
      "step": 4056
    },
    {
      "epoch": 0.654565989028719,
      "grad_norm": 1.2477672100067139,
      "learning_rate": 0.0005,
      "loss": 3.0362355709075928,
      "step": 4057
    },
    {
      "epoch": 0.6547273313972249,
      "grad_norm": 1.281033992767334,
      "learning_rate": 0.0005,
      "loss": 2.7306408882141113,
      "step": 4058
    },
    {
      "epoch": 0.6548886737657309,
      "grad_norm": 1.2631484270095825,
      "learning_rate": 0.0005,
      "loss": 2.548732280731201,
      "step": 4059
    },
    {
      "epoch": 0.6550500161342369,
      "grad_norm": 1.2007784843444824,
      "learning_rate": 0.0005,
      "loss": 2.0935988426208496,
      "step": 4060
    },
    {
      "epoch": 0.6552113585027428,
      "grad_norm": 5.490997791290283,
      "learning_rate": 0.0005,
      "loss": 2.509127140045166,
      "step": 4061
    },
    {
      "epoch": 0.6553727008712488,
      "grad_norm": 1.3333262205123901,
      "learning_rate": 0.0005,
      "loss": 2.4383232593536377,
      "step": 4062
    },
    {
      "epoch": 0.6555340432397547,
      "grad_norm": 1.0933054685592651,
      "learning_rate": 0.0005,
      "loss": 2.6442813873291016,
      "step": 4063
    },
    {
      "epoch": 0.6556953856082607,
      "grad_norm": 1.3192741870880127,
      "learning_rate": 0.0005,
      "loss": 2.8430984020233154,
      "step": 4064
    },
    {
      "epoch": 0.6558567279767668,
      "grad_norm": 0.9777157306671143,
      "learning_rate": 0.0005,
      "loss": 2.582615375518799,
      "step": 4065
    },
    {
      "epoch": 0.6560180703452727,
      "grad_norm": 1.067923665046692,
      "learning_rate": 0.0005,
      "loss": 2.7648708820343018,
      "step": 4066
    },
    {
      "epoch": 0.6561794127137787,
      "grad_norm": 1.0634955167770386,
      "learning_rate": 0.0005,
      "loss": 2.264796733856201,
      "step": 4067
    },
    {
      "epoch": 0.6563407550822846,
      "grad_norm": 1.3450416326522827,
      "learning_rate": 0.0005,
      "loss": 2.7901487350463867,
      "step": 4068
    },
    {
      "epoch": 0.6565020974507906,
      "grad_norm": 1.3371795415878296,
      "learning_rate": 0.0005,
      "loss": 2.8125834465026855,
      "step": 4069
    },
    {
      "epoch": 0.6566634398192965,
      "grad_norm": 1.2329782247543335,
      "learning_rate": 0.0005,
      "loss": 2.1337270736694336,
      "step": 4070
    },
    {
      "epoch": 0.6568247821878025,
      "grad_norm": 1.1248277425765991,
      "learning_rate": 0.0005,
      "loss": 2.969020366668701,
      "step": 4071
    },
    {
      "epoch": 0.6569861245563084,
      "grad_norm": 1.2277400493621826,
      "learning_rate": 0.0005,
      "loss": 2.878711223602295,
      "step": 4072
    },
    {
      "epoch": 0.6571474669248144,
      "grad_norm": 1.1489969491958618,
      "learning_rate": 0.0005,
      "loss": 2.512300729751587,
      "step": 4073
    },
    {
      "epoch": 0.6573088092933205,
      "grad_norm": 1.0261318683624268,
      "learning_rate": 0.0005,
      "loss": 2.5676331520080566,
      "step": 4074
    },
    {
      "epoch": 0.6574701516618264,
      "grad_norm": 1.2260767221450806,
      "learning_rate": 0.0005,
      "loss": 2.6179051399230957,
      "step": 4075
    },
    {
      "epoch": 0.6576314940303324,
      "grad_norm": 1.03656804561615,
      "learning_rate": 0.0005,
      "loss": 2.4859838485717773,
      "step": 4076
    },
    {
      "epoch": 0.6577928363988383,
      "grad_norm": 1.267841100692749,
      "learning_rate": 0.0005,
      "loss": 2.669032096862793,
      "step": 4077
    },
    {
      "epoch": 0.6579541787673443,
      "grad_norm": 0.9125861525535583,
      "learning_rate": 0.0005,
      "loss": 2.009308338165283,
      "step": 4078
    },
    {
      "epoch": 0.6581155211358503,
      "grad_norm": 1.118844747543335,
      "learning_rate": 0.0005,
      "loss": 2.6417226791381836,
      "step": 4079
    },
    {
      "epoch": 0.6582768635043562,
      "grad_norm": 1.454233169555664,
      "learning_rate": 0.0005,
      "loss": 3.125454902648926,
      "step": 4080
    },
    {
      "epoch": 0.6584382058728622,
      "grad_norm": 1.3809053897857666,
      "learning_rate": 0.0005,
      "loss": 2.5615599155426025,
      "step": 4081
    },
    {
      "epoch": 0.6585995482413682,
      "grad_norm": 1.125316858291626,
      "learning_rate": 0.0005,
      "loss": 2.425389289855957,
      "step": 4082
    },
    {
      "epoch": 0.6587608906098742,
      "grad_norm": 1.0794752836227417,
      "learning_rate": 0.0005,
      "loss": 2.4155678749084473,
      "step": 4083
    },
    {
      "epoch": 0.6589222329783802,
      "grad_norm": 1.1291965246200562,
      "learning_rate": 0.0005,
      "loss": 2.4451515674591064,
      "step": 4084
    },
    {
      "epoch": 0.6590835753468861,
      "grad_norm": 0.9840728640556335,
      "learning_rate": 0.0005,
      "loss": 2.679206371307373,
      "step": 4085
    },
    {
      "epoch": 0.6592449177153921,
      "grad_norm": 2.811037063598633,
      "learning_rate": 0.0005,
      "loss": 2.635664939880371,
      "step": 4086
    },
    {
      "epoch": 0.659406260083898,
      "grad_norm": 1.1759099960327148,
      "learning_rate": 0.0005,
      "loss": 2.759443759918213,
      "step": 4087
    },
    {
      "epoch": 0.659567602452404,
      "grad_norm": 1.7037197351455688,
      "learning_rate": 0.0005,
      "loss": 2.7998452186584473,
      "step": 4088
    },
    {
      "epoch": 0.6597289448209099,
      "grad_norm": 1.3179243803024292,
      "learning_rate": 0.0005,
      "loss": 2.561061143875122,
      "step": 4089
    },
    {
      "epoch": 0.6598902871894159,
      "grad_norm": 1.2546536922454834,
      "learning_rate": 0.0005,
      "loss": 2.774240016937256,
      "step": 4090
    },
    {
      "epoch": 0.660051629557922,
      "grad_norm": 1.1942847967147827,
      "learning_rate": 0.0005,
      "loss": 2.733828067779541,
      "step": 4091
    },
    {
      "epoch": 0.6602129719264279,
      "grad_norm": 1.0199836492538452,
      "learning_rate": 0.0005,
      "loss": 2.7860374450683594,
      "step": 4092
    },
    {
      "epoch": 0.6603743142949339,
      "grad_norm": 1.9713490009307861,
      "learning_rate": 0.0005,
      "loss": 2.880018949508667,
      "step": 4093
    },
    {
      "epoch": 0.6605356566634398,
      "grad_norm": 1.1341748237609863,
      "learning_rate": 0.0005,
      "loss": 2.9520320892333984,
      "step": 4094
    },
    {
      "epoch": 0.6606969990319458,
      "grad_norm": 1.1191500425338745,
      "learning_rate": 0.0005,
      "loss": 3.054748058319092,
      "step": 4095
    },
    {
      "epoch": 0.6608583414004517,
      "grad_norm": 1.30820894241333,
      "learning_rate": 0.0005,
      "loss": 2.659794330596924,
      "step": 4096
    },
    {
      "epoch": 0.6610196837689577,
      "grad_norm": 1.5105011463165283,
      "learning_rate": 0.0005,
      "loss": 2.710826873779297,
      "step": 4097
    },
    {
      "epoch": 0.6611810261374637,
      "grad_norm": 1.9184907674789429,
      "learning_rate": 0.0005,
      "loss": 2.9664957523345947,
      "step": 4098
    },
    {
      "epoch": 0.6613423685059696,
      "grad_norm": 2.3275105953216553,
      "learning_rate": 0.0005,
      "loss": 2.3983359336853027,
      "step": 4099
    },
    {
      "epoch": 0.6615037108744757,
      "grad_norm": 1.282881498336792,
      "learning_rate": 0.0005,
      "loss": 2.6583704948425293,
      "step": 4100
    },
    {
      "epoch": 0.6616650532429816,
      "grad_norm": 2.005760908126831,
      "learning_rate": 0.0005,
      "loss": 2.6345410346984863,
      "step": 4101
    },
    {
      "epoch": 0.6618263956114876,
      "grad_norm": 1.4662944078445435,
      "learning_rate": 0.0005,
      "loss": 2.667029619216919,
      "step": 4102
    },
    {
      "epoch": 0.6619877379799936,
      "grad_norm": 3.8895034790039062,
      "learning_rate": 0.0005,
      "loss": 2.7641873359680176,
      "step": 4103
    },
    {
      "epoch": 0.6621490803484995,
      "grad_norm": 1.2689471244812012,
      "learning_rate": 0.0005,
      "loss": 2.79986572265625,
      "step": 4104
    },
    {
      "epoch": 0.6623104227170055,
      "grad_norm": 1.180087924003601,
      "learning_rate": 0.0005,
      "loss": 2.514662742614746,
      "step": 4105
    },
    {
      "epoch": 0.6624717650855114,
      "grad_norm": 1.7238351106643677,
      "learning_rate": 0.0005,
      "loss": 2.3975634574890137,
      "step": 4106
    },
    {
      "epoch": 0.6626331074540174,
      "grad_norm": 1.426509141921997,
      "learning_rate": 0.0005,
      "loss": 2.565145492553711,
      "step": 4107
    },
    {
      "epoch": 0.6627944498225234,
      "grad_norm": 1.0468628406524658,
      "learning_rate": 0.0005,
      "loss": 2.3348312377929688,
      "step": 4108
    },
    {
      "epoch": 0.6629557921910294,
      "grad_norm": 1.2852839231491089,
      "learning_rate": 0.0005,
      "loss": 2.474459171295166,
      "step": 4109
    },
    {
      "epoch": 0.6631171345595354,
      "grad_norm": 1.014176368713379,
      "learning_rate": 0.0005,
      "loss": 2.5012283325195312,
      "step": 4110
    },
    {
      "epoch": 0.6632784769280413,
      "grad_norm": 1.2057996988296509,
      "learning_rate": 0.0005,
      "loss": 2.747323989868164,
      "step": 4111
    },
    {
      "epoch": 0.6634398192965473,
      "grad_norm": 1.3726623058319092,
      "learning_rate": 0.0005,
      "loss": 2.8232502937316895,
      "step": 4112
    },
    {
      "epoch": 0.6636011616650532,
      "grad_norm": 1.2420233488082886,
      "learning_rate": 0.0005,
      "loss": 2.7101681232452393,
      "step": 4113
    },
    {
      "epoch": 0.6637625040335592,
      "grad_norm": 1.3032232522964478,
      "learning_rate": 0.0005,
      "loss": 2.9276881217956543,
      "step": 4114
    },
    {
      "epoch": 0.6639238464020651,
      "grad_norm": 1.3165374994277954,
      "learning_rate": 0.0005,
      "loss": 2.677828788757324,
      "step": 4115
    },
    {
      "epoch": 0.6640851887705711,
      "grad_norm": 1.078810214996338,
      "learning_rate": 0.0005,
      "loss": 2.4976072311401367,
      "step": 4116
    },
    {
      "epoch": 0.6642465311390772,
      "grad_norm": 1.051223635673523,
      "learning_rate": 0.0005,
      "loss": 2.5776829719543457,
      "step": 4117
    },
    {
      "epoch": 0.6644078735075831,
      "grad_norm": 1.3187564611434937,
      "learning_rate": 0.0005,
      "loss": 2.564399242401123,
      "step": 4118
    },
    {
      "epoch": 0.6645692158760891,
      "grad_norm": 0.9691308736801147,
      "learning_rate": 0.0005,
      "loss": 2.0141992568969727,
      "step": 4119
    },
    {
      "epoch": 0.664730558244595,
      "grad_norm": 1.4231479167938232,
      "learning_rate": 0.0005,
      "loss": 2.5973622798919678,
      "step": 4120
    },
    {
      "epoch": 0.664891900613101,
      "grad_norm": 1.316522240638733,
      "learning_rate": 0.0005,
      "loss": 3.5891776084899902,
      "step": 4121
    },
    {
      "epoch": 0.665053242981607,
      "grad_norm": 1.1545841693878174,
      "learning_rate": 0.0005,
      "loss": 2.6452341079711914,
      "step": 4122
    },
    {
      "epoch": 0.6652145853501129,
      "grad_norm": 1.330678105354309,
      "learning_rate": 0.0005,
      "loss": 2.000704765319824,
      "step": 4123
    },
    {
      "epoch": 0.6653759277186189,
      "grad_norm": 1.4231929779052734,
      "learning_rate": 0.0005,
      "loss": 2.777650833129883,
      "step": 4124
    },
    {
      "epoch": 0.6655372700871249,
      "grad_norm": 1.1911325454711914,
      "learning_rate": 0.0005,
      "loss": 2.9498753547668457,
      "step": 4125
    },
    {
      "epoch": 0.6656986124556309,
      "grad_norm": 1.3673357963562012,
      "learning_rate": 0.0005,
      "loss": 2.5411486625671387,
      "step": 4126
    },
    {
      "epoch": 0.6658599548241368,
      "grad_norm": 1.3242322206497192,
      "learning_rate": 0.0005,
      "loss": 2.1920084953308105,
      "step": 4127
    },
    {
      "epoch": 0.6660212971926428,
      "grad_norm": 1.35744047164917,
      "learning_rate": 0.0005,
      "loss": 3.0928680896759033,
      "step": 4128
    },
    {
      "epoch": 0.6661826395611488,
      "grad_norm": 1.2964602708816528,
      "learning_rate": 0.0005,
      "loss": 2.7118237018585205,
      "step": 4129
    },
    {
      "epoch": 0.6663439819296547,
      "grad_norm": 1.2620799541473389,
      "learning_rate": 0.0005,
      "loss": 2.116143226623535,
      "step": 4130
    },
    {
      "epoch": 0.6665053242981607,
      "grad_norm": 1.3506356477737427,
      "learning_rate": 0.0005,
      "loss": 2.7131729125976562,
      "step": 4131
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 1.2796392440795898,
      "learning_rate": 0.0005,
      "loss": 2.400148391723633,
      "step": 4132
    },
    {
      "epoch": 0.6668280090351726,
      "grad_norm": 1.4123857021331787,
      "learning_rate": 0.0005,
      "loss": 2.533144474029541,
      "step": 4133
    },
    {
      "epoch": 0.6669893514036787,
      "grad_norm": 1.3071880340576172,
      "learning_rate": 0.0005,
      "loss": 2.5980124473571777,
      "step": 4134
    },
    {
      "epoch": 0.6671506937721846,
      "grad_norm": 1.0161458253860474,
      "learning_rate": 0.0005,
      "loss": 2.4627318382263184,
      "step": 4135
    },
    {
      "epoch": 0.6673120361406906,
      "grad_norm": 1.0918126106262207,
      "learning_rate": 0.0005,
      "loss": 2.5350046157836914,
      "step": 4136
    },
    {
      "epoch": 0.6674733785091965,
      "grad_norm": 1.3917407989501953,
      "learning_rate": 0.0005,
      "loss": 2.6478943824768066,
      "step": 4137
    },
    {
      "epoch": 0.6676347208777025,
      "grad_norm": 1.4759893417358398,
      "learning_rate": 0.0005,
      "loss": 2.789602756500244,
      "step": 4138
    },
    {
      "epoch": 0.6677960632462084,
      "grad_norm": 1.1292133331298828,
      "learning_rate": 0.0005,
      "loss": 2.5903875827789307,
      "step": 4139
    },
    {
      "epoch": 0.6679574056147144,
      "grad_norm": 1.097827672958374,
      "learning_rate": 0.0005,
      "loss": 2.70975661277771,
      "step": 4140
    },
    {
      "epoch": 0.6681187479832204,
      "grad_norm": 1.5092604160308838,
      "learning_rate": 0.0005,
      "loss": 2.7826476097106934,
      "step": 4141
    },
    {
      "epoch": 0.6682800903517263,
      "grad_norm": 1.2524131536483765,
      "learning_rate": 0.0005,
      "loss": 2.7448019981384277,
      "step": 4142
    },
    {
      "epoch": 0.6684414327202324,
      "grad_norm": 1.5568145513534546,
      "learning_rate": 0.0005,
      "loss": 2.8269410133361816,
      "step": 4143
    },
    {
      "epoch": 0.6686027750887383,
      "grad_norm": 1.12056565284729,
      "learning_rate": 0.0005,
      "loss": 2.8642330169677734,
      "step": 4144
    },
    {
      "epoch": 0.6687641174572443,
      "grad_norm": 0.8952769041061401,
      "learning_rate": 0.0005,
      "loss": 2.716022491455078,
      "step": 4145
    },
    {
      "epoch": 0.6689254598257502,
      "grad_norm": 1.382756233215332,
      "learning_rate": 0.0005,
      "loss": 2.6229753494262695,
      "step": 4146
    },
    {
      "epoch": 0.6690868021942562,
      "grad_norm": 1.1495500802993774,
      "learning_rate": 0.0005,
      "loss": 2.646371364593506,
      "step": 4147
    },
    {
      "epoch": 0.6692481445627622,
      "grad_norm": 1.1774749755859375,
      "learning_rate": 0.0005,
      "loss": 2.8252687454223633,
      "step": 4148
    },
    {
      "epoch": 0.6694094869312681,
      "grad_norm": 1.0621337890625,
      "learning_rate": 0.0005,
      "loss": 2.6290435791015625,
      "step": 4149
    },
    {
      "epoch": 0.6695708292997741,
      "grad_norm": 1.5764390230178833,
      "learning_rate": 0.0005,
      "loss": 2.5272717475891113,
      "step": 4150
    },
    {
      "epoch": 0.6697321716682801,
      "grad_norm": 1.4065905809402466,
      "learning_rate": 0.0005,
      "loss": 2.579322338104248,
      "step": 4151
    },
    {
      "epoch": 0.6698935140367861,
      "grad_norm": 1.3032134771347046,
      "learning_rate": 0.0005,
      "loss": 2.589984893798828,
      "step": 4152
    },
    {
      "epoch": 0.670054856405292,
      "grad_norm": 0.984115481376648,
      "learning_rate": 0.0005,
      "loss": 2.630801200866699,
      "step": 4153
    },
    {
      "epoch": 0.670216198773798,
      "grad_norm": 1.3203582763671875,
      "learning_rate": 0.0005,
      "loss": 3.039595603942871,
      "step": 4154
    },
    {
      "epoch": 0.670377541142304,
      "grad_norm": 1.4059574604034424,
      "learning_rate": 0.0005,
      "loss": 2.5882298946380615,
      "step": 4155
    },
    {
      "epoch": 0.6705388835108099,
      "grad_norm": 1.1577980518341064,
      "learning_rate": 0.0005,
      "loss": 2.573726177215576,
      "step": 4156
    },
    {
      "epoch": 0.6707002258793159,
      "grad_norm": 1.4741274118423462,
      "learning_rate": 0.0005,
      "loss": 2.5766515731811523,
      "step": 4157
    },
    {
      "epoch": 0.6708615682478218,
      "grad_norm": 1.1401036977767944,
      "learning_rate": 0.0005,
      "loss": 2.592255115509033,
      "step": 4158
    },
    {
      "epoch": 0.6710229106163278,
      "grad_norm": 1.1954227685928345,
      "learning_rate": 0.0005,
      "loss": 2.606218099594116,
      "step": 4159
    },
    {
      "epoch": 0.6711842529848339,
      "grad_norm": 1.2329505681991577,
      "learning_rate": 0.0005,
      "loss": 2.7096073627471924,
      "step": 4160
    },
    {
      "epoch": 0.6713455953533398,
      "grad_norm": 1.157309889793396,
      "learning_rate": 0.0005,
      "loss": 2.7145376205444336,
      "step": 4161
    },
    {
      "epoch": 0.6715069377218458,
      "grad_norm": 1.0621778964996338,
      "learning_rate": 0.0005,
      "loss": 2.4751951694488525,
      "step": 4162
    },
    {
      "epoch": 0.6716682800903517,
      "grad_norm": 1.163652777671814,
      "learning_rate": 0.0005,
      "loss": 2.920747756958008,
      "step": 4163
    },
    {
      "epoch": 0.6718296224588577,
      "grad_norm": 0.9622435569763184,
      "learning_rate": 0.0005,
      "loss": 2.5972564220428467,
      "step": 4164
    },
    {
      "epoch": 0.6719909648273636,
      "grad_norm": 0.8759825229644775,
      "learning_rate": 0.0005,
      "loss": 2.5101511478424072,
      "step": 4165
    },
    {
      "epoch": 0.6721523071958696,
      "grad_norm": 1.1327866315841675,
      "learning_rate": 0.0005,
      "loss": 2.6757218837738037,
      "step": 4166
    },
    {
      "epoch": 0.6723136495643756,
      "grad_norm": 1.0350618362426758,
      "learning_rate": 0.0005,
      "loss": 2.533275604248047,
      "step": 4167
    },
    {
      "epoch": 0.6724749919328816,
      "grad_norm": 1.1037455797195435,
      "learning_rate": 0.0005,
      "loss": 2.485715627670288,
      "step": 4168
    },
    {
      "epoch": 0.6726363343013876,
      "grad_norm": 1.5758945941925049,
      "learning_rate": 0.0005,
      "loss": 2.999082088470459,
      "step": 4169
    },
    {
      "epoch": 0.6727976766698935,
      "grad_norm": 1.1886019706726074,
      "learning_rate": 0.0005,
      "loss": 2.657254695892334,
      "step": 4170
    },
    {
      "epoch": 0.6729590190383995,
      "grad_norm": 0.9663862586021423,
      "learning_rate": 0.0005,
      "loss": 2.8108572959899902,
      "step": 4171
    },
    {
      "epoch": 0.6731203614069055,
      "grad_norm": 1.6002105474472046,
      "learning_rate": 0.0005,
      "loss": 2.7001593112945557,
      "step": 4172
    },
    {
      "epoch": 0.6732817037754114,
      "grad_norm": 0.9953100085258484,
      "learning_rate": 0.0005,
      "loss": 2.565920352935791,
      "step": 4173
    },
    {
      "epoch": 0.6734430461439174,
      "grad_norm": 1.0345028638839722,
      "learning_rate": 0.0005,
      "loss": 2.4970126152038574,
      "step": 4174
    },
    {
      "epoch": 0.6736043885124233,
      "grad_norm": 1.2705663442611694,
      "learning_rate": 0.0005,
      "loss": 2.685548782348633,
      "step": 4175
    },
    {
      "epoch": 0.6737657308809293,
      "grad_norm": 1.082732915878296,
      "learning_rate": 0.0005,
      "loss": 2.8808140754699707,
      "step": 4176
    },
    {
      "epoch": 0.6739270732494353,
      "grad_norm": 1.1557954549789429,
      "learning_rate": 0.0005,
      "loss": 3.0889205932617188,
      "step": 4177
    },
    {
      "epoch": 0.6740884156179413,
      "grad_norm": 1.1539051532745361,
      "learning_rate": 0.0005,
      "loss": 2.501514434814453,
      "step": 4178
    },
    {
      "epoch": 0.6742497579864473,
      "grad_norm": 1.2353874444961548,
      "learning_rate": 0.0005,
      "loss": 2.6242880821228027,
      "step": 4179
    },
    {
      "epoch": 0.6744111003549532,
      "grad_norm": 1.2898396253585815,
      "learning_rate": 0.0005,
      "loss": 3.054642677307129,
      "step": 4180
    },
    {
      "epoch": 0.6745724427234592,
      "grad_norm": 1.2549573183059692,
      "learning_rate": 0.0005,
      "loss": 2.8597326278686523,
      "step": 4181
    },
    {
      "epoch": 0.6747337850919651,
      "grad_norm": 1.163065791130066,
      "learning_rate": 0.0005,
      "loss": 2.4518845081329346,
      "step": 4182
    },
    {
      "epoch": 0.6748951274604711,
      "grad_norm": 1.4469270706176758,
      "learning_rate": 0.0005,
      "loss": 3.061727523803711,
      "step": 4183
    },
    {
      "epoch": 0.675056469828977,
      "grad_norm": 1.0100865364074707,
      "learning_rate": 0.0005,
      "loss": 2.659003496170044,
      "step": 4184
    },
    {
      "epoch": 0.6752178121974831,
      "grad_norm": 1.613961935043335,
      "learning_rate": 0.0005,
      "loss": 3.0159335136413574,
      "step": 4185
    },
    {
      "epoch": 0.6753791545659891,
      "grad_norm": 2.6925032138824463,
      "learning_rate": 0.0005,
      "loss": 2.6967363357543945,
      "step": 4186
    },
    {
      "epoch": 0.675540496934495,
      "grad_norm": 1.3034688234329224,
      "learning_rate": 0.0005,
      "loss": 3.2480177879333496,
      "step": 4187
    },
    {
      "epoch": 0.675701839303001,
      "grad_norm": 1.478234052658081,
      "learning_rate": 0.0005,
      "loss": 2.2730510234832764,
      "step": 4188
    },
    {
      "epoch": 0.6758631816715069,
      "grad_norm": 1.224445104598999,
      "learning_rate": 0.0005,
      "loss": 2.5998737812042236,
      "step": 4189
    },
    {
      "epoch": 0.6760245240400129,
      "grad_norm": 1.088907241821289,
      "learning_rate": 0.0005,
      "loss": 2.619490623474121,
      "step": 4190
    },
    {
      "epoch": 0.6761858664085189,
      "grad_norm": 1.0819604396820068,
      "learning_rate": 0.0005,
      "loss": 2.516571521759033,
      "step": 4191
    },
    {
      "epoch": 0.6763472087770248,
      "grad_norm": 1.0891494750976562,
      "learning_rate": 0.0005,
      "loss": 2.422992706298828,
      "step": 4192
    },
    {
      "epoch": 0.6765085511455308,
      "grad_norm": 1.1225117444992065,
      "learning_rate": 0.0005,
      "loss": 2.8549184799194336,
      "step": 4193
    },
    {
      "epoch": 0.6766698935140368,
      "grad_norm": 0.9920690059661865,
      "learning_rate": 0.0005,
      "loss": 2.629666805267334,
      "step": 4194
    },
    {
      "epoch": 0.6768312358825428,
      "grad_norm": 1.8634061813354492,
      "learning_rate": 0.0005,
      "loss": 2.790104866027832,
      "step": 4195
    },
    {
      "epoch": 0.6769925782510487,
      "grad_norm": 1.3537847995758057,
      "learning_rate": 0.0005,
      "loss": 2.488677978515625,
      "step": 4196
    },
    {
      "epoch": 0.6771539206195547,
      "grad_norm": 1.151623010635376,
      "learning_rate": 0.0005,
      "loss": 2.740252733230591,
      "step": 4197
    },
    {
      "epoch": 0.6773152629880607,
      "grad_norm": 1.330542802810669,
      "learning_rate": 0.0005,
      "loss": 2.597649335861206,
      "step": 4198
    },
    {
      "epoch": 0.6774766053565666,
      "grad_norm": 1.0837327241897583,
      "learning_rate": 0.0005,
      "loss": 2.6141293048858643,
      "step": 4199
    },
    {
      "epoch": 0.6776379477250726,
      "grad_norm": 1.0132684707641602,
      "learning_rate": 0.0005,
      "loss": 2.443101406097412,
      "step": 4200
    },
    {
      "epoch": 0.6777992900935785,
      "grad_norm": 1.3305720090866089,
      "learning_rate": 0.0005,
      "loss": 2.707731246948242,
      "step": 4201
    },
    {
      "epoch": 0.6779606324620845,
      "grad_norm": 1.4712436199188232,
      "learning_rate": 0.0005,
      "loss": 2.6216917037963867,
      "step": 4202
    },
    {
      "epoch": 0.6781219748305906,
      "grad_norm": 1.2421849966049194,
      "learning_rate": 0.0005,
      "loss": 2.5358152389526367,
      "step": 4203
    },
    {
      "epoch": 0.6782833171990965,
      "grad_norm": 1.1763030290603638,
      "learning_rate": 0.0005,
      "loss": 2.5785603523254395,
      "step": 4204
    },
    {
      "epoch": 0.6784446595676025,
      "grad_norm": 1.30049467086792,
      "learning_rate": 0.0005,
      "loss": 2.198280096054077,
      "step": 4205
    },
    {
      "epoch": 0.6786060019361084,
      "grad_norm": 1.0975278615951538,
      "learning_rate": 0.0005,
      "loss": 2.5282859802246094,
      "step": 4206
    },
    {
      "epoch": 0.6787673443046144,
      "grad_norm": 1.216774344444275,
      "learning_rate": 0.0005,
      "loss": 2.7917680740356445,
      "step": 4207
    },
    {
      "epoch": 0.6789286866731203,
      "grad_norm": 0.9682071805000305,
      "learning_rate": 0.0005,
      "loss": 2.583120822906494,
      "step": 4208
    },
    {
      "epoch": 0.6790900290416263,
      "grad_norm": 1.0086090564727783,
      "learning_rate": 0.0005,
      "loss": 2.571815013885498,
      "step": 4209
    },
    {
      "epoch": 0.6792513714101323,
      "grad_norm": 1.0844650268554688,
      "learning_rate": 0.0005,
      "loss": 2.730487823486328,
      "step": 4210
    },
    {
      "epoch": 0.6794127137786383,
      "grad_norm": 1.0690959692001343,
      "learning_rate": 0.0005,
      "loss": 2.344773054122925,
      "step": 4211
    },
    {
      "epoch": 0.6795740561471443,
      "grad_norm": 1.0133121013641357,
      "learning_rate": 0.0005,
      "loss": 2.7132978439331055,
      "step": 4212
    },
    {
      "epoch": 0.6797353985156502,
      "grad_norm": 0.925731360912323,
      "learning_rate": 0.0005,
      "loss": 2.7603940963745117,
      "step": 4213
    },
    {
      "epoch": 0.6798967408841562,
      "grad_norm": 1.5364669561386108,
      "learning_rate": 0.0005,
      "loss": 2.588686466217041,
      "step": 4214
    },
    {
      "epoch": 0.6800580832526621,
      "grad_norm": 1.5932390689849854,
      "learning_rate": 0.0005,
      "loss": 2.4961626529693604,
      "step": 4215
    },
    {
      "epoch": 0.6802194256211681,
      "grad_norm": 1.569256067276001,
      "learning_rate": 0.0005,
      "loss": 2.7376608848571777,
      "step": 4216
    },
    {
      "epoch": 0.6803807679896741,
      "grad_norm": 0.9984979629516602,
      "learning_rate": 0.0005,
      "loss": 2.498483657836914,
      "step": 4217
    },
    {
      "epoch": 0.68054211035818,
      "grad_norm": 1.2279465198516846,
      "learning_rate": 0.0005,
      "loss": 2.605848789215088,
      "step": 4218
    },
    {
      "epoch": 0.680703452726686,
      "grad_norm": 1.0248712301254272,
      "learning_rate": 0.0005,
      "loss": 2.584446668624878,
      "step": 4219
    },
    {
      "epoch": 0.680864795095192,
      "grad_norm": 1.6338461637496948,
      "learning_rate": 0.0005,
      "loss": 3.330331802368164,
      "step": 4220
    },
    {
      "epoch": 0.681026137463698,
      "grad_norm": 1.3686180114746094,
      "learning_rate": 0.0005,
      "loss": 2.902024269104004,
      "step": 4221
    },
    {
      "epoch": 0.681187479832204,
      "grad_norm": 2.3319339752197266,
      "learning_rate": 0.0005,
      "loss": 2.3029022216796875,
      "step": 4222
    },
    {
      "epoch": 0.6813488222007099,
      "grad_norm": 1.1751364469528198,
      "learning_rate": 0.0005,
      "loss": 2.565335750579834,
      "step": 4223
    },
    {
      "epoch": 0.6815101645692159,
      "grad_norm": 1.376888632774353,
      "learning_rate": 0.0005,
      "loss": 3.1199896335601807,
      "step": 4224
    },
    {
      "epoch": 0.6816715069377218,
      "grad_norm": 1.0778534412384033,
      "learning_rate": 0.0005,
      "loss": 3.011117458343506,
      "step": 4225
    },
    {
      "epoch": 0.6818328493062278,
      "grad_norm": 1.3871610164642334,
      "learning_rate": 0.0005,
      "loss": 2.837317705154419,
      "step": 4226
    },
    {
      "epoch": 0.6819941916747337,
      "grad_norm": 1.2338835000991821,
      "learning_rate": 0.0005,
      "loss": 2.7216310501098633,
      "step": 4227
    },
    {
      "epoch": 0.6821555340432398,
      "grad_norm": 1.0193153619766235,
      "learning_rate": 0.0005,
      "loss": 2.7204947471618652,
      "step": 4228
    },
    {
      "epoch": 0.6823168764117458,
      "grad_norm": 1.2107067108154297,
      "learning_rate": 0.0005,
      "loss": 2.700026750564575,
      "step": 4229
    },
    {
      "epoch": 0.6824782187802517,
      "grad_norm": 0.9520393013954163,
      "learning_rate": 0.0005,
      "loss": 2.2671797275543213,
      "step": 4230
    },
    {
      "epoch": 0.6826395611487577,
      "grad_norm": 1.0453983545303345,
      "learning_rate": 0.0005,
      "loss": 2.464063882827759,
      "step": 4231
    },
    {
      "epoch": 0.6828009035172636,
      "grad_norm": 1.100490689277649,
      "learning_rate": 0.0005,
      "loss": 2.6748995780944824,
      "step": 4232
    },
    {
      "epoch": 0.6829622458857696,
      "grad_norm": 1.3008148670196533,
      "learning_rate": 0.0005,
      "loss": 2.761357545852661,
      "step": 4233
    },
    {
      "epoch": 0.6831235882542755,
      "grad_norm": 1.2342664003372192,
      "learning_rate": 0.0005,
      "loss": 2.619274139404297,
      "step": 4234
    },
    {
      "epoch": 0.6832849306227815,
      "grad_norm": 1.5275169610977173,
      "learning_rate": 0.0005,
      "loss": 2.573021650314331,
      "step": 4235
    },
    {
      "epoch": 0.6834462729912875,
      "grad_norm": 1.057270884513855,
      "learning_rate": 0.0005,
      "loss": 2.393214225769043,
      "step": 4236
    },
    {
      "epoch": 0.6836076153597935,
      "grad_norm": 1.2774615287780762,
      "learning_rate": 0.0005,
      "loss": 2.4153075218200684,
      "step": 4237
    },
    {
      "epoch": 0.6837689577282995,
      "grad_norm": 1.21660578250885,
      "learning_rate": 0.0005,
      "loss": 2.522763729095459,
      "step": 4238
    },
    {
      "epoch": 0.6839303000968054,
      "grad_norm": 1.0774894952774048,
      "learning_rate": 0.0005,
      "loss": 2.689786434173584,
      "step": 4239
    },
    {
      "epoch": 0.6840916424653114,
      "grad_norm": 1.0757695436477661,
      "learning_rate": 0.0005,
      "loss": 2.547933340072632,
      "step": 4240
    },
    {
      "epoch": 0.6842529848338174,
      "grad_norm": 1.1774698495864868,
      "learning_rate": 0.0005,
      "loss": 2.7426414489746094,
      "step": 4241
    },
    {
      "epoch": 0.6844143272023233,
      "grad_norm": 1.1907838582992554,
      "learning_rate": 0.0005,
      "loss": 2.792241096496582,
      "step": 4242
    },
    {
      "epoch": 0.6845756695708293,
      "grad_norm": 1.0536720752716064,
      "learning_rate": 0.0005,
      "loss": 2.5572142601013184,
      "step": 4243
    },
    {
      "epoch": 0.6847370119393352,
      "grad_norm": 1.3971896171569824,
      "learning_rate": 0.0005,
      "loss": 2.1332144737243652,
      "step": 4244
    },
    {
      "epoch": 0.6848983543078412,
      "grad_norm": 1.3285967111587524,
      "learning_rate": 0.0005,
      "loss": 2.9223837852478027,
      "step": 4245
    },
    {
      "epoch": 0.6850596966763473,
      "grad_norm": 1.417879581451416,
      "learning_rate": 0.0005,
      "loss": 2.5293662548065186,
      "step": 4246
    },
    {
      "epoch": 0.6852210390448532,
      "grad_norm": 1.2997183799743652,
      "learning_rate": 0.0005,
      "loss": 2.782132625579834,
      "step": 4247
    },
    {
      "epoch": 0.6853823814133592,
      "grad_norm": 0.9826723337173462,
      "learning_rate": 0.0005,
      "loss": 2.002516746520996,
      "step": 4248
    },
    {
      "epoch": 0.6855437237818651,
      "grad_norm": 1.3859142065048218,
      "learning_rate": 0.0005,
      "loss": 2.5375452041625977,
      "step": 4249
    },
    {
      "epoch": 0.6857050661503711,
      "grad_norm": 1.1768219470977783,
      "learning_rate": 0.0005,
      "loss": 3.1994152069091797,
      "step": 4250
    },
    {
      "epoch": 0.685866408518877,
      "grad_norm": 1.0620815753936768,
      "learning_rate": 0.0005,
      "loss": 2.2051825523376465,
      "step": 4251
    },
    {
      "epoch": 0.686027750887383,
      "grad_norm": 1.0677378177642822,
      "learning_rate": 0.0005,
      "loss": 2.3548645973205566,
      "step": 4252
    },
    {
      "epoch": 0.686189093255889,
      "grad_norm": 1.6115810871124268,
      "learning_rate": 0.0005,
      "loss": 2.417306900024414,
      "step": 4253
    },
    {
      "epoch": 0.686350435624395,
      "grad_norm": 1.038526177406311,
      "learning_rate": 0.0005,
      "loss": 2.6624066829681396,
      "step": 4254
    },
    {
      "epoch": 0.686511777992901,
      "grad_norm": 1.0622936487197876,
      "learning_rate": 0.0005,
      "loss": 2.326395034790039,
      "step": 4255
    },
    {
      "epoch": 0.6866731203614069,
      "grad_norm": 1.1461071968078613,
      "learning_rate": 0.0005,
      "loss": 2.5159006118774414,
      "step": 4256
    },
    {
      "epoch": 0.6868344627299129,
      "grad_norm": 0.854144811630249,
      "learning_rate": 0.0005,
      "loss": 2.086803436279297,
      "step": 4257
    },
    {
      "epoch": 0.6869958050984188,
      "grad_norm": 1.0934221744537354,
      "learning_rate": 0.0005,
      "loss": 2.5924055576324463,
      "step": 4258
    },
    {
      "epoch": 0.6871571474669248,
      "grad_norm": 1.1256012916564941,
      "learning_rate": 0.0005,
      "loss": 2.6850953102111816,
      "step": 4259
    },
    {
      "epoch": 0.6873184898354308,
      "grad_norm": 1.3469024896621704,
      "learning_rate": 0.0005,
      "loss": 2.816962718963623,
      "step": 4260
    },
    {
      "epoch": 0.6874798322039367,
      "grad_norm": 1.138877272605896,
      "learning_rate": 0.0005,
      "loss": 2.3434038162231445,
      "step": 4261
    },
    {
      "epoch": 0.6876411745724427,
      "grad_norm": 1.2110002040863037,
      "learning_rate": 0.0005,
      "loss": 2.653294086456299,
      "step": 4262
    },
    {
      "epoch": 0.6878025169409487,
      "grad_norm": 1.223833680152893,
      "learning_rate": 0.0005,
      "loss": 2.76611328125,
      "step": 4263
    },
    {
      "epoch": 0.6879638593094547,
      "grad_norm": 1.1492433547973633,
      "learning_rate": 0.0005,
      "loss": 2.296595573425293,
      "step": 4264
    },
    {
      "epoch": 0.6881252016779607,
      "grad_norm": 1.2836706638336182,
      "learning_rate": 0.0005,
      "loss": 2.6139793395996094,
      "step": 4265
    },
    {
      "epoch": 0.6882865440464666,
      "grad_norm": 0.9766125679016113,
      "learning_rate": 0.0005,
      "loss": 2.421114683151245,
      "step": 4266
    },
    {
      "epoch": 0.6884478864149726,
      "grad_norm": 1.8313053846359253,
      "learning_rate": 0.0005,
      "loss": 2.5466043949127197,
      "step": 4267
    },
    {
      "epoch": 0.6886092287834785,
      "grad_norm": 1.1891613006591797,
      "learning_rate": 0.0005,
      "loss": 2.380479335784912,
      "step": 4268
    },
    {
      "epoch": 0.6887705711519845,
      "grad_norm": 1.1631842851638794,
      "learning_rate": 0.0005,
      "loss": 2.4699201583862305,
      "step": 4269
    },
    {
      "epoch": 0.6889319135204904,
      "grad_norm": 1.728432536125183,
      "learning_rate": 0.0005,
      "loss": 2.7784252166748047,
      "step": 4270
    },
    {
      "epoch": 0.6890932558889965,
      "grad_norm": 1.0042495727539062,
      "learning_rate": 0.0005,
      "loss": 2.642385721206665,
      "step": 4271
    },
    {
      "epoch": 0.6892545982575025,
      "grad_norm": 1.216499924659729,
      "learning_rate": 0.0005,
      "loss": 3.347574234008789,
      "step": 4272
    },
    {
      "epoch": 0.6894159406260084,
      "grad_norm": 1.2104612588882446,
      "learning_rate": 0.0005,
      "loss": 2.535073757171631,
      "step": 4273
    },
    {
      "epoch": 0.6895772829945144,
      "grad_norm": 1.7385120391845703,
      "learning_rate": 0.0005,
      "loss": 2.7160370349884033,
      "step": 4274
    },
    {
      "epoch": 0.6897386253630203,
      "grad_norm": 1.4963302612304688,
      "learning_rate": 0.0005,
      "loss": 3.016515016555786,
      "step": 4275
    },
    {
      "epoch": 0.6898999677315263,
      "grad_norm": 1.0684396028518677,
      "learning_rate": 0.0005,
      "loss": 2.8192496299743652,
      "step": 4276
    },
    {
      "epoch": 0.6900613101000322,
      "grad_norm": 0.96053546667099,
      "learning_rate": 0.0005,
      "loss": 2.397970676422119,
      "step": 4277
    },
    {
      "epoch": 0.6902226524685382,
      "grad_norm": 1.2071216106414795,
      "learning_rate": 0.0005,
      "loss": 2.4374279975891113,
      "step": 4278
    },
    {
      "epoch": 0.6903839948370442,
      "grad_norm": 1.3319369554519653,
      "learning_rate": 0.0005,
      "loss": 2.668301582336426,
      "step": 4279
    },
    {
      "epoch": 0.6905453372055502,
      "grad_norm": 1.1643435955047607,
      "learning_rate": 0.0005,
      "loss": 2.720707893371582,
      "step": 4280
    },
    {
      "epoch": 0.6907066795740562,
      "grad_norm": 1.4473015069961548,
      "learning_rate": 0.0005,
      "loss": 2.7478036880493164,
      "step": 4281
    },
    {
      "epoch": 0.6908680219425621,
      "grad_norm": 1.161092758178711,
      "learning_rate": 0.0005,
      "loss": 2.6853082180023193,
      "step": 4282
    },
    {
      "epoch": 0.6910293643110681,
      "grad_norm": 1.4382288455963135,
      "learning_rate": 0.0005,
      "loss": 2.757002830505371,
      "step": 4283
    },
    {
      "epoch": 0.691190706679574,
      "grad_norm": 1.0602760314941406,
      "learning_rate": 0.0005,
      "loss": 2.6494359970092773,
      "step": 4284
    },
    {
      "epoch": 0.69135204904808,
      "grad_norm": 1.083195686340332,
      "learning_rate": 0.0005,
      "loss": 2.202641487121582,
      "step": 4285
    },
    {
      "epoch": 0.691513391416586,
      "grad_norm": 1.5319340229034424,
      "learning_rate": 0.0005,
      "loss": 3.162597179412842,
      "step": 4286
    },
    {
      "epoch": 0.6916747337850919,
      "grad_norm": 1.3790476322174072,
      "learning_rate": 0.0005,
      "loss": 2.455561637878418,
      "step": 4287
    },
    {
      "epoch": 0.691836076153598,
      "grad_norm": 1.4068571329116821,
      "learning_rate": 0.0005,
      "loss": 2.2528562545776367,
      "step": 4288
    },
    {
      "epoch": 0.691997418522104,
      "grad_norm": 1.0188528299331665,
      "learning_rate": 0.0005,
      "loss": 2.485591411590576,
      "step": 4289
    },
    {
      "epoch": 0.6921587608906099,
      "grad_norm": 1.0855642557144165,
      "learning_rate": 0.0005,
      "loss": 2.375978469848633,
      "step": 4290
    },
    {
      "epoch": 0.6923201032591159,
      "grad_norm": 1.060429573059082,
      "learning_rate": 0.0005,
      "loss": 2.3212990760803223,
      "step": 4291
    },
    {
      "epoch": 0.6924814456276218,
      "grad_norm": 1.2221972942352295,
      "learning_rate": 0.0005,
      "loss": 2.9161312580108643,
      "step": 4292
    },
    {
      "epoch": 0.6926427879961278,
      "grad_norm": 1.1824359893798828,
      "learning_rate": 0.0005,
      "loss": 3.2717995643615723,
      "step": 4293
    },
    {
      "epoch": 0.6928041303646337,
      "grad_norm": 1.256733775138855,
      "learning_rate": 0.0005,
      "loss": 2.9251489639282227,
      "step": 4294
    },
    {
      "epoch": 0.6929654727331397,
      "grad_norm": 1.393523931503296,
      "learning_rate": 0.0005,
      "loss": 2.7846577167510986,
      "step": 4295
    },
    {
      "epoch": 0.6931268151016456,
      "grad_norm": 1.0313735008239746,
      "learning_rate": 0.0005,
      "loss": 2.574937343597412,
      "step": 4296
    },
    {
      "epoch": 0.6932881574701517,
      "grad_norm": 1.0327215194702148,
      "learning_rate": 0.0005,
      "loss": 2.1531448364257812,
      "step": 4297
    },
    {
      "epoch": 0.6934494998386577,
      "grad_norm": 1.1799460649490356,
      "learning_rate": 0.0005,
      "loss": 3.0011777877807617,
      "step": 4298
    },
    {
      "epoch": 0.6936108422071636,
      "grad_norm": 1.0098379850387573,
      "learning_rate": 0.0005,
      "loss": 2.5690958499908447,
      "step": 4299
    },
    {
      "epoch": 0.6937721845756696,
      "grad_norm": 1.0364463329315186,
      "learning_rate": 0.0005,
      "loss": 3.022449254989624,
      "step": 4300
    },
    {
      "epoch": 0.6939335269441755,
      "grad_norm": 1.1771966218948364,
      "learning_rate": 0.0005,
      "loss": 2.2320942878723145,
      "step": 4301
    },
    {
      "epoch": 0.6940948693126815,
      "grad_norm": 1.2286486625671387,
      "learning_rate": 0.0005,
      "loss": 2.7485289573669434,
      "step": 4302
    },
    {
      "epoch": 0.6942562116811875,
      "grad_norm": 1.3581188917160034,
      "learning_rate": 0.0005,
      "loss": 2.164371967315674,
      "step": 4303
    },
    {
      "epoch": 0.6944175540496934,
      "grad_norm": 1.0557706356048584,
      "learning_rate": 0.0005,
      "loss": 2.674389123916626,
      "step": 4304
    },
    {
      "epoch": 0.6945788964181994,
      "grad_norm": 1.2559456825256348,
      "learning_rate": 0.0005,
      "loss": 2.712273120880127,
      "step": 4305
    },
    {
      "epoch": 0.6947402387867054,
      "grad_norm": 1.675311803817749,
      "learning_rate": 0.0005,
      "loss": 2.673515796661377,
      "step": 4306
    },
    {
      "epoch": 0.6949015811552114,
      "grad_norm": 1.07296884059906,
      "learning_rate": 0.0005,
      "loss": 2.6503515243530273,
      "step": 4307
    },
    {
      "epoch": 0.6950629235237173,
      "grad_norm": 1.057946801185608,
      "learning_rate": 0.0005,
      "loss": 2.766883373260498,
      "step": 4308
    },
    {
      "epoch": 0.6952242658922233,
      "grad_norm": 1.4015620946884155,
      "learning_rate": 0.0005,
      "loss": 2.4534454345703125,
      "step": 4309
    },
    {
      "epoch": 0.6953856082607293,
      "grad_norm": 1.1588214635849,
      "learning_rate": 0.0005,
      "loss": 2.7297356128692627,
      "step": 4310
    },
    {
      "epoch": 0.6955469506292352,
      "grad_norm": 1.3314579725265503,
      "learning_rate": 0.0005,
      "loss": 3.122300148010254,
      "step": 4311
    },
    {
      "epoch": 0.6957082929977412,
      "grad_norm": 1.2616503238677979,
      "learning_rate": 0.0005,
      "loss": 2.8456764221191406,
      "step": 4312
    },
    {
      "epoch": 0.6958696353662471,
      "grad_norm": 1.466168999671936,
      "learning_rate": 0.0005,
      "loss": 2.6792848110198975,
      "step": 4313
    },
    {
      "epoch": 0.6960309777347532,
      "grad_norm": 1.2682393789291382,
      "learning_rate": 0.0005,
      "loss": 2.50671648979187,
      "step": 4314
    },
    {
      "epoch": 0.6961923201032592,
      "grad_norm": 1.459399700164795,
      "learning_rate": 0.0005,
      "loss": 2.50520396232605,
      "step": 4315
    },
    {
      "epoch": 0.6963536624717651,
      "grad_norm": 1.2763259410858154,
      "learning_rate": 0.0005,
      "loss": 2.4020373821258545,
      "step": 4316
    },
    {
      "epoch": 0.6965150048402711,
      "grad_norm": 1.479565978050232,
      "learning_rate": 0.0005,
      "loss": 2.9249181747436523,
      "step": 4317
    },
    {
      "epoch": 0.696676347208777,
      "grad_norm": 1.1765602827072144,
      "learning_rate": 0.0005,
      "loss": 2.3543550968170166,
      "step": 4318
    },
    {
      "epoch": 0.696837689577283,
      "grad_norm": 1.0912549495697021,
      "learning_rate": 0.0005,
      "loss": 2.4353606700897217,
      "step": 4319
    },
    {
      "epoch": 0.6969990319457889,
      "grad_norm": 2.03171706199646,
      "learning_rate": 0.0005,
      "loss": 2.651675224304199,
      "step": 4320
    },
    {
      "epoch": 0.6971603743142949,
      "grad_norm": 1.9236507415771484,
      "learning_rate": 0.0005,
      "loss": 2.467564582824707,
      "step": 4321
    },
    {
      "epoch": 0.6973217166828009,
      "grad_norm": 1.0934473276138306,
      "learning_rate": 0.0005,
      "loss": 2.3825221061706543,
      "step": 4322
    },
    {
      "epoch": 0.6974830590513069,
      "grad_norm": 1.3332617282867432,
      "learning_rate": 0.0005,
      "loss": 2.293341636657715,
      "step": 4323
    },
    {
      "epoch": 0.6976444014198129,
      "grad_norm": 1.301454782485962,
      "learning_rate": 0.0005,
      "loss": 2.3393709659576416,
      "step": 4324
    },
    {
      "epoch": 0.6978057437883188,
      "grad_norm": 1.3101270198822021,
      "learning_rate": 0.0005,
      "loss": 3.167435646057129,
      "step": 4325
    },
    {
      "epoch": 0.6979670861568248,
      "grad_norm": 1.2261748313903809,
      "learning_rate": 0.0005,
      "loss": 2.4380130767822266,
      "step": 4326
    },
    {
      "epoch": 0.6981284285253307,
      "grad_norm": 1.5031076669692993,
      "learning_rate": 0.0005,
      "loss": 2.7700035572052,
      "step": 4327
    },
    {
      "epoch": 0.6982897708938367,
      "grad_norm": 1.4178647994995117,
      "learning_rate": 0.0005,
      "loss": 2.641284465789795,
      "step": 4328
    },
    {
      "epoch": 0.6984511132623427,
      "grad_norm": 1.351775884628296,
      "learning_rate": 0.0005,
      "loss": 2.4092774391174316,
      "step": 4329
    },
    {
      "epoch": 0.6986124556308486,
      "grad_norm": 1.0329128503799438,
      "learning_rate": 0.0005,
      "loss": 2.5740222930908203,
      "step": 4330
    },
    {
      "epoch": 0.6987737979993547,
      "grad_norm": 1.547524094581604,
      "learning_rate": 0.0005,
      "loss": 2.5946407318115234,
      "step": 4331
    },
    {
      "epoch": 0.6989351403678606,
      "grad_norm": 1.3223299980163574,
      "learning_rate": 0.0005,
      "loss": 2.6585206985473633,
      "step": 4332
    },
    {
      "epoch": 0.6990964827363666,
      "grad_norm": 1.1976274251937866,
      "learning_rate": 0.0005,
      "loss": 2.354005813598633,
      "step": 4333
    },
    {
      "epoch": 0.6992578251048726,
      "grad_norm": 1.2580572366714478,
      "learning_rate": 0.0005,
      "loss": 2.7076218128204346,
      "step": 4334
    },
    {
      "epoch": 0.6994191674733785,
      "grad_norm": 1.181484341621399,
      "learning_rate": 0.0005,
      "loss": 2.956712007522583,
      "step": 4335
    },
    {
      "epoch": 0.6995805098418845,
      "grad_norm": 1.0671216249465942,
      "learning_rate": 0.0005,
      "loss": 2.7417008876800537,
      "step": 4336
    },
    {
      "epoch": 0.6997418522103904,
      "grad_norm": 1.226249098777771,
      "learning_rate": 0.0005,
      "loss": 2.602318048477173,
      "step": 4337
    },
    {
      "epoch": 0.6999031945788964,
      "grad_norm": 1.2037718296051025,
      "learning_rate": 0.0005,
      "loss": 2.660144805908203,
      "step": 4338
    },
    {
      "epoch": 0.7000645369474023,
      "grad_norm": 1.0918534994125366,
      "learning_rate": 0.0005,
      "loss": 2.4661271572113037,
      "step": 4339
    },
    {
      "epoch": 0.7002258793159084,
      "grad_norm": 1.1225390434265137,
      "learning_rate": 0.0005,
      "loss": 2.245403289794922,
      "step": 4340
    },
    {
      "epoch": 0.7003872216844144,
      "grad_norm": 1.0744870901107788,
      "learning_rate": 0.0005,
      "loss": 3.241179943084717,
      "step": 4341
    },
    {
      "epoch": 0.7005485640529203,
      "grad_norm": 1.7579764127731323,
      "learning_rate": 0.0005,
      "loss": 2.6713314056396484,
      "step": 4342
    },
    {
      "epoch": 0.7007099064214263,
      "grad_norm": 1.2542550563812256,
      "learning_rate": 0.0005,
      "loss": 2.7298383712768555,
      "step": 4343
    },
    {
      "epoch": 0.7008712487899322,
      "grad_norm": 1.3864096403121948,
      "learning_rate": 0.0005,
      "loss": 2.6198556423187256,
      "step": 4344
    },
    {
      "epoch": 0.7010325911584382,
      "grad_norm": 1.3527909517288208,
      "learning_rate": 0.0005,
      "loss": 2.732802391052246,
      "step": 4345
    },
    {
      "epoch": 0.7011939335269441,
      "grad_norm": 1.3253098726272583,
      "learning_rate": 0.0005,
      "loss": 2.5429224967956543,
      "step": 4346
    },
    {
      "epoch": 0.7013552758954501,
      "grad_norm": 1.1010472774505615,
      "learning_rate": 0.0005,
      "loss": 2.326199531555176,
      "step": 4347
    },
    {
      "epoch": 0.7015166182639561,
      "grad_norm": 1.3029654026031494,
      "learning_rate": 0.0005,
      "loss": 2.2119216918945312,
      "step": 4348
    },
    {
      "epoch": 0.7016779606324621,
      "grad_norm": 1.1104495525360107,
      "learning_rate": 0.0005,
      "loss": 3.1265275478363037,
      "step": 4349
    },
    {
      "epoch": 0.7018393030009681,
      "grad_norm": 1.0436605215072632,
      "learning_rate": 0.0005,
      "loss": 2.951955795288086,
      "step": 4350
    },
    {
      "epoch": 0.702000645369474,
      "grad_norm": 1.2684237957000732,
      "learning_rate": 0.0005,
      "loss": 2.2551519870758057,
      "step": 4351
    },
    {
      "epoch": 0.70216198773798,
      "grad_norm": 1.064245343208313,
      "learning_rate": 0.0005,
      "loss": 2.676079273223877,
      "step": 4352
    },
    {
      "epoch": 0.702323330106486,
      "grad_norm": 1.4330666065216064,
      "learning_rate": 0.0005,
      "loss": 2.3033061027526855,
      "step": 4353
    },
    {
      "epoch": 0.7024846724749919,
      "grad_norm": 1.2056584358215332,
      "learning_rate": 0.0005,
      "loss": 2.8102684020996094,
      "step": 4354
    },
    {
      "epoch": 0.7026460148434979,
      "grad_norm": 1.430150032043457,
      "learning_rate": 0.0005,
      "loss": 2.4469919204711914,
      "step": 4355
    },
    {
      "epoch": 0.7028073572120038,
      "grad_norm": 0.9443376064300537,
      "learning_rate": 0.0005,
      "loss": 1.9553499221801758,
      "step": 4356
    },
    {
      "epoch": 0.7029686995805099,
      "grad_norm": 1.034079670906067,
      "learning_rate": 0.0005,
      "loss": 2.497718095779419,
      "step": 4357
    },
    {
      "epoch": 0.7031300419490158,
      "grad_norm": 1.3414676189422607,
      "learning_rate": 0.0005,
      "loss": 2.354421615600586,
      "step": 4358
    },
    {
      "epoch": 0.7032913843175218,
      "grad_norm": 1.3877424001693726,
      "learning_rate": 0.0005,
      "loss": 2.386474847793579,
      "step": 4359
    },
    {
      "epoch": 0.7034527266860278,
      "grad_norm": 1.6145726442337036,
      "learning_rate": 0.0005,
      "loss": 2.4424309730529785,
      "step": 4360
    },
    {
      "epoch": 0.7036140690545337,
      "grad_norm": 1.3107548952102661,
      "learning_rate": 0.0005,
      "loss": 2.933908462524414,
      "step": 4361
    },
    {
      "epoch": 0.7037754114230397,
      "grad_norm": 1.662217617034912,
      "learning_rate": 0.0005,
      "loss": 3.0468740463256836,
      "step": 4362
    },
    {
      "epoch": 0.7039367537915456,
      "grad_norm": 1.6219843626022339,
      "learning_rate": 0.0005,
      "loss": 2.555065631866455,
      "step": 4363
    },
    {
      "epoch": 0.7040980961600516,
      "grad_norm": 1.3438385725021362,
      "learning_rate": 0.0005,
      "loss": 2.912980556488037,
      "step": 4364
    },
    {
      "epoch": 0.7042594385285575,
      "grad_norm": 1.5910874605178833,
      "learning_rate": 0.0005,
      "loss": 2.2862389087677,
      "step": 4365
    },
    {
      "epoch": 0.7044207808970636,
      "grad_norm": 1.4712166786193848,
      "learning_rate": 0.0005,
      "loss": 3.014319658279419,
      "step": 4366
    },
    {
      "epoch": 0.7045821232655696,
      "grad_norm": 1.4315662384033203,
      "learning_rate": 0.0005,
      "loss": 2.5047082901000977,
      "step": 4367
    },
    {
      "epoch": 0.7047434656340755,
      "grad_norm": 1.2868629693984985,
      "learning_rate": 0.0005,
      "loss": 2.8288538455963135,
      "step": 4368
    },
    {
      "epoch": 0.7049048080025815,
      "grad_norm": 1.2700867652893066,
      "learning_rate": 0.0005,
      "loss": 2.7089526653289795,
      "step": 4369
    },
    {
      "epoch": 0.7050661503710874,
      "grad_norm": 1.4375507831573486,
      "learning_rate": 0.0005,
      "loss": 2.7187881469726562,
      "step": 4370
    },
    {
      "epoch": 0.7052274927395934,
      "grad_norm": 1.0775458812713623,
      "learning_rate": 0.0005,
      "loss": 2.7467384338378906,
      "step": 4371
    },
    {
      "epoch": 0.7053888351080994,
      "grad_norm": 0.9413638710975647,
      "learning_rate": 0.0005,
      "loss": 2.305546522140503,
      "step": 4372
    },
    {
      "epoch": 0.7055501774766053,
      "grad_norm": 1.059064269065857,
      "learning_rate": 0.0005,
      "loss": 2.7211239337921143,
      "step": 4373
    },
    {
      "epoch": 0.7057115198451114,
      "grad_norm": 1.1291438341140747,
      "learning_rate": 0.0005,
      "loss": 2.6069557666778564,
      "step": 4374
    },
    {
      "epoch": 0.7058728622136173,
      "grad_norm": 1.5694177150726318,
      "learning_rate": 0.0005,
      "loss": 2.756411552429199,
      "step": 4375
    },
    {
      "epoch": 0.7060342045821233,
      "grad_norm": 1.2312270402908325,
      "learning_rate": 0.0005,
      "loss": 2.8354923725128174,
      "step": 4376
    },
    {
      "epoch": 0.7061955469506292,
      "grad_norm": 1.2572275400161743,
      "learning_rate": 0.0005,
      "loss": 2.7749574184417725,
      "step": 4377
    },
    {
      "epoch": 0.7063568893191352,
      "grad_norm": 1.174553394317627,
      "learning_rate": 0.0005,
      "loss": 3.106534004211426,
      "step": 4378
    },
    {
      "epoch": 0.7065182316876412,
      "grad_norm": 1.304879903793335,
      "learning_rate": 0.0005,
      "loss": 2.4904251098632812,
      "step": 4379
    },
    {
      "epoch": 0.7066795740561471,
      "grad_norm": 1.0897789001464844,
      "learning_rate": 0.0005,
      "loss": 2.467088460922241,
      "step": 4380
    },
    {
      "epoch": 0.7068409164246531,
      "grad_norm": 1.2234857082366943,
      "learning_rate": 0.0005,
      "loss": 2.490147590637207,
      "step": 4381
    },
    {
      "epoch": 0.707002258793159,
      "grad_norm": 1.1329936981201172,
      "learning_rate": 0.0005,
      "loss": 2.7613227367401123,
      "step": 4382
    },
    {
      "epoch": 0.7071636011616651,
      "grad_norm": 1.083865761756897,
      "learning_rate": 0.0005,
      "loss": 2.3226099014282227,
      "step": 4383
    },
    {
      "epoch": 0.7073249435301711,
      "grad_norm": 1.3984719514846802,
      "learning_rate": 0.0005,
      "loss": 2.4673259258270264,
      "step": 4384
    },
    {
      "epoch": 0.707486285898677,
      "grad_norm": 1.7569888830184937,
      "learning_rate": 0.0005,
      "loss": 3.0361533164978027,
      "step": 4385
    },
    {
      "epoch": 0.707647628267183,
      "grad_norm": 1.1413837671279907,
      "learning_rate": 0.0005,
      "loss": 2.3370933532714844,
      "step": 4386
    },
    {
      "epoch": 0.7078089706356889,
      "grad_norm": 1.1901743412017822,
      "learning_rate": 0.0005,
      "loss": 2.8326783180236816,
      "step": 4387
    },
    {
      "epoch": 0.7079703130041949,
      "grad_norm": 0.8734435439109802,
      "learning_rate": 0.0005,
      "loss": 2.3167591094970703,
      "step": 4388
    },
    {
      "epoch": 0.7081316553727008,
      "grad_norm": 0.9891631007194519,
      "learning_rate": 0.0005,
      "loss": 2.9654788970947266,
      "step": 4389
    },
    {
      "epoch": 0.7082929977412068,
      "grad_norm": 1.4566751718521118,
      "learning_rate": 0.0005,
      "loss": 2.3441238403320312,
      "step": 4390
    },
    {
      "epoch": 0.7084543401097129,
      "grad_norm": 1.032167673110962,
      "learning_rate": 0.0005,
      "loss": 2.2255780696868896,
      "step": 4391
    },
    {
      "epoch": 0.7086156824782188,
      "grad_norm": 1.2554861307144165,
      "learning_rate": 0.0005,
      "loss": 2.768688440322876,
      "step": 4392
    },
    {
      "epoch": 0.7087770248467248,
      "grad_norm": 1.1005662679672241,
      "learning_rate": 0.0005,
      "loss": 2.5900144577026367,
      "step": 4393
    },
    {
      "epoch": 0.7089383672152307,
      "grad_norm": 1.3007231950759888,
      "learning_rate": 0.0005,
      "loss": 2.5088863372802734,
      "step": 4394
    },
    {
      "epoch": 0.7090997095837367,
      "grad_norm": 0.9609440565109253,
      "learning_rate": 0.0005,
      "loss": 2.7609057426452637,
      "step": 4395
    },
    {
      "epoch": 0.7092610519522426,
      "grad_norm": 8.172270774841309,
      "learning_rate": 0.0005,
      "loss": 2.938964366912842,
      "step": 4396
    },
    {
      "epoch": 0.7094223943207486,
      "grad_norm": 0.9589486122131348,
      "learning_rate": 0.0005,
      "loss": 2.5836310386657715,
      "step": 4397
    },
    {
      "epoch": 0.7095837366892546,
      "grad_norm": 1.1649447679519653,
      "learning_rate": 0.0005,
      "loss": 2.319976568222046,
      "step": 4398
    },
    {
      "epoch": 0.7097450790577605,
      "grad_norm": 1.4247794151306152,
      "learning_rate": 0.0005,
      "loss": 3.008103847503662,
      "step": 4399
    },
    {
      "epoch": 0.7099064214262666,
      "grad_norm": 0.9774237275123596,
      "learning_rate": 0.0005,
      "loss": 2.2675745487213135,
      "step": 4400
    },
    {
      "epoch": 0.7100677637947725,
      "grad_norm": 1.1849106550216675,
      "learning_rate": 0.0005,
      "loss": 2.590421199798584,
      "step": 4401
    },
    {
      "epoch": 0.7102291061632785,
      "grad_norm": 1.1061723232269287,
      "learning_rate": 0.0005,
      "loss": 2.641820192337036,
      "step": 4402
    },
    {
      "epoch": 0.7103904485317845,
      "grad_norm": 1.144736647605896,
      "learning_rate": 0.0005,
      "loss": 2.612569808959961,
      "step": 4403
    },
    {
      "epoch": 0.7105517909002904,
      "grad_norm": 1.0017132759094238,
      "learning_rate": 0.0005,
      "loss": 2.7093236446380615,
      "step": 4404
    },
    {
      "epoch": 0.7107131332687964,
      "grad_norm": 0.9502840638160706,
      "learning_rate": 0.0005,
      "loss": 2.4143049716949463,
      "step": 4405
    },
    {
      "epoch": 0.7108744756373023,
      "grad_norm": 1.7704797983169556,
      "learning_rate": 0.0005,
      "loss": 2.6315884590148926,
      "step": 4406
    },
    {
      "epoch": 0.7110358180058083,
      "grad_norm": 1.151017427444458,
      "learning_rate": 0.0005,
      "loss": 2.594808578491211,
      "step": 4407
    },
    {
      "epoch": 0.7111971603743142,
      "grad_norm": 1.301056981086731,
      "learning_rate": 0.0005,
      "loss": 2.477151393890381,
      "step": 4408
    },
    {
      "epoch": 0.7113585027428203,
      "grad_norm": 1.0889369249343872,
      "learning_rate": 0.0005,
      "loss": 2.1105329990386963,
      "step": 4409
    },
    {
      "epoch": 0.7115198451113263,
      "grad_norm": 0.9961727857589722,
      "learning_rate": 0.0005,
      "loss": 2.518589496612549,
      "step": 4410
    },
    {
      "epoch": 0.7116811874798322,
      "grad_norm": 1.242053508758545,
      "learning_rate": 0.0005,
      "loss": 2.569704532623291,
      "step": 4411
    },
    {
      "epoch": 0.7118425298483382,
      "grad_norm": 1.067924976348877,
      "learning_rate": 0.0005,
      "loss": 2.486783981323242,
      "step": 4412
    },
    {
      "epoch": 0.7120038722168441,
      "grad_norm": 1.3803553581237793,
      "learning_rate": 0.0005,
      "loss": 2.359555959701538,
      "step": 4413
    },
    {
      "epoch": 0.7121652145853501,
      "grad_norm": 3.207611083984375,
      "learning_rate": 0.0005,
      "loss": 2.567939519882202,
      "step": 4414
    },
    {
      "epoch": 0.712326556953856,
      "grad_norm": 1.404727816581726,
      "learning_rate": 0.0005,
      "loss": 2.7661585807800293,
      "step": 4415
    },
    {
      "epoch": 0.712487899322362,
      "grad_norm": 1.2601978778839111,
      "learning_rate": 0.0005,
      "loss": 2.3638675212860107,
      "step": 4416
    },
    {
      "epoch": 0.7126492416908681,
      "grad_norm": 1.713732123374939,
      "learning_rate": 0.0005,
      "loss": 2.267958641052246,
      "step": 4417
    },
    {
      "epoch": 0.712810584059374,
      "grad_norm": 1.210799217224121,
      "learning_rate": 0.0005,
      "loss": 2.5462534427642822,
      "step": 4418
    },
    {
      "epoch": 0.71297192642788,
      "grad_norm": 1.0716218948364258,
      "learning_rate": 0.0005,
      "loss": 2.5042381286621094,
      "step": 4419
    },
    {
      "epoch": 0.7131332687963859,
      "grad_norm": 1.2766340970993042,
      "learning_rate": 0.0005,
      "loss": 2.163595199584961,
      "step": 4420
    },
    {
      "epoch": 0.7132946111648919,
      "grad_norm": 1.1910885572433472,
      "learning_rate": 0.0005,
      "loss": 2.5287551879882812,
      "step": 4421
    },
    {
      "epoch": 0.7134559535333979,
      "grad_norm": 1.2059329748153687,
      "learning_rate": 0.0005,
      "loss": 2.691087245941162,
      "step": 4422
    },
    {
      "epoch": 0.7136172959019038,
      "grad_norm": 1.225374698638916,
      "learning_rate": 0.0005,
      "loss": 3.0742006301879883,
      "step": 4423
    },
    {
      "epoch": 0.7137786382704098,
      "grad_norm": 1.0861386060714722,
      "learning_rate": 0.0005,
      "loss": 2.553562641143799,
      "step": 4424
    },
    {
      "epoch": 0.7139399806389157,
      "grad_norm": 2.1374902725219727,
      "learning_rate": 0.0005,
      "loss": 2.4870145320892334,
      "step": 4425
    },
    {
      "epoch": 0.7141013230074218,
      "grad_norm": 1.1084740161895752,
      "learning_rate": 0.0005,
      "loss": 2.788289785385132,
      "step": 4426
    },
    {
      "epoch": 0.7142626653759278,
      "grad_norm": 1.3943181037902832,
      "learning_rate": 0.0005,
      "loss": 2.5039191246032715,
      "step": 4427
    },
    {
      "epoch": 0.7144240077444337,
      "grad_norm": 1.124120831489563,
      "learning_rate": 0.0005,
      "loss": 2.7952024936676025,
      "step": 4428
    },
    {
      "epoch": 0.7145853501129397,
      "grad_norm": 1.263365387916565,
      "learning_rate": 0.0005,
      "loss": 2.328505277633667,
      "step": 4429
    },
    {
      "epoch": 0.7147466924814456,
      "grad_norm": 1.160799503326416,
      "learning_rate": 0.0005,
      "loss": 2.6259284019470215,
      "step": 4430
    },
    {
      "epoch": 0.7149080348499516,
      "grad_norm": 1.1571753025054932,
      "learning_rate": 0.0005,
      "loss": 2.4968366622924805,
      "step": 4431
    },
    {
      "epoch": 0.7150693772184575,
      "grad_norm": 1.4355041980743408,
      "learning_rate": 0.0005,
      "loss": 2.2196872234344482,
      "step": 4432
    },
    {
      "epoch": 0.7152307195869635,
      "grad_norm": 1.5049070119857788,
      "learning_rate": 0.0005,
      "loss": 2.992917060852051,
      "step": 4433
    },
    {
      "epoch": 0.7153920619554696,
      "grad_norm": 1.330875277519226,
      "learning_rate": 0.0005,
      "loss": 2.775733470916748,
      "step": 4434
    },
    {
      "epoch": 0.7155534043239755,
      "grad_norm": 0.9694841504096985,
      "learning_rate": 0.0005,
      "loss": 2.619741201400757,
      "step": 4435
    },
    {
      "epoch": 0.7157147466924815,
      "grad_norm": 1.1943567991256714,
      "learning_rate": 0.0005,
      "loss": 2.3464105129241943,
      "step": 4436
    },
    {
      "epoch": 0.7158760890609874,
      "grad_norm": 1.4316891431808472,
      "learning_rate": 0.0005,
      "loss": 2.3405628204345703,
      "step": 4437
    },
    {
      "epoch": 0.7160374314294934,
      "grad_norm": 1.0300495624542236,
      "learning_rate": 0.0005,
      "loss": 2.5778648853302,
      "step": 4438
    },
    {
      "epoch": 0.7161987737979993,
      "grad_norm": 1.0075750350952148,
      "learning_rate": 0.0005,
      "loss": 2.341501235961914,
      "step": 4439
    },
    {
      "epoch": 0.7163601161665053,
      "grad_norm": 0.9390974640846252,
      "learning_rate": 0.0005,
      "loss": 2.5327978134155273,
      "step": 4440
    },
    {
      "epoch": 0.7165214585350113,
      "grad_norm": 1.2657616138458252,
      "learning_rate": 0.0005,
      "loss": 2.6763246059417725,
      "step": 4441
    },
    {
      "epoch": 0.7166828009035172,
      "grad_norm": 1.1136010885238647,
      "learning_rate": 0.0005,
      "loss": 2.468461275100708,
      "step": 4442
    },
    {
      "epoch": 0.7168441432720233,
      "grad_norm": 1.1692286729812622,
      "learning_rate": 0.0005,
      "loss": 2.5977423191070557,
      "step": 4443
    },
    {
      "epoch": 0.7170054856405292,
      "grad_norm": 1.3781143426895142,
      "learning_rate": 0.0005,
      "loss": 2.5545239448547363,
      "step": 4444
    },
    {
      "epoch": 0.7171668280090352,
      "grad_norm": 1.1836775541305542,
      "learning_rate": 0.0005,
      "loss": 2.594785213470459,
      "step": 4445
    },
    {
      "epoch": 0.7173281703775412,
      "grad_norm": 8.393026351928711,
      "learning_rate": 0.0005,
      "loss": 2.3285536766052246,
      "step": 4446
    },
    {
      "epoch": 0.7174895127460471,
      "grad_norm": 0.8657118678092957,
      "learning_rate": 0.0005,
      "loss": 1.906479835510254,
      "step": 4447
    },
    {
      "epoch": 0.7176508551145531,
      "grad_norm": 1.2284297943115234,
      "learning_rate": 0.0005,
      "loss": 2.575551748275757,
      "step": 4448
    },
    {
      "epoch": 0.717812197483059,
      "grad_norm": 1.2617896795272827,
      "learning_rate": 0.0005,
      "loss": 2.600644826889038,
      "step": 4449
    },
    {
      "epoch": 0.717973539851565,
      "grad_norm": 1.2861782312393188,
      "learning_rate": 0.0005,
      "loss": 2.8103065490722656,
      "step": 4450
    },
    {
      "epoch": 0.7181348822200709,
      "grad_norm": 1.0600767135620117,
      "learning_rate": 0.0005,
      "loss": 2.8866286277770996,
      "step": 4451
    },
    {
      "epoch": 0.718296224588577,
      "grad_norm": 0.9962307214736938,
      "learning_rate": 0.0005,
      "loss": 2.5146262645721436,
      "step": 4452
    },
    {
      "epoch": 0.718457566957083,
      "grad_norm": 1.0117932558059692,
      "learning_rate": 0.0005,
      "loss": 2.608112335205078,
      "step": 4453
    },
    {
      "epoch": 0.7186189093255889,
      "grad_norm": 1.090909719467163,
      "learning_rate": 0.0005,
      "loss": 2.2799227237701416,
      "step": 4454
    },
    {
      "epoch": 0.7187802516940949,
      "grad_norm": 1.2781370878219604,
      "learning_rate": 0.0005,
      "loss": 2.50673508644104,
      "step": 4455
    },
    {
      "epoch": 0.7189415940626008,
      "grad_norm": 1.0728954076766968,
      "learning_rate": 0.0005,
      "loss": 2.6609559059143066,
      "step": 4456
    },
    {
      "epoch": 0.7191029364311068,
      "grad_norm": 1.269737958908081,
      "learning_rate": 0.0005,
      "loss": 2.6054513454437256,
      "step": 4457
    },
    {
      "epoch": 0.7192642787996127,
      "grad_norm": 0.9994696974754333,
      "learning_rate": 0.0005,
      "loss": 2.1944403648376465,
      "step": 4458
    },
    {
      "epoch": 0.7194256211681187,
      "grad_norm": 1.2147940397262573,
      "learning_rate": 0.0005,
      "loss": 2.4864768981933594,
      "step": 4459
    },
    {
      "epoch": 0.7195869635366248,
      "grad_norm": 1.174887776374817,
      "learning_rate": 0.0005,
      "loss": 2.8600003719329834,
      "step": 4460
    },
    {
      "epoch": 0.7197483059051307,
      "grad_norm": 1.0378109216690063,
      "learning_rate": 0.0005,
      "loss": 2.6651504039764404,
      "step": 4461
    },
    {
      "epoch": 0.7199096482736367,
      "grad_norm": 1.2350982427597046,
      "learning_rate": 0.0005,
      "loss": 2.1273794174194336,
      "step": 4462
    },
    {
      "epoch": 0.7200709906421426,
      "grad_norm": 2.988586664199829,
      "learning_rate": 0.0005,
      "loss": 2.861400604248047,
      "step": 4463
    },
    {
      "epoch": 0.7202323330106486,
      "grad_norm": 1.108393907546997,
      "learning_rate": 0.0005,
      "loss": 2.8704586029052734,
      "step": 4464
    },
    {
      "epoch": 0.7203936753791546,
      "grad_norm": 1.4910640716552734,
      "learning_rate": 0.0005,
      "loss": 2.354534149169922,
      "step": 4465
    },
    {
      "epoch": 0.7205550177476605,
      "grad_norm": 1.140335202217102,
      "learning_rate": 0.0005,
      "loss": 2.4543399810791016,
      "step": 4466
    },
    {
      "epoch": 0.7207163601161665,
      "grad_norm": 1.252384066581726,
      "learning_rate": 0.0005,
      "loss": 2.3646039962768555,
      "step": 4467
    },
    {
      "epoch": 0.7208777024846724,
      "grad_norm": 1.2848896980285645,
      "learning_rate": 0.0005,
      "loss": 2.3539254665374756,
      "step": 4468
    },
    {
      "epoch": 0.7210390448531785,
      "grad_norm": 1.7579725980758667,
      "learning_rate": 0.0005,
      "loss": 2.67490816116333,
      "step": 4469
    },
    {
      "epoch": 0.7212003872216844,
      "grad_norm": 1.5110617876052856,
      "learning_rate": 0.0005,
      "loss": 2.5532164573669434,
      "step": 4470
    },
    {
      "epoch": 0.7213617295901904,
      "grad_norm": 1.7962831258773804,
      "learning_rate": 0.0005,
      "loss": 2.3453898429870605,
      "step": 4471
    },
    {
      "epoch": 0.7215230719586964,
      "grad_norm": 1.1453977823257446,
      "learning_rate": 0.0005,
      "loss": 2.6174237728118896,
      "step": 4472
    },
    {
      "epoch": 0.7216844143272023,
      "grad_norm": 1.0474916696548462,
      "learning_rate": 0.0005,
      "loss": 2.6911840438842773,
      "step": 4473
    },
    {
      "epoch": 0.7218457566957083,
      "grad_norm": 1.0030232667922974,
      "learning_rate": 0.0005,
      "loss": 2.544907331466675,
      "step": 4474
    },
    {
      "epoch": 0.7220070990642142,
      "grad_norm": 1.3450770378112793,
      "learning_rate": 0.0005,
      "loss": 3.139037609100342,
      "step": 4475
    },
    {
      "epoch": 0.7221684414327202,
      "grad_norm": 1.295335292816162,
      "learning_rate": 0.0005,
      "loss": 2.645327568054199,
      "step": 4476
    },
    {
      "epoch": 0.7223297838012263,
      "grad_norm": 1.0745964050292969,
      "learning_rate": 0.0005,
      "loss": 2.428286552429199,
      "step": 4477
    },
    {
      "epoch": 0.7224911261697322,
      "grad_norm": 1.002286434173584,
      "learning_rate": 0.0005,
      "loss": 2.496696710586548,
      "step": 4478
    },
    {
      "epoch": 0.7226524685382382,
      "grad_norm": 1.9381272792816162,
      "learning_rate": 0.0005,
      "loss": 2.110726833343506,
      "step": 4479
    },
    {
      "epoch": 0.7228138109067441,
      "grad_norm": 1.2204939126968384,
      "learning_rate": 0.0005,
      "loss": 2.512110710144043,
      "step": 4480
    },
    {
      "epoch": 0.7229751532752501,
      "grad_norm": 1.0795278549194336,
      "learning_rate": 0.0005,
      "loss": 2.5424435138702393,
      "step": 4481
    },
    {
      "epoch": 0.723136495643756,
      "grad_norm": 1.1036419868469238,
      "learning_rate": 0.0005,
      "loss": 2.450794219970703,
      "step": 4482
    },
    {
      "epoch": 0.723297838012262,
      "grad_norm": 1.2363778352737427,
      "learning_rate": 0.0005,
      "loss": 2.489701747894287,
      "step": 4483
    },
    {
      "epoch": 0.723459180380768,
      "grad_norm": 26.25476837158203,
      "learning_rate": 0.0005,
      "loss": 2.7171502113342285,
      "step": 4484
    },
    {
      "epoch": 0.7236205227492739,
      "grad_norm": 0.9643695950508118,
      "learning_rate": 0.0005,
      "loss": 2.7924346923828125,
      "step": 4485
    },
    {
      "epoch": 0.72378186511778,
      "grad_norm": 0.8446159362792969,
      "learning_rate": 0.0005,
      "loss": 2.3074357509613037,
      "step": 4486
    },
    {
      "epoch": 0.7239432074862859,
      "grad_norm": 1.3652105331420898,
      "learning_rate": 0.0005,
      "loss": 2.303316593170166,
      "step": 4487
    },
    {
      "epoch": 0.7241045498547919,
      "grad_norm": 1.1791143417358398,
      "learning_rate": 0.0005,
      "loss": 2.354280948638916,
      "step": 4488
    },
    {
      "epoch": 0.7242658922232978,
      "grad_norm": 1.0244773626327515,
      "learning_rate": 0.0005,
      "loss": 2.8963489532470703,
      "step": 4489
    },
    {
      "epoch": 0.7244272345918038,
      "grad_norm": 1.4277665615081787,
      "learning_rate": 0.0005,
      "loss": 2.565002202987671,
      "step": 4490
    },
    {
      "epoch": 0.7245885769603098,
      "grad_norm": 0.9646983742713928,
      "learning_rate": 0.0005,
      "loss": 2.6272683143615723,
      "step": 4491
    },
    {
      "epoch": 0.7247499193288157,
      "grad_norm": 1.621889591217041,
      "learning_rate": 0.0005,
      "loss": 2.5768401622772217,
      "step": 4492
    },
    {
      "epoch": 0.7249112616973217,
      "grad_norm": 1.2631986141204834,
      "learning_rate": 0.0005,
      "loss": 2.2992000579833984,
      "step": 4493
    },
    {
      "epoch": 0.7250726040658277,
      "grad_norm": 1.0669809579849243,
      "learning_rate": 0.0005,
      "loss": 2.6821537017822266,
      "step": 4494
    },
    {
      "epoch": 0.7252339464343337,
      "grad_norm": 1.4297970533370972,
      "learning_rate": 0.0005,
      "loss": 2.9611480236053467,
      "step": 4495
    },
    {
      "epoch": 0.7253952888028397,
      "grad_norm": 1.1419013738632202,
      "learning_rate": 0.0005,
      "loss": 2.5228540897369385,
      "step": 4496
    },
    {
      "epoch": 0.7255566311713456,
      "grad_norm": 0.958114504814148,
      "learning_rate": 0.0005,
      "loss": 2.196169137954712,
      "step": 4497
    },
    {
      "epoch": 0.7257179735398516,
      "grad_norm": 0.9943202137947083,
      "learning_rate": 0.0005,
      "loss": 2.54752254486084,
      "step": 4498
    },
    {
      "epoch": 0.7258793159083575,
      "grad_norm": 0.9525582194328308,
      "learning_rate": 0.0005,
      "loss": 2.186842679977417,
      "step": 4499
    },
    {
      "epoch": 0.7260406582768635,
      "grad_norm": 1.521873116493225,
      "learning_rate": 0.0005,
      "loss": 3.0647010803222656,
      "step": 4500
    },
    {
      "epoch": 0.7262020006453694,
      "grad_norm": 0.9748245477676392,
      "learning_rate": 0.0005,
      "loss": 2.649048089981079,
      "step": 4501
    },
    {
      "epoch": 0.7263633430138754,
      "grad_norm": 1.2296136617660522,
      "learning_rate": 0.0005,
      "loss": 2.575206995010376,
      "step": 4502
    },
    {
      "epoch": 0.7265246853823815,
      "grad_norm": 1.308455467224121,
      "learning_rate": 0.0005,
      "loss": 2.800516128540039,
      "step": 4503
    },
    {
      "epoch": 0.7266860277508874,
      "grad_norm": 0.9843746423721313,
      "learning_rate": 0.0005,
      "loss": 2.7865893840789795,
      "step": 4504
    },
    {
      "epoch": 0.7268473701193934,
      "grad_norm": 0.9618057012557983,
      "learning_rate": 0.0005,
      "loss": 2.860959529876709,
      "step": 4505
    },
    {
      "epoch": 0.7270087124878993,
      "grad_norm": 1.0118284225463867,
      "learning_rate": 0.0005,
      "loss": 2.5936994552612305,
      "step": 4506
    },
    {
      "epoch": 0.7271700548564053,
      "grad_norm": 1.0502946376800537,
      "learning_rate": 0.0005,
      "loss": 2.702277660369873,
      "step": 4507
    },
    {
      "epoch": 0.7273313972249112,
      "grad_norm": 1.674452543258667,
      "learning_rate": 0.0005,
      "loss": 2.452205181121826,
      "step": 4508
    },
    {
      "epoch": 0.7274927395934172,
      "grad_norm": 1.8551321029663086,
      "learning_rate": 0.0005,
      "loss": 2.537290573120117,
      "step": 4509
    },
    {
      "epoch": 0.7276540819619232,
      "grad_norm": 1.1987664699554443,
      "learning_rate": 0.0005,
      "loss": 2.5958855152130127,
      "step": 4510
    },
    {
      "epoch": 0.7278154243304291,
      "grad_norm": 1.656888484954834,
      "learning_rate": 0.0005,
      "loss": 2.8197555541992188,
      "step": 4511
    },
    {
      "epoch": 0.7279767666989352,
      "grad_norm": 1.8383605480194092,
      "learning_rate": 0.0005,
      "loss": 2.3401451110839844,
      "step": 4512
    },
    {
      "epoch": 0.7281381090674411,
      "grad_norm": 1.0263527631759644,
      "learning_rate": 0.0005,
      "loss": 2.295545816421509,
      "step": 4513
    },
    {
      "epoch": 0.7282994514359471,
      "grad_norm": 1.1139893531799316,
      "learning_rate": 0.0005,
      "loss": 2.695981502532959,
      "step": 4514
    },
    {
      "epoch": 0.728460793804453,
      "grad_norm": 1.0003726482391357,
      "learning_rate": 0.0005,
      "loss": 2.8447418212890625,
      "step": 4515
    },
    {
      "epoch": 0.728622136172959,
      "grad_norm": 3.5667853355407715,
      "learning_rate": 0.0005,
      "loss": 2.263812780380249,
      "step": 4516
    },
    {
      "epoch": 0.728783478541465,
      "grad_norm": 2.4892921447753906,
      "learning_rate": 0.0005,
      "loss": 2.637979507446289,
      "step": 4517
    },
    {
      "epoch": 0.7289448209099709,
      "grad_norm": 1.9523582458496094,
      "learning_rate": 0.0005,
      "loss": 3.1211774349212646,
      "step": 4518
    },
    {
      "epoch": 0.7291061632784769,
      "grad_norm": 1.1098809242248535,
      "learning_rate": 0.0005,
      "loss": 3.0742204189300537,
      "step": 4519
    },
    {
      "epoch": 0.729267505646983,
      "grad_norm": 1.104744553565979,
      "learning_rate": 0.0005,
      "loss": 2.632260322570801,
      "step": 4520
    },
    {
      "epoch": 0.7294288480154889,
      "grad_norm": 1.1853890419006348,
      "learning_rate": 0.0005,
      "loss": 2.1110141277313232,
      "step": 4521
    },
    {
      "epoch": 0.7295901903839949,
      "grad_norm": 1.464074969291687,
      "learning_rate": 0.0005,
      "loss": 2.8582029342651367,
      "step": 4522
    },
    {
      "epoch": 0.7297515327525008,
      "grad_norm": 1.1271202564239502,
      "learning_rate": 0.0005,
      "loss": 2.6396985054016113,
      "step": 4523
    },
    {
      "epoch": 0.7299128751210068,
      "grad_norm": 0.9445173740386963,
      "learning_rate": 0.0005,
      "loss": 2.329472541809082,
      "step": 4524
    },
    {
      "epoch": 0.7300742174895127,
      "grad_norm": 0.997973620891571,
      "learning_rate": 0.0005,
      "loss": 2.563516855239868,
      "step": 4525
    },
    {
      "epoch": 0.7302355598580187,
      "grad_norm": 1.3795801401138306,
      "learning_rate": 0.0005,
      "loss": 2.6935596466064453,
      "step": 4526
    },
    {
      "epoch": 0.7303969022265246,
      "grad_norm": 1.0983023643493652,
      "learning_rate": 0.0005,
      "loss": 2.515834331512451,
      "step": 4527
    },
    {
      "epoch": 0.7305582445950306,
      "grad_norm": 1.2840576171875,
      "learning_rate": 0.0005,
      "loss": 2.8436264991760254,
      "step": 4528
    },
    {
      "epoch": 0.7307195869635367,
      "grad_norm": 1.3650853633880615,
      "learning_rate": 0.0005,
      "loss": 2.384472370147705,
      "step": 4529
    },
    {
      "epoch": 0.7308809293320426,
      "grad_norm": 1.0387135744094849,
      "learning_rate": 0.0005,
      "loss": 3.011195182800293,
      "step": 4530
    },
    {
      "epoch": 0.7310422717005486,
      "grad_norm": 1.256974458694458,
      "learning_rate": 0.0005,
      "loss": 2.7315444946289062,
      "step": 4531
    },
    {
      "epoch": 0.7312036140690545,
      "grad_norm": 1.005070447921753,
      "learning_rate": 0.0005,
      "loss": 2.3653275966644287,
      "step": 4532
    },
    {
      "epoch": 0.7313649564375605,
      "grad_norm": 1.2695863246917725,
      "learning_rate": 0.0005,
      "loss": 2.6719627380371094,
      "step": 4533
    },
    {
      "epoch": 0.7315262988060665,
      "grad_norm": 1.5402860641479492,
      "learning_rate": 0.0005,
      "loss": 2.650974750518799,
      "step": 4534
    },
    {
      "epoch": 0.7316876411745724,
      "grad_norm": 1.2958519458770752,
      "learning_rate": 0.0005,
      "loss": 2.640206813812256,
      "step": 4535
    },
    {
      "epoch": 0.7318489835430784,
      "grad_norm": 1.2829241752624512,
      "learning_rate": 0.0005,
      "loss": 2.7083632946014404,
      "step": 4536
    },
    {
      "epoch": 0.7320103259115844,
      "grad_norm": 1.1423546075820923,
      "learning_rate": 0.0005,
      "loss": 2.347456932067871,
      "step": 4537
    },
    {
      "epoch": 0.7321716682800904,
      "grad_norm": 1.1124229431152344,
      "learning_rate": 0.0005,
      "loss": 2.526492118835449,
      "step": 4538
    },
    {
      "epoch": 0.7323330106485963,
      "grad_norm": 1.4760165214538574,
      "learning_rate": 0.0005,
      "loss": 2.7234725952148438,
      "step": 4539
    },
    {
      "epoch": 0.7324943530171023,
      "grad_norm": 1.707310676574707,
      "learning_rate": 0.0005,
      "loss": 3.3656814098358154,
      "step": 4540
    },
    {
      "epoch": 0.7326556953856083,
      "grad_norm": 1.026191234588623,
      "learning_rate": 0.0005,
      "loss": 2.352813959121704,
      "step": 4541
    },
    {
      "epoch": 0.7328170377541142,
      "grad_norm": 1.1068063974380493,
      "learning_rate": 0.0005,
      "loss": 2.6281018257141113,
      "step": 4542
    },
    {
      "epoch": 0.7329783801226202,
      "grad_norm": 1.6518261432647705,
      "learning_rate": 0.0005,
      "loss": 3.0958590507507324,
      "step": 4543
    },
    {
      "epoch": 0.7331397224911261,
      "grad_norm": 1.0908405780792236,
      "learning_rate": 0.0005,
      "loss": 2.79177188873291,
      "step": 4544
    },
    {
      "epoch": 0.7333010648596321,
      "grad_norm": 1.15807044506073,
      "learning_rate": 0.0005,
      "loss": 2.386673927307129,
      "step": 4545
    },
    {
      "epoch": 0.7334624072281382,
      "grad_norm": 1.1061997413635254,
      "learning_rate": 0.0005,
      "loss": 2.382564067840576,
      "step": 4546
    },
    {
      "epoch": 0.7336237495966441,
      "grad_norm": 1.3178789615631104,
      "learning_rate": 0.0005,
      "loss": 2.797508716583252,
      "step": 4547
    },
    {
      "epoch": 0.7337850919651501,
      "grad_norm": 1.064788579940796,
      "learning_rate": 0.0005,
      "loss": 2.964979410171509,
      "step": 4548
    },
    {
      "epoch": 0.733946434333656,
      "grad_norm": 1.202730417251587,
      "learning_rate": 0.0005,
      "loss": 2.379718780517578,
      "step": 4549
    },
    {
      "epoch": 0.734107776702162,
      "grad_norm": 1.2969143390655518,
      "learning_rate": 0.0005,
      "loss": 2.7281482219696045,
      "step": 4550
    },
    {
      "epoch": 0.7342691190706679,
      "grad_norm": 1.1314321756362915,
      "learning_rate": 0.0005,
      "loss": 2.681438446044922,
      "step": 4551
    },
    {
      "epoch": 0.7344304614391739,
      "grad_norm": 8.534379005432129,
      "learning_rate": 0.0005,
      "loss": 2.8242435455322266,
      "step": 4552
    },
    {
      "epoch": 0.7345918038076799,
      "grad_norm": 1.4117604494094849,
      "learning_rate": 0.0005,
      "loss": 2.6479954719543457,
      "step": 4553
    },
    {
      "epoch": 0.7347531461761858,
      "grad_norm": 1.3265609741210938,
      "learning_rate": 0.0005,
      "loss": 2.385901927947998,
      "step": 4554
    },
    {
      "epoch": 0.7349144885446919,
      "grad_norm": 1.2216260433197021,
      "learning_rate": 0.0005,
      "loss": 2.4969072341918945,
      "step": 4555
    },
    {
      "epoch": 0.7350758309131978,
      "grad_norm": 1.0161995887756348,
      "learning_rate": 0.0005,
      "loss": 2.8797473907470703,
      "step": 4556
    },
    {
      "epoch": 0.7352371732817038,
      "grad_norm": 1.180174708366394,
      "learning_rate": 0.0005,
      "loss": 2.484074115753174,
      "step": 4557
    },
    {
      "epoch": 0.7353985156502097,
      "grad_norm": 1.6457598209381104,
      "learning_rate": 0.0005,
      "loss": 3.042921781539917,
      "step": 4558
    },
    {
      "epoch": 0.7355598580187157,
      "grad_norm": 1.0869932174682617,
      "learning_rate": 0.0005,
      "loss": 2.5027565956115723,
      "step": 4559
    },
    {
      "epoch": 0.7357212003872217,
      "grad_norm": 0.93024080991745,
      "learning_rate": 0.0005,
      "loss": 2.556893825531006,
      "step": 4560
    },
    {
      "epoch": 0.7358825427557276,
      "grad_norm": 0.9589362144470215,
      "learning_rate": 0.0005,
      "loss": 2.47310209274292,
      "step": 4561
    },
    {
      "epoch": 0.7360438851242336,
      "grad_norm": 2.5812368392944336,
      "learning_rate": 0.0005,
      "loss": 2.4251174926757812,
      "step": 4562
    },
    {
      "epoch": 0.7362052274927396,
      "grad_norm": 1.241429090499878,
      "learning_rate": 0.0005,
      "loss": 2.7724432945251465,
      "step": 4563
    },
    {
      "epoch": 0.7363665698612456,
      "grad_norm": 1.2309257984161377,
      "learning_rate": 0.0005,
      "loss": 2.8199493885040283,
      "step": 4564
    },
    {
      "epoch": 0.7365279122297516,
      "grad_norm": 1.0173430442810059,
      "learning_rate": 0.0005,
      "loss": 2.286975860595703,
      "step": 4565
    },
    {
      "epoch": 0.7366892545982575,
      "grad_norm": 1.5592281818389893,
      "learning_rate": 0.0005,
      "loss": 2.851632595062256,
      "step": 4566
    },
    {
      "epoch": 0.7368505969667635,
      "grad_norm": 1.423831820487976,
      "learning_rate": 0.0005,
      "loss": 3.062488317489624,
      "step": 4567
    },
    {
      "epoch": 0.7370119393352694,
      "grad_norm": 1.482826828956604,
      "learning_rate": 0.0005,
      "loss": 2.2502169609069824,
      "step": 4568
    },
    {
      "epoch": 0.7371732817037754,
      "grad_norm": 1.3937844038009644,
      "learning_rate": 0.0005,
      "loss": 2.6399283409118652,
      "step": 4569
    },
    {
      "epoch": 0.7373346240722813,
      "grad_norm": 1.165657639503479,
      "learning_rate": 0.0005,
      "loss": 2.8740899562835693,
      "step": 4570
    },
    {
      "epoch": 0.7374959664407873,
      "grad_norm": 1.320839762687683,
      "learning_rate": 0.0005,
      "loss": 2.899916648864746,
      "step": 4571
    },
    {
      "epoch": 0.7376573088092934,
      "grad_norm": 1.0172351598739624,
      "learning_rate": 0.0005,
      "loss": 2.6505661010742188,
      "step": 4572
    },
    {
      "epoch": 0.7378186511777993,
      "grad_norm": 1.9308903217315674,
      "learning_rate": 0.0005,
      "loss": 2.8144521713256836,
      "step": 4573
    },
    {
      "epoch": 0.7379799935463053,
      "grad_norm": 1.058741807937622,
      "learning_rate": 0.0005,
      "loss": 2.571256160736084,
      "step": 4574
    },
    {
      "epoch": 0.7381413359148112,
      "grad_norm": 1.2805368900299072,
      "learning_rate": 0.0005,
      "loss": 2.4702346324920654,
      "step": 4575
    },
    {
      "epoch": 0.7383026782833172,
      "grad_norm": 1.1097887754440308,
      "learning_rate": 0.0005,
      "loss": 2.705995559692383,
      "step": 4576
    },
    {
      "epoch": 0.7384640206518231,
      "grad_norm": 1.2300785779953003,
      "learning_rate": 0.0005,
      "loss": 2.7748866081237793,
      "step": 4577
    },
    {
      "epoch": 0.7386253630203291,
      "grad_norm": 1.1103532314300537,
      "learning_rate": 0.0005,
      "loss": 2.553804636001587,
      "step": 4578
    },
    {
      "epoch": 0.7387867053888351,
      "grad_norm": 1.3594739437103271,
      "learning_rate": 0.0005,
      "loss": 2.335552930831909,
      "step": 4579
    },
    {
      "epoch": 0.7389480477573411,
      "grad_norm": 1.104950189590454,
      "learning_rate": 0.0005,
      "loss": 2.476191282272339,
      "step": 4580
    },
    {
      "epoch": 0.7391093901258471,
      "grad_norm": 1.0547404289245605,
      "learning_rate": 0.0005,
      "loss": 2.5774261951446533,
      "step": 4581
    },
    {
      "epoch": 0.739270732494353,
      "grad_norm": 1.298581600189209,
      "learning_rate": 0.0005,
      "loss": 2.7228541374206543,
      "step": 4582
    },
    {
      "epoch": 0.739432074862859,
      "grad_norm": 1.2042220830917358,
      "learning_rate": 0.0005,
      "loss": 2.5044796466827393,
      "step": 4583
    },
    {
      "epoch": 0.739593417231365,
      "grad_norm": 1.0614309310913086,
      "learning_rate": 0.0005,
      "loss": 2.8533387184143066,
      "step": 4584
    },
    {
      "epoch": 0.7397547595998709,
      "grad_norm": 1.1529269218444824,
      "learning_rate": 0.0005,
      "loss": 2.9042868614196777,
      "step": 4585
    },
    {
      "epoch": 0.7399161019683769,
      "grad_norm": 1.2624951601028442,
      "learning_rate": 0.0005,
      "loss": 2.6640567779541016,
      "step": 4586
    },
    {
      "epoch": 0.7400774443368828,
      "grad_norm": 1.136612892150879,
      "learning_rate": 0.0005,
      "loss": 2.2834415435791016,
      "step": 4587
    },
    {
      "epoch": 0.7402387867053888,
      "grad_norm": 1.1962237358093262,
      "learning_rate": 0.0005,
      "loss": 2.553440570831299,
      "step": 4588
    },
    {
      "epoch": 0.7404001290738949,
      "grad_norm": 1.1063861846923828,
      "learning_rate": 0.0005,
      "loss": 2.3658432960510254,
      "step": 4589
    },
    {
      "epoch": 0.7405614714424008,
      "grad_norm": 1.0182466506958008,
      "learning_rate": 0.0005,
      "loss": 2.648996353149414,
      "step": 4590
    },
    {
      "epoch": 0.7407228138109068,
      "grad_norm": 1.1191602945327759,
      "learning_rate": 0.0005,
      "loss": 2.9223008155822754,
      "step": 4591
    },
    {
      "epoch": 0.7408841561794127,
      "grad_norm": 0.9502724409103394,
      "learning_rate": 0.0005,
      "loss": 2.8100602626800537,
      "step": 4592
    },
    {
      "epoch": 0.7410454985479187,
      "grad_norm": 1.3151593208312988,
      "learning_rate": 0.0005,
      "loss": 2.115626096725464,
      "step": 4593
    },
    {
      "epoch": 0.7412068409164246,
      "grad_norm": 1.2198396921157837,
      "learning_rate": 0.0005,
      "loss": 2.3262007236480713,
      "step": 4594
    },
    {
      "epoch": 0.7413681832849306,
      "grad_norm": 1.7159956693649292,
      "learning_rate": 0.0005,
      "loss": 2.819244146347046,
      "step": 4595
    },
    {
      "epoch": 0.7415295256534365,
      "grad_norm": 1.0852477550506592,
      "learning_rate": 0.0005,
      "loss": 2.442269802093506,
      "step": 4596
    },
    {
      "epoch": 0.7416908680219426,
      "grad_norm": 0.9092139601707458,
      "learning_rate": 0.0005,
      "loss": 1.96380615234375,
      "step": 4597
    },
    {
      "epoch": 0.7418522103904486,
      "grad_norm": 2.3747456073760986,
      "learning_rate": 0.0005,
      "loss": 3.090489387512207,
      "step": 4598
    },
    {
      "epoch": 0.7420135527589545,
      "grad_norm": 1.241660475730896,
      "learning_rate": 0.0005,
      "loss": 2.6464991569519043,
      "step": 4599
    },
    {
      "epoch": 0.7421748951274605,
      "grad_norm": 1.2880886793136597,
      "learning_rate": 0.0005,
      "loss": 2.713701009750366,
      "step": 4600
    },
    {
      "epoch": 0.7423362374959664,
      "grad_norm": 1.2932302951812744,
      "learning_rate": 0.0005,
      "loss": 2.69221830368042,
      "step": 4601
    },
    {
      "epoch": 0.7424975798644724,
      "grad_norm": 1.3056490421295166,
      "learning_rate": 0.0005,
      "loss": 2.5498085021972656,
      "step": 4602
    },
    {
      "epoch": 0.7426589222329784,
      "grad_norm": 0.938247561454773,
      "learning_rate": 0.0005,
      "loss": 2.2783875465393066,
      "step": 4603
    },
    {
      "epoch": 0.7428202646014843,
      "grad_norm": 1.5958126783370972,
      "learning_rate": 0.0005,
      "loss": 2.3632664680480957,
      "step": 4604
    },
    {
      "epoch": 0.7429816069699903,
      "grad_norm": 1.35136878490448,
      "learning_rate": 0.0005,
      "loss": 3.063037395477295,
      "step": 4605
    },
    {
      "epoch": 0.7431429493384963,
      "grad_norm": 1.1173193454742432,
      "learning_rate": 0.0005,
      "loss": 2.8066632747650146,
      "step": 4606
    },
    {
      "epoch": 0.7433042917070023,
      "grad_norm": 1.5255686044692993,
      "learning_rate": 0.0005,
      "loss": 2.1422171592712402,
      "step": 4607
    },
    {
      "epoch": 0.7434656340755083,
      "grad_norm": 1.105491042137146,
      "learning_rate": 0.0005,
      "loss": 2.689819812774658,
      "step": 4608
    },
    {
      "epoch": 0.7436269764440142,
      "grad_norm": 1.9400352239608765,
      "learning_rate": 0.0005,
      "loss": 2.695361614227295,
      "step": 4609
    },
    {
      "epoch": 0.7437883188125202,
      "grad_norm": 1.5192738771438599,
      "learning_rate": 0.0005,
      "loss": 2.7311370372772217,
      "step": 4610
    },
    {
      "epoch": 0.7439496611810261,
      "grad_norm": 1.3157992362976074,
      "learning_rate": 0.0005,
      "loss": 2.746326446533203,
      "step": 4611
    },
    {
      "epoch": 0.7441110035495321,
      "grad_norm": 1.2452561855316162,
      "learning_rate": 0.0005,
      "loss": 2.343677520751953,
      "step": 4612
    },
    {
      "epoch": 0.744272345918038,
      "grad_norm": 1.2366795539855957,
      "learning_rate": 0.0005,
      "loss": 2.521847724914551,
      "step": 4613
    },
    {
      "epoch": 0.744433688286544,
      "grad_norm": 1.0773106813430786,
      "learning_rate": 0.0005,
      "loss": 2.648914337158203,
      "step": 4614
    },
    {
      "epoch": 0.7445950306550501,
      "grad_norm": 1.4594192504882812,
      "learning_rate": 0.0005,
      "loss": 2.681828022003174,
      "step": 4615
    },
    {
      "epoch": 0.744756373023556,
      "grad_norm": 1.1591590642929077,
      "learning_rate": 0.0005,
      "loss": 2.6994035243988037,
      "step": 4616
    },
    {
      "epoch": 0.744917715392062,
      "grad_norm": 1.7019168138504028,
      "learning_rate": 0.0005,
      "loss": 2.903942108154297,
      "step": 4617
    },
    {
      "epoch": 0.7450790577605679,
      "grad_norm": 1.3810607194900513,
      "learning_rate": 0.0005,
      "loss": 2.6026642322540283,
      "step": 4618
    },
    {
      "epoch": 0.7452404001290739,
      "grad_norm": 1.2670550346374512,
      "learning_rate": 0.0005,
      "loss": 2.774149179458618,
      "step": 4619
    },
    {
      "epoch": 0.7454017424975798,
      "grad_norm": 1.1739633083343506,
      "learning_rate": 0.0005,
      "loss": 2.7148818969726562,
      "step": 4620
    },
    {
      "epoch": 0.7455630848660858,
      "grad_norm": 1.5037639141082764,
      "learning_rate": 0.0005,
      "loss": 2.4998250007629395,
      "step": 4621
    },
    {
      "epoch": 0.7457244272345918,
      "grad_norm": 1.1347709894180298,
      "learning_rate": 0.0005,
      "loss": 2.474877119064331,
      "step": 4622
    },
    {
      "epoch": 0.7458857696030978,
      "grad_norm": 1.6644691228866577,
      "learning_rate": 0.0005,
      "loss": 2.7385778427124023,
      "step": 4623
    },
    {
      "epoch": 0.7460471119716038,
      "grad_norm": 1.6678813695907593,
      "learning_rate": 0.0005,
      "loss": 2.1250667572021484,
      "step": 4624
    },
    {
      "epoch": 0.7462084543401097,
      "grad_norm": 1.1665654182434082,
      "learning_rate": 0.0005,
      "loss": 2.609761953353882,
      "step": 4625
    },
    {
      "epoch": 0.7463697967086157,
      "grad_norm": 0.9347933530807495,
      "learning_rate": 0.0005,
      "loss": 2.711853504180908,
      "step": 4626
    },
    {
      "epoch": 0.7465311390771217,
      "grad_norm": 1.0853419303894043,
      "learning_rate": 0.0005,
      "loss": 2.232111930847168,
      "step": 4627
    },
    {
      "epoch": 0.7466924814456276,
      "grad_norm": 1.0398378372192383,
      "learning_rate": 0.0005,
      "loss": 2.248166799545288,
      "step": 4628
    },
    {
      "epoch": 0.7468538238141336,
      "grad_norm": 0.9581737518310547,
      "learning_rate": 0.0005,
      "loss": 1.8096139430999756,
      "step": 4629
    },
    {
      "epoch": 0.7470151661826395,
      "grad_norm": 1.3091117143630981,
      "learning_rate": 0.0005,
      "loss": 2.397829532623291,
      "step": 4630
    },
    {
      "epoch": 0.7471765085511455,
      "grad_norm": 1.0077519416809082,
      "learning_rate": 0.0005,
      "loss": 2.5898962020874023,
      "step": 4631
    },
    {
      "epoch": 0.7473378509196515,
      "grad_norm": 1.1488274335861206,
      "learning_rate": 0.0005,
      "loss": 2.781865358352661,
      "step": 4632
    },
    {
      "epoch": 0.7474991932881575,
      "grad_norm": 1.1716150045394897,
      "learning_rate": 0.0005,
      "loss": 2.908708095550537,
      "step": 4633
    },
    {
      "epoch": 0.7476605356566635,
      "grad_norm": 1.2520091533660889,
      "learning_rate": 0.0005,
      "loss": 2.286212921142578,
      "step": 4634
    },
    {
      "epoch": 0.7478218780251694,
      "grad_norm": 1.427263855934143,
      "learning_rate": 0.0005,
      "loss": 3.3100109100341797,
      "step": 4635
    },
    {
      "epoch": 0.7479832203936754,
      "grad_norm": 1.017006754875183,
      "learning_rate": 0.0005,
      "loss": 2.5754003524780273,
      "step": 4636
    },
    {
      "epoch": 0.7481445627621813,
      "grad_norm": 1.1015406847000122,
      "learning_rate": 0.0005,
      "loss": 2.512202739715576,
      "step": 4637
    },
    {
      "epoch": 0.7483059051306873,
      "grad_norm": 1.1083645820617676,
      "learning_rate": 0.0005,
      "loss": 2.5082526206970215,
      "step": 4638
    },
    {
      "epoch": 0.7484672474991932,
      "grad_norm": 1.150871992111206,
      "learning_rate": 0.0005,
      "loss": 2.477182388305664,
      "step": 4639
    },
    {
      "epoch": 0.7486285898676993,
      "grad_norm": 1.1545854806900024,
      "learning_rate": 0.0005,
      "loss": 2.998800754547119,
      "step": 4640
    },
    {
      "epoch": 0.7487899322362053,
      "grad_norm": 1.0323668718338013,
      "learning_rate": 0.0005,
      "loss": 2.47440505027771,
      "step": 4641
    },
    {
      "epoch": 0.7489512746047112,
      "grad_norm": 1.0993589162826538,
      "learning_rate": 0.0005,
      "loss": 2.667829990386963,
      "step": 4642
    },
    {
      "epoch": 0.7491126169732172,
      "grad_norm": 0.9577547907829285,
      "learning_rate": 0.0005,
      "loss": 2.6281495094299316,
      "step": 4643
    },
    {
      "epoch": 0.7492739593417231,
      "grad_norm": 0.878539502620697,
      "learning_rate": 0.0005,
      "loss": 2.325216770172119,
      "step": 4644
    },
    {
      "epoch": 0.7494353017102291,
      "grad_norm": 1.102976679801941,
      "learning_rate": 0.0005,
      "loss": 2.36348557472229,
      "step": 4645
    },
    {
      "epoch": 0.749596644078735,
      "grad_norm": 1.1427357196807861,
      "learning_rate": 0.0005,
      "loss": 2.1306099891662598,
      "step": 4646
    },
    {
      "epoch": 0.749757986447241,
      "grad_norm": 1.1286011934280396,
      "learning_rate": 0.0005,
      "loss": 2.202406883239746,
      "step": 4647
    },
    {
      "epoch": 0.749919328815747,
      "grad_norm": 1.2031915187835693,
      "learning_rate": 0.0005,
      "loss": 2.4929494857788086,
      "step": 4648
    },
    {
      "epoch": 0.750080671184253,
      "grad_norm": 2.3220272064208984,
      "learning_rate": 0.0005,
      "loss": 2.4385032653808594,
      "step": 4649
    },
    {
      "epoch": 0.750242013552759,
      "grad_norm": 1.157933235168457,
      "learning_rate": 0.0005,
      "loss": 2.568286895751953,
      "step": 4650
    },
    {
      "epoch": 0.750403355921265,
      "grad_norm": 1.5078924894332886,
      "learning_rate": 0.0005,
      "loss": 2.561213254928589,
      "step": 4651
    },
    {
      "epoch": 0.7505646982897709,
      "grad_norm": 1.293861985206604,
      "learning_rate": 0.0005,
      "loss": 2.753904342651367,
      "step": 4652
    },
    {
      "epoch": 0.7507260406582769,
      "grad_norm": 2.3514187335968018,
      "learning_rate": 0.0005,
      "loss": 2.243616819381714,
      "step": 4653
    },
    {
      "epoch": 0.7508873830267828,
      "grad_norm": 2.1819937229156494,
      "learning_rate": 0.0005,
      "loss": 2.964728832244873,
      "step": 4654
    },
    {
      "epoch": 0.7510487253952888,
      "grad_norm": 1.2949321269989014,
      "learning_rate": 0.0005,
      "loss": 2.596524477005005,
      "step": 4655
    },
    {
      "epoch": 0.7512100677637947,
      "grad_norm": 1.4735987186431885,
      "learning_rate": 0.0005,
      "loss": 2.5744738578796387,
      "step": 4656
    },
    {
      "epoch": 0.7513714101323007,
      "grad_norm": 1.7168599367141724,
      "learning_rate": 0.0005,
      "loss": 2.7062861919403076,
      "step": 4657
    },
    {
      "epoch": 0.7515327525008068,
      "grad_norm": 1.0506234169006348,
      "learning_rate": 0.0005,
      "loss": 2.412269115447998,
      "step": 4658
    },
    {
      "epoch": 0.7516940948693127,
      "grad_norm": 1.1410497426986694,
      "learning_rate": 0.0005,
      "loss": 2.3657541275024414,
      "step": 4659
    },
    {
      "epoch": 0.7518554372378187,
      "grad_norm": 0.8963696956634521,
      "learning_rate": 0.0005,
      "loss": 2.548206329345703,
      "step": 4660
    },
    {
      "epoch": 0.7520167796063246,
      "grad_norm": 1.2479294538497925,
      "learning_rate": 0.0005,
      "loss": 2.5311546325683594,
      "step": 4661
    },
    {
      "epoch": 0.7521781219748306,
      "grad_norm": 15.442442893981934,
      "learning_rate": 0.0005,
      "loss": 2.2367103099823,
      "step": 4662
    },
    {
      "epoch": 0.7523394643433365,
      "grad_norm": 0.8941348791122437,
      "learning_rate": 0.0005,
      "loss": 2.865884780883789,
      "step": 4663
    },
    {
      "epoch": 0.7525008067118425,
      "grad_norm": 1.3624918460845947,
      "learning_rate": 0.0005,
      "loss": 2.8463475704193115,
      "step": 4664
    },
    {
      "epoch": 0.7526621490803485,
      "grad_norm": 1.3337135314941406,
      "learning_rate": 0.0005,
      "loss": 2.626272201538086,
      "step": 4665
    },
    {
      "epoch": 0.7528234914488545,
      "grad_norm": 1.460379958152771,
      "learning_rate": 0.0005,
      "loss": 2.738070011138916,
      "step": 4666
    },
    {
      "epoch": 0.7529848338173605,
      "grad_norm": 1.074344515800476,
      "learning_rate": 0.0005,
      "loss": 2.6947479248046875,
      "step": 4667
    },
    {
      "epoch": 0.7531461761858664,
      "grad_norm": 0.9284229874610901,
      "learning_rate": 0.0005,
      "loss": 2.8241286277770996,
      "step": 4668
    },
    {
      "epoch": 0.7533075185543724,
      "grad_norm": 1.2931134700775146,
      "learning_rate": 0.0005,
      "loss": 1.8603572845458984,
      "step": 4669
    },
    {
      "epoch": 0.7534688609228783,
      "grad_norm": 1.2720921039581299,
      "learning_rate": 0.0005,
      "loss": 2.8843743801116943,
      "step": 4670
    },
    {
      "epoch": 0.7536302032913843,
      "grad_norm": 1.189326286315918,
      "learning_rate": 0.0005,
      "loss": 2.5551700592041016,
      "step": 4671
    },
    {
      "epoch": 0.7537915456598903,
      "grad_norm": 1.225605845451355,
      "learning_rate": 0.0005,
      "loss": 2.5098276138305664,
      "step": 4672
    },
    {
      "epoch": 0.7539528880283962,
      "grad_norm": 1.018060326576233,
      "learning_rate": 0.0005,
      "loss": 2.2763113975524902,
      "step": 4673
    },
    {
      "epoch": 0.7541142303969022,
      "grad_norm": 1.1377928256988525,
      "learning_rate": 0.0005,
      "loss": 2.340789794921875,
      "step": 4674
    },
    {
      "epoch": 0.7542755727654082,
      "grad_norm": 1.2614049911499023,
      "learning_rate": 0.0005,
      "loss": 2.1514430046081543,
      "step": 4675
    },
    {
      "epoch": 0.7544369151339142,
      "grad_norm": 1.0697991847991943,
      "learning_rate": 0.0005,
      "loss": 2.3360755443573,
      "step": 4676
    },
    {
      "epoch": 0.7545982575024202,
      "grad_norm": 1.2521599531173706,
      "learning_rate": 0.0005,
      "loss": 2.853684902191162,
      "step": 4677
    },
    {
      "epoch": 0.7547595998709261,
      "grad_norm": 1.1858165264129639,
      "learning_rate": 0.0005,
      "loss": 2.4552292823791504,
      "step": 4678
    },
    {
      "epoch": 0.7549209422394321,
      "grad_norm": 1.2990946769714355,
      "learning_rate": 0.0005,
      "loss": 2.8598995208740234,
      "step": 4679
    },
    {
      "epoch": 0.755082284607938,
      "grad_norm": 1.5047860145568848,
      "learning_rate": 0.0005,
      "loss": 2.399169921875,
      "step": 4680
    },
    {
      "epoch": 0.755243626976444,
      "grad_norm": 1.0592575073242188,
      "learning_rate": 0.0005,
      "loss": 2.083441734313965,
      "step": 4681
    },
    {
      "epoch": 0.7554049693449499,
      "grad_norm": 1.1592856645584106,
      "learning_rate": 0.0005,
      "loss": 2.610870361328125,
      "step": 4682
    },
    {
      "epoch": 0.755566311713456,
      "grad_norm": 1.024383306503296,
      "learning_rate": 0.0005,
      "loss": 2.6815202236175537,
      "step": 4683
    },
    {
      "epoch": 0.755727654081962,
      "grad_norm": 0.9345232248306274,
      "learning_rate": 0.0005,
      "loss": 2.24798583984375,
      "step": 4684
    },
    {
      "epoch": 0.7558889964504679,
      "grad_norm": 1.3866344690322876,
      "learning_rate": 0.0005,
      "loss": 2.3313493728637695,
      "step": 4685
    },
    {
      "epoch": 0.7560503388189739,
      "grad_norm": 1.5060750246047974,
      "learning_rate": 0.0005,
      "loss": 2.703822135925293,
      "step": 4686
    },
    {
      "epoch": 0.7562116811874798,
      "grad_norm": 1.2301498651504517,
      "learning_rate": 0.0005,
      "loss": 2.3521976470947266,
      "step": 4687
    },
    {
      "epoch": 0.7563730235559858,
      "grad_norm": 1.2322595119476318,
      "learning_rate": 0.0005,
      "loss": 2.6434648036956787,
      "step": 4688
    },
    {
      "epoch": 0.7565343659244917,
      "grad_norm": 1.2606967687606812,
      "learning_rate": 0.0005,
      "loss": 2.310542106628418,
      "step": 4689
    },
    {
      "epoch": 0.7566957082929977,
      "grad_norm": 1.2224737405776978,
      "learning_rate": 0.0005,
      "loss": 2.5131728649139404,
      "step": 4690
    },
    {
      "epoch": 0.7568570506615037,
      "grad_norm": 0.9580301642417908,
      "learning_rate": 0.0005,
      "loss": 2.032003879547119,
      "step": 4691
    },
    {
      "epoch": 0.7570183930300097,
      "grad_norm": 1.0782580375671387,
      "learning_rate": 0.0005,
      "loss": 2.1372597217559814,
      "step": 4692
    },
    {
      "epoch": 0.7571797353985157,
      "grad_norm": 1.0925501585006714,
      "learning_rate": 0.0005,
      "loss": 2.629286050796509,
      "step": 4693
    },
    {
      "epoch": 0.7573410777670216,
      "grad_norm": 1.3905524015426636,
      "learning_rate": 0.0005,
      "loss": 2.7198686599731445,
      "step": 4694
    },
    {
      "epoch": 0.7575024201355276,
      "grad_norm": 1.3694506883621216,
      "learning_rate": 0.0005,
      "loss": 2.402116298675537,
      "step": 4695
    },
    {
      "epoch": 0.7576637625040336,
      "grad_norm": 1.2587389945983887,
      "learning_rate": 0.0005,
      "loss": 2.4838497638702393,
      "step": 4696
    },
    {
      "epoch": 0.7578251048725395,
      "grad_norm": 1.4565706253051758,
      "learning_rate": 0.0005,
      "loss": 2.850034475326538,
      "step": 4697
    },
    {
      "epoch": 0.7579864472410455,
      "grad_norm": 1.3759244680404663,
      "learning_rate": 0.0005,
      "loss": 2.4933204650878906,
      "step": 4698
    },
    {
      "epoch": 0.7581477896095514,
      "grad_norm": 1.164394497871399,
      "learning_rate": 0.0005,
      "loss": 2.600783109664917,
      "step": 4699
    },
    {
      "epoch": 0.7583091319780574,
      "grad_norm": 1.1120359897613525,
      "learning_rate": 0.0005,
      "loss": 3.1072919368743896,
      "step": 4700
    },
    {
      "epoch": 0.7584704743465635,
      "grad_norm": 1.3352903127670288,
      "learning_rate": 0.0005,
      "loss": 2.156982421875,
      "step": 4701
    },
    {
      "epoch": 0.7586318167150694,
      "grad_norm": 1.0034834146499634,
      "learning_rate": 0.0005,
      "loss": 2.6506729125976562,
      "step": 4702
    },
    {
      "epoch": 0.7587931590835754,
      "grad_norm": 1.2886875867843628,
      "learning_rate": 0.0005,
      "loss": 2.4502429962158203,
      "step": 4703
    },
    {
      "epoch": 0.7589545014520813,
      "grad_norm": 1.7062387466430664,
      "learning_rate": 0.0005,
      "loss": 2.791823148727417,
      "step": 4704
    },
    {
      "epoch": 0.7591158438205873,
      "grad_norm": 1.4219805002212524,
      "learning_rate": 0.0005,
      "loss": 2.5508298873901367,
      "step": 4705
    },
    {
      "epoch": 0.7592771861890932,
      "grad_norm": 0.9787278175354004,
      "learning_rate": 0.0005,
      "loss": 2.447948694229126,
      "step": 4706
    },
    {
      "epoch": 0.7594385285575992,
      "grad_norm": 0.8171472549438477,
      "learning_rate": 0.0005,
      "loss": 2.228295087814331,
      "step": 4707
    },
    {
      "epoch": 0.7595998709261051,
      "grad_norm": 1.3819468021392822,
      "learning_rate": 0.0005,
      "loss": 3.159604549407959,
      "step": 4708
    },
    {
      "epoch": 0.7597612132946112,
      "grad_norm": 1.3384122848510742,
      "learning_rate": 0.0005,
      "loss": 2.5521397590637207,
      "step": 4709
    },
    {
      "epoch": 0.7599225556631172,
      "grad_norm": 1.2727769613265991,
      "learning_rate": 0.0005,
      "loss": 2.466869354248047,
      "step": 4710
    },
    {
      "epoch": 0.7600838980316231,
      "grad_norm": 1.0903966426849365,
      "learning_rate": 0.0005,
      "loss": 2.6203179359436035,
      "step": 4711
    },
    {
      "epoch": 0.7602452404001291,
      "grad_norm": 2.595531702041626,
      "learning_rate": 0.0005,
      "loss": 2.864992380142212,
      "step": 4712
    },
    {
      "epoch": 0.760406582768635,
      "grad_norm": 0.9903154969215393,
      "learning_rate": 0.0005,
      "loss": 2.4945743083953857,
      "step": 4713
    },
    {
      "epoch": 0.760567925137141,
      "grad_norm": 1.2791246175765991,
      "learning_rate": 0.0005,
      "loss": 2.835003614425659,
      "step": 4714
    },
    {
      "epoch": 0.760729267505647,
      "grad_norm": 1.573399543762207,
      "learning_rate": 0.0005,
      "loss": 2.420999050140381,
      "step": 4715
    },
    {
      "epoch": 0.7608906098741529,
      "grad_norm": 1.1304162740707397,
      "learning_rate": 0.0005,
      "loss": 2.4285330772399902,
      "step": 4716
    },
    {
      "epoch": 0.7610519522426589,
      "grad_norm": 0.982288122177124,
      "learning_rate": 0.0005,
      "loss": 2.647062063217163,
      "step": 4717
    },
    {
      "epoch": 0.7612132946111649,
      "grad_norm": 4.059304237365723,
      "learning_rate": 0.0005,
      "loss": 2.539731740951538,
      "step": 4718
    },
    {
      "epoch": 0.7613746369796709,
      "grad_norm": 1.7484300136566162,
      "learning_rate": 0.0005,
      "loss": 2.6853179931640625,
      "step": 4719
    },
    {
      "epoch": 0.7615359793481769,
      "grad_norm": 1.367309808731079,
      "learning_rate": 0.0005,
      "loss": 2.466203212738037,
      "step": 4720
    },
    {
      "epoch": 0.7616973217166828,
      "grad_norm": 1.3958914279937744,
      "learning_rate": 0.0005,
      "loss": 2.463331699371338,
      "step": 4721
    },
    {
      "epoch": 0.7618586640851888,
      "grad_norm": 0.9230026602745056,
      "learning_rate": 0.0005,
      "loss": 1.9628753662109375,
      "step": 4722
    },
    {
      "epoch": 0.7620200064536947,
      "grad_norm": 1.2155431509017944,
      "learning_rate": 0.0005,
      "loss": 2.7992849349975586,
      "step": 4723
    },
    {
      "epoch": 0.7621813488222007,
      "grad_norm": 1.1245975494384766,
      "learning_rate": 0.0005,
      "loss": 2.4214062690734863,
      "step": 4724
    },
    {
      "epoch": 0.7623426911907066,
      "grad_norm": 1.1536399126052856,
      "learning_rate": 0.0005,
      "loss": 2.1188507080078125,
      "step": 4725
    },
    {
      "epoch": 0.7625040335592127,
      "grad_norm": 1.1223424673080444,
      "learning_rate": 0.0005,
      "loss": 2.756027936935425,
      "step": 4726
    },
    {
      "epoch": 0.7626653759277187,
      "grad_norm": 1.7753173112869263,
      "learning_rate": 0.0005,
      "loss": 2.41129207611084,
      "step": 4727
    },
    {
      "epoch": 0.7628267182962246,
      "grad_norm": 0.9527396559715271,
      "learning_rate": 0.0005,
      "loss": 2.1755919456481934,
      "step": 4728
    },
    {
      "epoch": 0.7629880606647306,
      "grad_norm": 1.5884718894958496,
      "learning_rate": 0.0005,
      "loss": 2.775846242904663,
      "step": 4729
    },
    {
      "epoch": 0.7631494030332365,
      "grad_norm": 1.191834568977356,
      "learning_rate": 0.0005,
      "loss": 2.758370876312256,
      "step": 4730
    },
    {
      "epoch": 0.7633107454017425,
      "grad_norm": 1.2052184343338013,
      "learning_rate": 0.0005,
      "loss": 2.460280418395996,
      "step": 4731
    },
    {
      "epoch": 0.7634720877702484,
      "grad_norm": 0.8725925087928772,
      "learning_rate": 0.0005,
      "loss": 2.0267415046691895,
      "step": 4732
    },
    {
      "epoch": 0.7636334301387544,
      "grad_norm": 1.146536946296692,
      "learning_rate": 0.0005,
      "loss": 2.6605935096740723,
      "step": 4733
    },
    {
      "epoch": 0.7637947725072604,
      "grad_norm": 1.2883992195129395,
      "learning_rate": 0.0005,
      "loss": 2.789203643798828,
      "step": 4734
    },
    {
      "epoch": 0.7639561148757664,
      "grad_norm": 1.1644757986068726,
      "learning_rate": 0.0005,
      "loss": 2.2719759941101074,
      "step": 4735
    },
    {
      "epoch": 0.7641174572442724,
      "grad_norm": 0.9286680221557617,
      "learning_rate": 0.0005,
      "loss": 2.450927734375,
      "step": 4736
    },
    {
      "epoch": 0.7642787996127783,
      "grad_norm": 1.2902590036392212,
      "learning_rate": 0.0005,
      "loss": 2.5736236572265625,
      "step": 4737
    },
    {
      "epoch": 0.7644401419812843,
      "grad_norm": 1.6604326963424683,
      "learning_rate": 0.0005,
      "loss": 2.7196834087371826,
      "step": 4738
    },
    {
      "epoch": 0.7646014843497903,
      "grad_norm": 1.009372591972351,
      "learning_rate": 0.0005,
      "loss": 2.7496821880340576,
      "step": 4739
    },
    {
      "epoch": 0.7647628267182962,
      "grad_norm": 1.2541451454162598,
      "learning_rate": 0.0005,
      "loss": 2.7297353744506836,
      "step": 4740
    },
    {
      "epoch": 0.7649241690868022,
      "grad_norm": 1.099994421005249,
      "learning_rate": 0.0005,
      "loss": 2.5197553634643555,
      "step": 4741
    },
    {
      "epoch": 0.7650855114553081,
      "grad_norm": 0.9795411825180054,
      "learning_rate": 0.0005,
      "loss": 2.687713146209717,
      "step": 4742
    },
    {
      "epoch": 0.7652468538238142,
      "grad_norm": 0.9232755303382874,
      "learning_rate": 0.0005,
      "loss": 2.5880770683288574,
      "step": 4743
    },
    {
      "epoch": 0.7654081961923201,
      "grad_norm": 1.3713281154632568,
      "learning_rate": 0.0005,
      "loss": 2.742034912109375,
      "step": 4744
    },
    {
      "epoch": 0.7655695385608261,
      "grad_norm": 1.5898914337158203,
      "learning_rate": 0.0005,
      "loss": 2.7570104598999023,
      "step": 4745
    },
    {
      "epoch": 0.7657308809293321,
      "grad_norm": 1.095023274421692,
      "learning_rate": 0.0005,
      "loss": 2.519939422607422,
      "step": 4746
    },
    {
      "epoch": 0.765892223297838,
      "grad_norm": 1.2248528003692627,
      "learning_rate": 0.0005,
      "loss": 2.3212833404541016,
      "step": 4747
    },
    {
      "epoch": 0.766053565666344,
      "grad_norm": 1.033294439315796,
      "learning_rate": 0.0005,
      "loss": 2.3896427154541016,
      "step": 4748
    },
    {
      "epoch": 0.7662149080348499,
      "grad_norm": 1.0817327499389648,
      "learning_rate": 0.0005,
      "loss": 2.729421377182007,
      "step": 4749
    },
    {
      "epoch": 0.7663762504033559,
      "grad_norm": 1.5850474834442139,
      "learning_rate": 0.0005,
      "loss": 2.906853437423706,
      "step": 4750
    },
    {
      "epoch": 0.7665375927718618,
      "grad_norm": 0.8816360235214233,
      "learning_rate": 0.0005,
      "loss": 2.5290074348449707,
      "step": 4751
    },
    {
      "epoch": 0.7666989351403679,
      "grad_norm": 1.109032154083252,
      "learning_rate": 0.0005,
      "loss": 2.541802167892456,
      "step": 4752
    },
    {
      "epoch": 0.7668602775088739,
      "grad_norm": 1.0493084192276,
      "learning_rate": 0.0005,
      "loss": 2.639425754547119,
      "step": 4753
    },
    {
      "epoch": 0.7670216198773798,
      "grad_norm": 1.1441656351089478,
      "learning_rate": 0.0005,
      "loss": 2.8506920337677,
      "step": 4754
    },
    {
      "epoch": 0.7671829622458858,
      "grad_norm": 1.313643455505371,
      "learning_rate": 0.0005,
      "loss": 2.6047611236572266,
      "step": 4755
    },
    {
      "epoch": 0.7673443046143917,
      "grad_norm": 1.4312729835510254,
      "learning_rate": 0.0005,
      "loss": 2.9139325618743896,
      "step": 4756
    },
    {
      "epoch": 0.7675056469828977,
      "grad_norm": 0.9561951160430908,
      "learning_rate": 0.0005,
      "loss": 2.5698060989379883,
      "step": 4757
    },
    {
      "epoch": 0.7676669893514037,
      "grad_norm": 1.115596890449524,
      "learning_rate": 0.0005,
      "loss": 2.1626458168029785,
      "step": 4758
    },
    {
      "epoch": 0.7678283317199096,
      "grad_norm": 1.3198438882827759,
      "learning_rate": 0.0005,
      "loss": 2.389496326446533,
      "step": 4759
    },
    {
      "epoch": 0.7679896740884156,
      "grad_norm": 1.4310219287872314,
      "learning_rate": 0.0005,
      "loss": 2.4689135551452637,
      "step": 4760
    },
    {
      "epoch": 0.7681510164569216,
      "grad_norm": 1.2429672479629517,
      "learning_rate": 0.0005,
      "loss": 2.6786231994628906,
      "step": 4761
    },
    {
      "epoch": 0.7683123588254276,
      "grad_norm": 0.97584068775177,
      "learning_rate": 0.0005,
      "loss": 2.1539230346679688,
      "step": 4762
    },
    {
      "epoch": 0.7684737011939335,
      "grad_norm": 1.2028213739395142,
      "learning_rate": 0.0005,
      "loss": 2.548470973968506,
      "step": 4763
    },
    {
      "epoch": 0.7686350435624395,
      "grad_norm": 0.9235853552818298,
      "learning_rate": 0.0005,
      "loss": 2.558095693588257,
      "step": 4764
    },
    {
      "epoch": 0.7687963859309455,
      "grad_norm": 1.1179394721984863,
      "learning_rate": 0.0005,
      "loss": 2.775550365447998,
      "step": 4765
    },
    {
      "epoch": 0.7689577282994514,
      "grad_norm": 1.4568872451782227,
      "learning_rate": 0.0005,
      "loss": 2.389486312866211,
      "step": 4766
    },
    {
      "epoch": 0.7691190706679574,
      "grad_norm": 1.0286709070205688,
      "learning_rate": 0.0005,
      "loss": 2.7005722522735596,
      "step": 4767
    },
    {
      "epoch": 0.7692804130364633,
      "grad_norm": 1.3061280250549316,
      "learning_rate": 0.0005,
      "loss": 2.826547145843506,
      "step": 4768
    },
    {
      "epoch": 0.7694417554049694,
      "grad_norm": 1.1424894332885742,
      "learning_rate": 0.0005,
      "loss": 2.245265483856201,
      "step": 4769
    },
    {
      "epoch": 0.7696030977734754,
      "grad_norm": 1.1584523916244507,
      "learning_rate": 0.0005,
      "loss": 2.5538246631622314,
      "step": 4770
    },
    {
      "epoch": 0.7697644401419813,
      "grad_norm": 1.038426160812378,
      "learning_rate": 0.0005,
      "loss": 2.5578064918518066,
      "step": 4771
    },
    {
      "epoch": 0.7699257825104873,
      "grad_norm": 1.087706208229065,
      "learning_rate": 0.0005,
      "loss": 2.854694366455078,
      "step": 4772
    },
    {
      "epoch": 0.7700871248789932,
      "grad_norm": 1.4811286926269531,
      "learning_rate": 0.0005,
      "loss": 2.864288806915283,
      "step": 4773
    },
    {
      "epoch": 0.7702484672474992,
      "grad_norm": 1.1478747129440308,
      "learning_rate": 0.0005,
      "loss": 2.41278338432312,
      "step": 4774
    },
    {
      "epoch": 0.7704098096160051,
      "grad_norm": 1.448837399482727,
      "learning_rate": 0.0005,
      "loss": 2.581630229949951,
      "step": 4775
    },
    {
      "epoch": 0.7705711519845111,
      "grad_norm": 1.0059410333633423,
      "learning_rate": 0.0005,
      "loss": 2.229264974594116,
      "step": 4776
    },
    {
      "epoch": 0.770732494353017,
      "grad_norm": 0.9941322803497314,
      "learning_rate": 0.0005,
      "loss": 2.4852123260498047,
      "step": 4777
    },
    {
      "epoch": 0.7708938367215231,
      "grad_norm": 1.2987147569656372,
      "learning_rate": 0.0005,
      "loss": 2.7946739196777344,
      "step": 4778
    },
    {
      "epoch": 0.7710551790900291,
      "grad_norm": 1.064120888710022,
      "learning_rate": 0.0005,
      "loss": 2.1042871475219727,
      "step": 4779
    },
    {
      "epoch": 0.771216521458535,
      "grad_norm": 1.9485136270523071,
      "learning_rate": 0.0005,
      "loss": 2.767822265625,
      "step": 4780
    },
    {
      "epoch": 0.771377863827041,
      "grad_norm": 1.11818528175354,
      "learning_rate": 0.0005,
      "loss": 2.1922569274902344,
      "step": 4781
    },
    {
      "epoch": 0.771539206195547,
      "grad_norm": 1.2532339096069336,
      "learning_rate": 0.0005,
      "loss": 2.94484543800354,
      "step": 4782
    },
    {
      "epoch": 0.7717005485640529,
      "grad_norm": 1.3399407863616943,
      "learning_rate": 0.0005,
      "loss": 2.769120693206787,
      "step": 4783
    },
    {
      "epoch": 0.7718618909325589,
      "grad_norm": 1.0985413789749146,
      "learning_rate": 0.0005,
      "loss": 2.8127260208129883,
      "step": 4784
    },
    {
      "epoch": 0.7720232333010648,
      "grad_norm": 1.1901522874832153,
      "learning_rate": 0.0005,
      "loss": 2.7009401321411133,
      "step": 4785
    },
    {
      "epoch": 0.7721845756695709,
      "grad_norm": 1.3409759998321533,
      "learning_rate": 0.0005,
      "loss": 2.476810932159424,
      "step": 4786
    },
    {
      "epoch": 0.7723459180380768,
      "grad_norm": 1.2551900148391724,
      "learning_rate": 0.0005,
      "loss": 2.5738930702209473,
      "step": 4787
    },
    {
      "epoch": 0.7725072604065828,
      "grad_norm": 1.1540127992630005,
      "learning_rate": 0.0005,
      "loss": 2.6221957206726074,
      "step": 4788
    },
    {
      "epoch": 0.7726686027750888,
      "grad_norm": 1.0922372341156006,
      "learning_rate": 0.0005,
      "loss": 2.1631321907043457,
      "step": 4789
    },
    {
      "epoch": 0.7728299451435947,
      "grad_norm": 1.3104314804077148,
      "learning_rate": 0.0005,
      "loss": 2.459355592727661,
      "step": 4790
    },
    {
      "epoch": 0.7729912875121007,
      "grad_norm": 1.1201428174972534,
      "learning_rate": 0.0005,
      "loss": 2.2408931255340576,
      "step": 4791
    },
    {
      "epoch": 0.7731526298806066,
      "grad_norm": 1.4800268411636353,
      "learning_rate": 0.0005,
      "loss": 2.3980307579040527,
      "step": 4792
    },
    {
      "epoch": 0.7733139722491126,
      "grad_norm": 0.9753093719482422,
      "learning_rate": 0.0005,
      "loss": 2.4784927368164062,
      "step": 4793
    },
    {
      "epoch": 0.7734753146176185,
      "grad_norm": 1.1698499917984009,
      "learning_rate": 0.0005,
      "loss": 2.883375883102417,
      "step": 4794
    },
    {
      "epoch": 0.7736366569861246,
      "grad_norm": 1.120516061782837,
      "learning_rate": 0.0005,
      "loss": 2.3063178062438965,
      "step": 4795
    },
    {
      "epoch": 0.7737979993546306,
      "grad_norm": 1.2895666360855103,
      "learning_rate": 0.0005,
      "loss": 2.450406789779663,
      "step": 4796
    },
    {
      "epoch": 0.7739593417231365,
      "grad_norm": 1.0465668439865112,
      "learning_rate": 0.0005,
      "loss": 2.12811279296875,
      "step": 4797
    },
    {
      "epoch": 0.7741206840916425,
      "grad_norm": 1.184342861175537,
      "learning_rate": 0.0005,
      "loss": 2.7342605590820312,
      "step": 4798
    },
    {
      "epoch": 0.7742820264601484,
      "grad_norm": 0.9283798933029175,
      "learning_rate": 0.0005,
      "loss": 2.545121431350708,
      "step": 4799
    },
    {
      "epoch": 0.7744433688286544,
      "grad_norm": 1.0587377548217773,
      "learning_rate": 0.0005,
      "loss": 2.2942123413085938,
      "step": 4800
    },
    {
      "epoch": 0.7746047111971603,
      "grad_norm": 1.0491082668304443,
      "learning_rate": 0.0005,
      "loss": 2.4091527462005615,
      "step": 4801
    },
    {
      "epoch": 0.7747660535656663,
      "grad_norm": 1.1260673999786377,
      "learning_rate": 0.0005,
      "loss": 2.4937448501586914,
      "step": 4802
    },
    {
      "epoch": 0.7749273959341723,
      "grad_norm": 1.3848463296890259,
      "learning_rate": 0.0005,
      "loss": 2.3980979919433594,
      "step": 4803
    },
    {
      "epoch": 0.7750887383026783,
      "grad_norm": 1.3726729154586792,
      "learning_rate": 0.0005,
      "loss": 2.4778127670288086,
      "step": 4804
    },
    {
      "epoch": 0.7752500806711843,
      "grad_norm": 1.3836538791656494,
      "learning_rate": 0.0005,
      "loss": 2.5665810108184814,
      "step": 4805
    },
    {
      "epoch": 0.7754114230396902,
      "grad_norm": 1.0755888223648071,
      "learning_rate": 0.0005,
      "loss": 2.552396297454834,
      "step": 4806
    },
    {
      "epoch": 0.7755727654081962,
      "grad_norm": 1.5073199272155762,
      "learning_rate": 0.0005,
      "loss": 2.3014447689056396,
      "step": 4807
    },
    {
      "epoch": 0.7757341077767022,
      "grad_norm": 1.2251784801483154,
      "learning_rate": 0.0005,
      "loss": 3.06508207321167,
      "step": 4808
    },
    {
      "epoch": 0.7758954501452081,
      "grad_norm": 0.940430760383606,
      "learning_rate": 0.0005,
      "loss": 2.4139585494995117,
      "step": 4809
    },
    {
      "epoch": 0.7760567925137141,
      "grad_norm": 1.1762531995773315,
      "learning_rate": 0.0005,
      "loss": 2.436310291290283,
      "step": 4810
    },
    {
      "epoch": 0.77621813488222,
      "grad_norm": 1.0723377466201782,
      "learning_rate": 0.0005,
      "loss": 2.6406612396240234,
      "step": 4811
    },
    {
      "epoch": 0.7763794772507261,
      "grad_norm": 1.2748745679855347,
      "learning_rate": 0.0005,
      "loss": 2.511629581451416,
      "step": 4812
    },
    {
      "epoch": 0.776540819619232,
      "grad_norm": 1.1447501182556152,
      "learning_rate": 0.0005,
      "loss": 2.4984359741210938,
      "step": 4813
    },
    {
      "epoch": 0.776702161987738,
      "grad_norm": 1.3986680507659912,
      "learning_rate": 0.0005,
      "loss": 2.793071746826172,
      "step": 4814
    },
    {
      "epoch": 0.776863504356244,
      "grad_norm": 1.130669355392456,
      "learning_rate": 0.0005,
      "loss": 2.2914092540740967,
      "step": 4815
    },
    {
      "epoch": 0.7770248467247499,
      "grad_norm": 1.4551198482513428,
      "learning_rate": 0.0005,
      "loss": 2.6291356086730957,
      "step": 4816
    },
    {
      "epoch": 0.7771861890932559,
      "grad_norm": 1.1236692667007446,
      "learning_rate": 0.0005,
      "loss": 2.2716293334960938,
      "step": 4817
    },
    {
      "epoch": 0.7773475314617618,
      "grad_norm": 1.030311942100525,
      "learning_rate": 0.0005,
      "loss": 2.441452741622925,
      "step": 4818
    },
    {
      "epoch": 0.7775088738302678,
      "grad_norm": 1.2201811075210571,
      "learning_rate": 0.0005,
      "loss": 2.5221798419952393,
      "step": 4819
    },
    {
      "epoch": 0.7776702161987737,
      "grad_norm": 1.0212409496307373,
      "learning_rate": 0.0005,
      "loss": 2.364482879638672,
      "step": 4820
    },
    {
      "epoch": 0.7778315585672798,
      "grad_norm": 0.8877346515655518,
      "learning_rate": 0.0005,
      "loss": 2.6658902168273926,
      "step": 4821
    },
    {
      "epoch": 0.7779929009357858,
      "grad_norm": 1.5165807008743286,
      "learning_rate": 0.0005,
      "loss": 2.2041711807250977,
      "step": 4822
    },
    {
      "epoch": 0.7781542433042917,
      "grad_norm": 1.1978111267089844,
      "learning_rate": 0.0005,
      "loss": 2.813978433609009,
      "step": 4823
    },
    {
      "epoch": 0.7783155856727977,
      "grad_norm": 1.0666753053665161,
      "learning_rate": 0.0005,
      "loss": 2.668973445892334,
      "step": 4824
    },
    {
      "epoch": 0.7784769280413036,
      "grad_norm": 1.3602607250213623,
      "learning_rate": 0.0005,
      "loss": 2.4775476455688477,
      "step": 4825
    },
    {
      "epoch": 0.7786382704098096,
      "grad_norm": 1.1803624629974365,
      "learning_rate": 0.0005,
      "loss": 2.157592535018921,
      "step": 4826
    },
    {
      "epoch": 0.7787996127783156,
      "grad_norm": 1.3324776887893677,
      "learning_rate": 0.0005,
      "loss": 2.397258758544922,
      "step": 4827
    },
    {
      "epoch": 0.7789609551468215,
      "grad_norm": 2.2335574626922607,
      "learning_rate": 0.0005,
      "loss": 2.983187198638916,
      "step": 4828
    },
    {
      "epoch": 0.7791222975153276,
      "grad_norm": 1.2665596008300781,
      "learning_rate": 0.0005,
      "loss": 2.877446174621582,
      "step": 4829
    },
    {
      "epoch": 0.7792836398838335,
      "grad_norm": 0.9752010107040405,
      "learning_rate": 0.0005,
      "loss": 2.241554021835327,
      "step": 4830
    },
    {
      "epoch": 0.7794449822523395,
      "grad_norm": 1.0717895030975342,
      "learning_rate": 0.0005,
      "loss": 2.7284507751464844,
      "step": 4831
    },
    {
      "epoch": 0.7796063246208454,
      "grad_norm": 1.181462287902832,
      "learning_rate": 0.0005,
      "loss": 2.398672342300415,
      "step": 4832
    },
    {
      "epoch": 0.7797676669893514,
      "grad_norm": 1.077890396118164,
      "learning_rate": 0.0005,
      "loss": 2.7199649810791016,
      "step": 4833
    },
    {
      "epoch": 0.7799290093578574,
      "grad_norm": 1.2637962102890015,
      "learning_rate": 0.0005,
      "loss": 3.155644178390503,
      "step": 4834
    },
    {
      "epoch": 0.7800903517263633,
      "grad_norm": 1.1948884725570679,
      "learning_rate": 0.0005,
      "loss": 2.65687894821167,
      "step": 4835
    },
    {
      "epoch": 0.7802516940948693,
      "grad_norm": 1.2308766841888428,
      "learning_rate": 0.0005,
      "loss": 2.3411924839019775,
      "step": 4836
    },
    {
      "epoch": 0.7804130364633752,
      "grad_norm": 1.272266149520874,
      "learning_rate": 0.0005,
      "loss": 2.846104383468628,
      "step": 4837
    },
    {
      "epoch": 0.7805743788318813,
      "grad_norm": 1.1163266897201538,
      "learning_rate": 0.0005,
      "loss": 2.8571105003356934,
      "step": 4838
    },
    {
      "epoch": 0.7807357212003873,
      "grad_norm": 1.0156053304672241,
      "learning_rate": 0.0005,
      "loss": 2.743259906768799,
      "step": 4839
    },
    {
      "epoch": 0.7808970635688932,
      "grad_norm": 1.8591859340667725,
      "learning_rate": 0.0005,
      "loss": 2.2797722816467285,
      "step": 4840
    },
    {
      "epoch": 0.7810584059373992,
      "grad_norm": 1.5318032503128052,
      "learning_rate": 0.0005,
      "loss": 2.733175754547119,
      "step": 4841
    },
    {
      "epoch": 0.7812197483059051,
      "grad_norm": 1.0177937746047974,
      "learning_rate": 0.0005,
      "loss": 2.658738613128662,
      "step": 4842
    },
    {
      "epoch": 0.7813810906744111,
      "grad_norm": 1.2756168842315674,
      "learning_rate": 0.0005,
      "loss": 2.4532155990600586,
      "step": 4843
    },
    {
      "epoch": 0.781542433042917,
      "grad_norm": 1.1695257425308228,
      "learning_rate": 0.0005,
      "loss": 2.0599164962768555,
      "step": 4844
    },
    {
      "epoch": 0.781703775411423,
      "grad_norm": 1.2773046493530273,
      "learning_rate": 0.0005,
      "loss": 2.3477842807769775,
      "step": 4845
    },
    {
      "epoch": 0.7818651177799291,
      "grad_norm": 1.8710218667984009,
      "learning_rate": 0.0005,
      "loss": 3.099979877471924,
      "step": 4846
    },
    {
      "epoch": 0.782026460148435,
      "grad_norm": 1.2499091625213623,
      "learning_rate": 0.0005,
      "loss": 2.345806837081909,
      "step": 4847
    },
    {
      "epoch": 0.782187802516941,
      "grad_norm": 1.1254827976226807,
      "learning_rate": 0.0005,
      "loss": 2.4914088249206543,
      "step": 4848
    },
    {
      "epoch": 0.7823491448854469,
      "grad_norm": 0.9738696813583374,
      "learning_rate": 0.0005,
      "loss": 2.154680013656616,
      "step": 4849
    },
    {
      "epoch": 0.7825104872539529,
      "grad_norm": 1.1389678716659546,
      "learning_rate": 0.0005,
      "loss": 2.5945703983306885,
      "step": 4850
    },
    {
      "epoch": 0.7826718296224588,
      "grad_norm": 1.16068696975708,
      "learning_rate": 0.0005,
      "loss": 2.2030482292175293,
      "step": 4851
    },
    {
      "epoch": 0.7828331719909648,
      "grad_norm": 1.2384529113769531,
      "learning_rate": 0.0005,
      "loss": 2.658395528793335,
      "step": 4852
    },
    {
      "epoch": 0.7829945143594708,
      "grad_norm": 1.2440440654754639,
      "learning_rate": 0.0005,
      "loss": 2.447829246520996,
      "step": 4853
    },
    {
      "epoch": 0.7831558567279767,
      "grad_norm": 1.05245041847229,
      "learning_rate": 0.0005,
      "loss": 2.541250705718994,
      "step": 4854
    },
    {
      "epoch": 0.7833171990964828,
      "grad_norm": 1.2273372411727905,
      "learning_rate": 0.0005,
      "loss": 2.5613853931427,
      "step": 4855
    },
    {
      "epoch": 0.7834785414649887,
      "grad_norm": 0.9923788905143738,
      "learning_rate": 0.0005,
      "loss": 2.8373537063598633,
      "step": 4856
    },
    {
      "epoch": 0.7836398838334947,
      "grad_norm": 1.0704749822616577,
      "learning_rate": 0.0005,
      "loss": 2.3811302185058594,
      "step": 4857
    },
    {
      "epoch": 0.7838012262020007,
      "grad_norm": 1.1902008056640625,
      "learning_rate": 0.0005,
      "loss": 2.6378719806671143,
      "step": 4858
    },
    {
      "epoch": 0.7839625685705066,
      "grad_norm": 0.919959545135498,
      "learning_rate": 0.0005,
      "loss": 3.181828260421753,
      "step": 4859
    },
    {
      "epoch": 0.7841239109390126,
      "grad_norm": 1.1331061124801636,
      "learning_rate": 0.0005,
      "loss": 2.518670082092285,
      "step": 4860
    },
    {
      "epoch": 0.7842852533075185,
      "grad_norm": 1.035094976425171,
      "learning_rate": 0.0005,
      "loss": 2.4587230682373047,
      "step": 4861
    },
    {
      "epoch": 0.7844465956760245,
      "grad_norm": 0.9523653984069824,
      "learning_rate": 0.0005,
      "loss": 2.4961822032928467,
      "step": 4862
    },
    {
      "epoch": 0.7846079380445304,
      "grad_norm": 0.9570661187171936,
      "learning_rate": 0.0005,
      "loss": 2.3408563137054443,
      "step": 4863
    },
    {
      "epoch": 0.7847692804130365,
      "grad_norm": 1.0882176160812378,
      "learning_rate": 0.0005,
      "loss": 2.770321846008301,
      "step": 4864
    },
    {
      "epoch": 0.7849306227815425,
      "grad_norm": 1.1457092761993408,
      "learning_rate": 0.0005,
      "loss": 2.613750696182251,
      "step": 4865
    },
    {
      "epoch": 0.7850919651500484,
      "grad_norm": 1.0577973127365112,
      "learning_rate": 0.0005,
      "loss": 2.4082717895507812,
      "step": 4866
    },
    {
      "epoch": 0.7852533075185544,
      "grad_norm": 0.8888022303581238,
      "learning_rate": 0.0005,
      "loss": 2.387295961380005,
      "step": 4867
    },
    {
      "epoch": 0.7854146498870603,
      "grad_norm": 0.9024696946144104,
      "learning_rate": 0.0005,
      "loss": 2.3019566535949707,
      "step": 4868
    },
    {
      "epoch": 0.7855759922555663,
      "grad_norm": 0.9861170649528503,
      "learning_rate": 0.0005,
      "loss": 2.439725637435913,
      "step": 4869
    },
    {
      "epoch": 0.7857373346240722,
      "grad_norm": 1.1602497100830078,
      "learning_rate": 0.0005,
      "loss": 2.579151153564453,
      "step": 4870
    },
    {
      "epoch": 0.7858986769925782,
      "grad_norm": 1.1731077432632446,
      "learning_rate": 0.0005,
      "loss": 2.871466636657715,
      "step": 4871
    },
    {
      "epoch": 0.7860600193610843,
      "grad_norm": 1.1766555309295654,
      "learning_rate": 0.0005,
      "loss": 2.584531545639038,
      "step": 4872
    },
    {
      "epoch": 0.7862213617295902,
      "grad_norm": 1.0400735139846802,
      "learning_rate": 0.0005,
      "loss": 2.650137424468994,
      "step": 4873
    },
    {
      "epoch": 0.7863827040980962,
      "grad_norm": 1.2471539974212646,
      "learning_rate": 0.0005,
      "loss": 2.8480327129364014,
      "step": 4874
    },
    {
      "epoch": 0.7865440464666021,
      "grad_norm": 1.1729462146759033,
      "learning_rate": 0.0005,
      "loss": 2.9398374557495117,
      "step": 4875
    },
    {
      "epoch": 0.7867053888351081,
      "grad_norm": 0.9397968649864197,
      "learning_rate": 0.0005,
      "loss": 2.544037342071533,
      "step": 4876
    },
    {
      "epoch": 0.7868667312036141,
      "grad_norm": 1.328538179397583,
      "learning_rate": 0.0005,
      "loss": 2.740818977355957,
      "step": 4877
    },
    {
      "epoch": 0.78702807357212,
      "grad_norm": 1.3713637590408325,
      "learning_rate": 0.0005,
      "loss": 2.4134268760681152,
      "step": 4878
    },
    {
      "epoch": 0.787189415940626,
      "grad_norm": 1.2177844047546387,
      "learning_rate": 0.0005,
      "loss": 2.346907615661621,
      "step": 4879
    },
    {
      "epoch": 0.7873507583091319,
      "grad_norm": 1.4583152532577515,
      "learning_rate": 0.0005,
      "loss": 2.5338242053985596,
      "step": 4880
    },
    {
      "epoch": 0.787512100677638,
      "grad_norm": 1.2659332752227783,
      "learning_rate": 0.0005,
      "loss": 2.728276491165161,
      "step": 4881
    },
    {
      "epoch": 0.787673443046144,
      "grad_norm": 1.1893035173416138,
      "learning_rate": 0.0005,
      "loss": 2.812384843826294,
      "step": 4882
    },
    {
      "epoch": 0.7878347854146499,
      "grad_norm": 1.114518404006958,
      "learning_rate": 0.0005,
      "loss": 2.5338611602783203,
      "step": 4883
    },
    {
      "epoch": 0.7879961277831559,
      "grad_norm": 1.005948781967163,
      "learning_rate": 0.0005,
      "loss": 2.487649440765381,
      "step": 4884
    },
    {
      "epoch": 0.7881574701516618,
      "grad_norm": 0.8584744334220886,
      "learning_rate": 0.0005,
      "loss": 2.9017081260681152,
      "step": 4885
    },
    {
      "epoch": 0.7883188125201678,
      "grad_norm": 1.2450363636016846,
      "learning_rate": 0.0005,
      "loss": 2.354605197906494,
      "step": 4886
    },
    {
      "epoch": 0.7884801548886737,
      "grad_norm": 1.2600888013839722,
      "learning_rate": 0.0005,
      "loss": 2.220262050628662,
      "step": 4887
    },
    {
      "epoch": 0.7886414972571797,
      "grad_norm": 1.1588985919952393,
      "learning_rate": 0.0005,
      "loss": 2.668168544769287,
      "step": 4888
    },
    {
      "epoch": 0.7888028396256858,
      "grad_norm": 1.2492923736572266,
      "learning_rate": 0.0005,
      "loss": 2.472844123840332,
      "step": 4889
    },
    {
      "epoch": 0.7889641819941917,
      "grad_norm": 1.2599142789840698,
      "learning_rate": 0.0005,
      "loss": 2.5533523559570312,
      "step": 4890
    },
    {
      "epoch": 0.7891255243626977,
      "grad_norm": 1.1228004693984985,
      "learning_rate": 0.0005,
      "loss": 2.6967716217041016,
      "step": 4891
    },
    {
      "epoch": 0.7892868667312036,
      "grad_norm": 1.0517048835754395,
      "learning_rate": 0.0005,
      "loss": 2.1337180137634277,
      "step": 4892
    },
    {
      "epoch": 0.7894482090997096,
      "grad_norm": 0.9160770773887634,
      "learning_rate": 0.0005,
      "loss": 2.314624309539795,
      "step": 4893
    },
    {
      "epoch": 0.7896095514682155,
      "grad_norm": 2.048135757446289,
      "learning_rate": 0.0005,
      "loss": 2.9052529335021973,
      "step": 4894
    },
    {
      "epoch": 0.7897708938367215,
      "grad_norm": 0.9967055320739746,
      "learning_rate": 0.0005,
      "loss": 2.7696080207824707,
      "step": 4895
    },
    {
      "epoch": 0.7899322362052275,
      "grad_norm": 1.0366419553756714,
      "learning_rate": 0.0005,
      "loss": 2.934171199798584,
      "step": 4896
    },
    {
      "epoch": 0.7900935785737334,
      "grad_norm": 1.003130555152893,
      "learning_rate": 0.0005,
      "loss": 2.2518351078033447,
      "step": 4897
    },
    {
      "epoch": 0.7902549209422395,
      "grad_norm": 1.4868249893188477,
      "learning_rate": 0.0005,
      "loss": 2.8517651557922363,
      "step": 4898
    },
    {
      "epoch": 0.7904162633107454,
      "grad_norm": 1.0105875730514526,
      "learning_rate": 0.0005,
      "loss": 2.196436882019043,
      "step": 4899
    },
    {
      "epoch": 0.7905776056792514,
      "grad_norm": 1.1598117351531982,
      "learning_rate": 0.0005,
      "loss": 2.5936665534973145,
      "step": 4900
    },
    {
      "epoch": 0.7907389480477574,
      "grad_norm": 0.950390100479126,
      "learning_rate": 0.0005,
      "loss": 2.695467472076416,
      "step": 4901
    },
    {
      "epoch": 0.7909002904162633,
      "grad_norm": 1.2332947254180908,
      "learning_rate": 0.0005,
      "loss": 1.9251865148544312,
      "step": 4902
    },
    {
      "epoch": 0.7910616327847693,
      "grad_norm": 1.4000327587127686,
      "learning_rate": 0.0005,
      "loss": 2.812438488006592,
      "step": 4903
    },
    {
      "epoch": 0.7912229751532752,
      "grad_norm": 1.2828117609024048,
      "learning_rate": 0.0005,
      "loss": 2.471768856048584,
      "step": 4904
    },
    {
      "epoch": 0.7913843175217812,
      "grad_norm": 1.3205335140228271,
      "learning_rate": 0.0005,
      "loss": 2.42630672454834,
      "step": 4905
    },
    {
      "epoch": 0.7915456598902871,
      "grad_norm": 1.2413386106491089,
      "learning_rate": 0.0005,
      "loss": 2.7621026039123535,
      "step": 4906
    },
    {
      "epoch": 0.7917070022587932,
      "grad_norm": 1.045593023300171,
      "learning_rate": 0.0005,
      "loss": 2.8378067016601562,
      "step": 4907
    },
    {
      "epoch": 0.7918683446272992,
      "grad_norm": 1.309727430343628,
      "learning_rate": 0.0005,
      "loss": 2.2215325832366943,
      "step": 4908
    },
    {
      "epoch": 0.7920296869958051,
      "grad_norm": 1.4096795320510864,
      "learning_rate": 0.0005,
      "loss": 2.236907720565796,
      "step": 4909
    },
    {
      "epoch": 0.7921910293643111,
      "grad_norm": 1.0276256799697876,
      "learning_rate": 0.0005,
      "loss": 2.427997350692749,
      "step": 4910
    },
    {
      "epoch": 0.792352371732817,
      "grad_norm": 1.031556248664856,
      "learning_rate": 0.0005,
      "loss": 2.3514814376831055,
      "step": 4911
    },
    {
      "epoch": 0.792513714101323,
      "grad_norm": 1.2535488605499268,
      "learning_rate": 0.0005,
      "loss": 2.315953254699707,
      "step": 4912
    },
    {
      "epoch": 0.7926750564698289,
      "grad_norm": 0.9491416215896606,
      "learning_rate": 0.0005,
      "loss": 2.2951416969299316,
      "step": 4913
    },
    {
      "epoch": 0.7928363988383349,
      "grad_norm": 1.2707703113555908,
      "learning_rate": 0.0005,
      "loss": 2.251107692718506,
      "step": 4914
    },
    {
      "epoch": 0.792997741206841,
      "grad_norm": 1.0491389036178589,
      "learning_rate": 0.0005,
      "loss": 2.4421792030334473,
      "step": 4915
    },
    {
      "epoch": 0.7931590835753469,
      "grad_norm": 1.3303285837173462,
      "learning_rate": 0.0005,
      "loss": 2.690476894378662,
      "step": 4916
    },
    {
      "epoch": 0.7933204259438529,
      "grad_norm": 1.1615025997161865,
      "learning_rate": 0.0005,
      "loss": 2.241252899169922,
      "step": 4917
    },
    {
      "epoch": 0.7934817683123588,
      "grad_norm": 1.0747473239898682,
      "learning_rate": 0.0005,
      "loss": 2.4424848556518555,
      "step": 4918
    },
    {
      "epoch": 0.7936431106808648,
      "grad_norm": 1.0348073244094849,
      "learning_rate": 0.0005,
      "loss": 2.2880845069885254,
      "step": 4919
    },
    {
      "epoch": 0.7938044530493708,
      "grad_norm": 1.0977243185043335,
      "learning_rate": 0.0005,
      "loss": 2.2727034091949463,
      "step": 4920
    },
    {
      "epoch": 0.7939657954178767,
      "grad_norm": 0.9937576651573181,
      "learning_rate": 0.0005,
      "loss": 2.380612850189209,
      "step": 4921
    },
    {
      "epoch": 0.7941271377863827,
      "grad_norm": 1.0429956912994385,
      "learning_rate": 0.0005,
      "loss": 2.452012538909912,
      "step": 4922
    },
    {
      "epoch": 0.7942884801548886,
      "grad_norm": 0.9998892545700073,
      "learning_rate": 0.0005,
      "loss": 3.036637783050537,
      "step": 4923
    },
    {
      "epoch": 0.7944498225233947,
      "grad_norm": 1.9879385232925415,
      "learning_rate": 0.0005,
      "loss": 2.748729705810547,
      "step": 4924
    },
    {
      "epoch": 0.7946111648919006,
      "grad_norm": 0.8152735233306885,
      "learning_rate": 0.0005,
      "loss": 2.5578675270080566,
      "step": 4925
    },
    {
      "epoch": 0.7947725072604066,
      "grad_norm": 1.3945509195327759,
      "learning_rate": 0.0005,
      "loss": 3.057487964630127,
      "step": 4926
    },
    {
      "epoch": 0.7949338496289126,
      "grad_norm": 1.3151799440383911,
      "learning_rate": 0.0005,
      "loss": 2.4387733936309814,
      "step": 4927
    },
    {
      "epoch": 0.7950951919974185,
      "grad_norm": 1.0769224166870117,
      "learning_rate": 0.0005,
      "loss": 2.6571950912475586,
      "step": 4928
    },
    {
      "epoch": 0.7952565343659245,
      "grad_norm": 0.7871328592300415,
      "learning_rate": 0.0005,
      "loss": 2.152712106704712,
      "step": 4929
    },
    {
      "epoch": 0.7954178767344304,
      "grad_norm": 1.1259725093841553,
      "learning_rate": 0.0005,
      "loss": 2.487098217010498,
      "step": 4930
    },
    {
      "epoch": 0.7955792191029364,
      "grad_norm": 1.1982911825180054,
      "learning_rate": 0.0005,
      "loss": 2.5557265281677246,
      "step": 4931
    },
    {
      "epoch": 0.7957405614714425,
      "grad_norm": 0.8954219222068787,
      "learning_rate": 0.0005,
      "loss": 2.4626574516296387,
      "step": 4932
    },
    {
      "epoch": 0.7959019038399484,
      "grad_norm": 1.1646705865859985,
      "learning_rate": 0.0005,
      "loss": 2.4329123497009277,
      "step": 4933
    },
    {
      "epoch": 0.7960632462084544,
      "grad_norm": 1.024631381034851,
      "learning_rate": 0.0005,
      "loss": 2.700425624847412,
      "step": 4934
    },
    {
      "epoch": 0.7962245885769603,
      "grad_norm": 1.629607915878296,
      "learning_rate": 0.0005,
      "loss": 2.2622501850128174,
      "step": 4935
    },
    {
      "epoch": 0.7963859309454663,
      "grad_norm": 1.4579685926437378,
      "learning_rate": 0.0005,
      "loss": 2.7243270874023438,
      "step": 4936
    },
    {
      "epoch": 0.7965472733139722,
      "grad_norm": 1.0695019960403442,
      "learning_rate": 0.0005,
      "loss": 2.349435806274414,
      "step": 4937
    },
    {
      "epoch": 0.7967086156824782,
      "grad_norm": 1.2183924913406372,
      "learning_rate": 0.0005,
      "loss": 2.8862245082855225,
      "step": 4938
    },
    {
      "epoch": 0.7968699580509842,
      "grad_norm": 1.0680583715438843,
      "learning_rate": 0.0005,
      "loss": 2.395595073699951,
      "step": 4939
    },
    {
      "epoch": 0.7970313004194901,
      "grad_norm": 0.861362636089325,
      "learning_rate": 0.0005,
      "loss": 2.660428524017334,
      "step": 4940
    },
    {
      "epoch": 0.7971926427879962,
      "grad_norm": 1.1125693321228027,
      "learning_rate": 0.0005,
      "loss": 2.320725202560425,
      "step": 4941
    },
    {
      "epoch": 0.7973539851565021,
      "grad_norm": 1.1546952724456787,
      "learning_rate": 0.0005,
      "loss": 2.483896017074585,
      "step": 4942
    },
    {
      "epoch": 0.7975153275250081,
      "grad_norm": 1.6028671264648438,
      "learning_rate": 0.0005,
      "loss": 2.395249128341675,
      "step": 4943
    },
    {
      "epoch": 0.797676669893514,
      "grad_norm": 1.0817445516586304,
      "learning_rate": 0.0005,
      "loss": 2.6096808910369873,
      "step": 4944
    },
    {
      "epoch": 0.79783801226202,
      "grad_norm": 0.8880256414413452,
      "learning_rate": 0.0005,
      "loss": 2.7115700244903564,
      "step": 4945
    },
    {
      "epoch": 0.797999354630526,
      "grad_norm": 1.2632923126220703,
      "learning_rate": 0.0005,
      "loss": 2.367415428161621,
      "step": 4946
    },
    {
      "epoch": 0.7981606969990319,
      "grad_norm": 1.226621389389038,
      "learning_rate": 0.0005,
      "loss": 2.359393835067749,
      "step": 4947
    },
    {
      "epoch": 0.7983220393675379,
      "grad_norm": 1.2339285612106323,
      "learning_rate": 0.0005,
      "loss": 2.567802906036377,
      "step": 4948
    },
    {
      "epoch": 0.7984833817360439,
      "grad_norm": 1.3938720226287842,
      "learning_rate": 0.0005,
      "loss": 2.3183436393737793,
      "step": 4949
    },
    {
      "epoch": 0.7986447241045499,
      "grad_norm": 1.3473292589187622,
      "learning_rate": 0.0005,
      "loss": 2.7526962757110596,
      "step": 4950
    },
    {
      "epoch": 0.7988060664730559,
      "grad_norm": 1.2138787508010864,
      "learning_rate": 0.0005,
      "loss": 2.6055006980895996,
      "step": 4951
    },
    {
      "epoch": 0.7989674088415618,
      "grad_norm": 1.5830943584442139,
      "learning_rate": 0.0005,
      "loss": 2.259244918823242,
      "step": 4952
    },
    {
      "epoch": 0.7991287512100678,
      "grad_norm": 1.3217254877090454,
      "learning_rate": 0.0005,
      "loss": 3.0091166496276855,
      "step": 4953
    },
    {
      "epoch": 0.7992900935785737,
      "grad_norm": 1.2758795022964478,
      "learning_rate": 0.0005,
      "loss": 2.827038288116455,
      "step": 4954
    },
    {
      "epoch": 0.7994514359470797,
      "grad_norm": 1.269493818283081,
      "learning_rate": 0.0005,
      "loss": 2.557893753051758,
      "step": 4955
    },
    {
      "epoch": 0.7996127783155856,
      "grad_norm": 1.1956868171691895,
      "learning_rate": 0.0005,
      "loss": 2.740394115447998,
      "step": 4956
    },
    {
      "epoch": 0.7997741206840916,
      "grad_norm": 0.9806995391845703,
      "learning_rate": 0.0005,
      "loss": 2.1893150806427,
      "step": 4957
    },
    {
      "epoch": 0.7999354630525977,
      "grad_norm": 0.964911699295044,
      "learning_rate": 0.0005,
      "loss": 2.789311647415161,
      "step": 4958
    },
    {
      "epoch": 0.8000968054211036,
      "grad_norm": 1.1416707038879395,
      "learning_rate": 0.0005,
      "loss": 2.679680347442627,
      "step": 4959
    },
    {
      "epoch": 0.8002581477896096,
      "grad_norm": 1.3212330341339111,
      "learning_rate": 0.0005,
      "loss": 2.7542943954467773,
      "step": 4960
    },
    {
      "epoch": 0.8004194901581155,
      "grad_norm": 1.2885029315948486,
      "learning_rate": 0.0004999991963500667,
      "loss": 2.760896682739258,
      "step": 4961
    },
    {
      "epoch": 0.8005808325266215,
      "grad_norm": 0.9627770781517029,
      "learning_rate": 0.0004999967854054339,
      "loss": 2.619070291519165,
      "step": 4962
    },
    {
      "epoch": 0.8007421748951274,
      "grad_norm": 1.1110098361968994,
      "learning_rate": 0.0004999927671816018,
      "loss": 2.9021196365356445,
      "step": 4963
    },
    {
      "epoch": 0.8009035172636334,
      "grad_norm": 1.0582584142684937,
      "learning_rate": 0.0004999871417044044,
      "loss": 2.613558530807495,
      "step": 4964
    },
    {
      "epoch": 0.8010648596321394,
      "grad_norm": 1.1939748525619507,
      "learning_rate": 0.0004999799090100092,
      "loss": 2.7491893768310547,
      "step": 4965
    },
    {
      "epoch": 0.8012262020006453,
      "grad_norm": 1.3736604452133179,
      "learning_rate": 0.0004999710691449165,
      "loss": 2.3811933994293213,
      "step": 4966
    },
    {
      "epoch": 0.8013875443691514,
      "grad_norm": 0.9629430770874023,
      "learning_rate": 0.0004999606221659595,
      "loss": 2.66998291015625,
      "step": 4967
    },
    {
      "epoch": 0.8015488867376573,
      "grad_norm": 1.1242125034332275,
      "learning_rate": 0.000499948568140304,
      "loss": 2.8831048011779785,
      "step": 4968
    },
    {
      "epoch": 0.8017102291061633,
      "grad_norm": 1.721794843673706,
      "learning_rate": 0.0004999349071454477,
      "loss": 2.358563184738159,
      "step": 4969
    },
    {
      "epoch": 0.8018715714746693,
      "grad_norm": 0.9743290543556213,
      "learning_rate": 0.0004999196392692198,
      "loss": 2.1140589714050293,
      "step": 4970
    },
    {
      "epoch": 0.8020329138431752,
      "grad_norm": 1.0038601160049438,
      "learning_rate": 0.0004999027646097807,
      "loss": 2.777146816253662,
      "step": 4971
    },
    {
      "epoch": 0.8021942562116812,
      "grad_norm": 1.0279964208602905,
      "learning_rate": 0.0004998842832756209,
      "loss": 2.9261884689331055,
      "step": 4972
    },
    {
      "epoch": 0.8023555985801871,
      "grad_norm": 1.0038455724716187,
      "learning_rate": 0.0004998641953855604,
      "loss": 2.161945343017578,
      "step": 4973
    },
    {
      "epoch": 0.8025169409486931,
      "grad_norm": 0.8763434886932373,
      "learning_rate": 0.0004998425010687484,
      "loss": 2.2574222087860107,
      "step": 4974
    },
    {
      "epoch": 0.8026782833171991,
      "grad_norm": 1.1947427988052368,
      "learning_rate": 0.000499819200464662,
      "loss": 2.237081527709961,
      "step": 4975
    },
    {
      "epoch": 0.8028396256857051,
      "grad_norm": 0.9257453083992004,
      "learning_rate": 0.0004997942937231054,
      "loss": 2.329653263092041,
      "step": 4976
    },
    {
      "epoch": 0.8030009680542111,
      "grad_norm": 1.2956626415252686,
      "learning_rate": 0.000499767781004209,
      "loss": 2.52846097946167,
      "step": 4977
    },
    {
      "epoch": 0.803162310422717,
      "grad_norm": 1.2623580694198608,
      "learning_rate": 0.0004997396624784284,
      "loss": 2.5113236904144287,
      "step": 4978
    },
    {
      "epoch": 0.803323652791223,
      "grad_norm": 4.6904449462890625,
      "learning_rate": 0.0004997099383265432,
      "loss": 2.238696575164795,
      "step": 4979
    },
    {
      "epoch": 0.8034849951597289,
      "grad_norm": 1.625442624092102,
      "learning_rate": 0.0004996786087396558,
      "loss": 2.4194438457489014,
      "step": 4980
    },
    {
      "epoch": 0.8036463375282349,
      "grad_norm": 1.0249446630477905,
      "learning_rate": 0.0004996456739191905,
      "loss": 2.035515308380127,
      "step": 4981
    },
    {
      "epoch": 0.8038076798967408,
      "grad_norm": 1.005046010017395,
      "learning_rate": 0.0004996111340768917,
      "loss": 2.3964970111846924,
      "step": 4982
    },
    {
      "epoch": 0.8039690222652468,
      "grad_norm": 1.239622950553894,
      "learning_rate": 0.0004995749894348232,
      "loss": 2.418114423751831,
      "step": 4983
    },
    {
      "epoch": 0.8041303646337529,
      "grad_norm": 1.205029010772705,
      "learning_rate": 0.0004995372402253657,
      "loss": 2.118318557739258,
      "step": 4984
    },
    {
      "epoch": 0.8042917070022588,
      "grad_norm": 1.2226656675338745,
      "learning_rate": 0.0004994978866912168,
      "loss": 2.2875187397003174,
      "step": 4985
    },
    {
      "epoch": 0.8044530493707648,
      "grad_norm": 1.346955418586731,
      "learning_rate": 0.000499456929085388,
      "loss": 2.9359123706817627,
      "step": 4986
    },
    {
      "epoch": 0.8046143917392707,
      "grad_norm": 1.3896229267120361,
      "learning_rate": 0.0004994143676712038,
      "loss": 2.365868091583252,
      "step": 4987
    },
    {
      "epoch": 0.8047757341077767,
      "grad_norm": 1.1992013454437256,
      "learning_rate": 0.0004993702027223004,
      "loss": 2.431342840194702,
      "step": 4988
    },
    {
      "epoch": 0.8049370764762827,
      "grad_norm": 0.8935186266899109,
      "learning_rate": 0.0004993244345226227,
      "loss": 2.499390125274658,
      "step": 4989
    },
    {
      "epoch": 0.8050984188447886,
      "grad_norm": 1.859905481338501,
      "learning_rate": 0.0004992770633664239,
      "loss": 2.8093156814575195,
      "step": 4990
    },
    {
      "epoch": 0.8052597612132946,
      "grad_norm": 1.051729440689087,
      "learning_rate": 0.0004992280895582623,
      "loss": 2.341883659362793,
      "step": 4991
    },
    {
      "epoch": 0.8054211035818006,
      "grad_norm": 1.4460991621017456,
      "learning_rate": 0.0004991775134130004,
      "loss": 2.396693706512451,
      "step": 4992
    },
    {
      "epoch": 0.8055824459503066,
      "grad_norm": 1.2552375793457031,
      "learning_rate": 0.0004991253352558024,
      "loss": 2.3051605224609375,
      "step": 4993
    },
    {
      "epoch": 0.8057437883188125,
      "grad_norm": 1.0653140544891357,
      "learning_rate": 0.0004990715554221321,
      "loss": 2.550072193145752,
      "step": 4994
    },
    {
      "epoch": 0.8059051306873185,
      "grad_norm": 1.2431329488754272,
      "learning_rate": 0.0004990161742577506,
      "loss": 2.9720168113708496,
      "step": 4995
    },
    {
      "epoch": 0.8060664730558245,
      "grad_norm": 1.3493221998214722,
      "learning_rate": 0.0004989591921187147,
      "loss": 2.380251884460449,
      "step": 4996
    },
    {
      "epoch": 0.8062278154243304,
      "grad_norm": 2.329496145248413,
      "learning_rate": 0.0004989006093713737,
      "loss": 2.6977834701538086,
      "step": 4997
    },
    {
      "epoch": 0.8063891577928364,
      "grad_norm": 1.4595279693603516,
      "learning_rate": 0.0004988404263923679,
      "loss": 2.493877410888672,
      "step": 4998
    },
    {
      "epoch": 0.8065505001613423,
      "grad_norm": 0.9957364201545715,
      "learning_rate": 0.0004987786435686257,
      "loss": 2.124185562133789,
      "step": 4999
    },
    {
      "epoch": 0.8067118425298483,
      "grad_norm": 1.2285385131835938,
      "learning_rate": 0.000498715261297361,
      "loss": 3.0173537731170654,
      "step": 5000
    },
    {
      "epoch": 0.8068731848983544,
      "grad_norm": 0.9999834895133972,
      "learning_rate": 0.0004986502799860712,
      "loss": 2.2842323780059814,
      "step": 5001
    },
    {
      "epoch": 0.8070345272668603,
      "grad_norm": 1.0894867181777954,
      "learning_rate": 0.0004985837000525343,
      "loss": 2.629405975341797,
      "step": 5002
    },
    {
      "epoch": 0.8071958696353663,
      "grad_norm": 1.1571074724197388,
      "learning_rate": 0.0004985155219248057,
      "loss": 3.2588400840759277,
      "step": 5003
    },
    {
      "epoch": 0.8073572120038722,
      "grad_norm": 1.3294461965560913,
      "learning_rate": 0.0004984457460412162,
      "loss": 2.6650795936584473,
      "step": 5004
    },
    {
      "epoch": 0.8075185543723782,
      "grad_norm": 1.350268006324768,
      "learning_rate": 0.0004983743728503691,
      "loss": 2.0063722133636475,
      "step": 5005
    },
    {
      "epoch": 0.8076798967408841,
      "grad_norm": 1.1232199668884277,
      "learning_rate": 0.0004983014028111367,
      "loss": 2.8257288932800293,
      "step": 5006
    },
    {
      "epoch": 0.8078412391093901,
      "grad_norm": 3.011221408843994,
      "learning_rate": 0.0004982268363926579,
      "loss": 2.399345874786377,
      "step": 5007
    },
    {
      "epoch": 0.808002581477896,
      "grad_norm": 1.1105473041534424,
      "learning_rate": 0.0004981506740743351,
      "loss": 2.5124881267547607,
      "step": 5008
    },
    {
      "epoch": 0.808163923846402,
      "grad_norm": 1.2113012075424194,
      "learning_rate": 0.0004980729163458311,
      "loss": 2.408893585205078,
      "step": 5009
    },
    {
      "epoch": 0.8083252662149081,
      "grad_norm": 1.1591178178787231,
      "learning_rate": 0.0004979935637070659,
      "loss": 2.443493127822876,
      "step": 5010
    },
    {
      "epoch": 0.808486608583414,
      "grad_norm": 1.1209735870361328,
      "learning_rate": 0.0004979126166682133,
      "loss": 2.452153205871582,
      "step": 5011
    },
    {
      "epoch": 0.80864795095192,
      "grad_norm": 1.1598180532455444,
      "learning_rate": 0.0004978300757496981,
      "loss": 2.4563746452331543,
      "step": 5012
    },
    {
      "epoch": 0.808809293320426,
      "grad_norm": 1.4054261445999146,
      "learning_rate": 0.0004977459414821922,
      "loss": 2.831145763397217,
      "step": 5013
    },
    {
      "epoch": 0.8089706356889319,
      "grad_norm": 1.0368999242782593,
      "learning_rate": 0.0004976602144066116,
      "loss": 2.5287671089172363,
      "step": 5014
    },
    {
      "epoch": 0.8091319780574379,
      "grad_norm": 1.4524691104888916,
      "learning_rate": 0.000497572895074113,
      "loss": 2.3673880100250244,
      "step": 5015
    },
    {
      "epoch": 0.8092933204259438,
      "grad_norm": 1.5894577503204346,
      "learning_rate": 0.0004974839840460895,
      "loss": 2.522307872772217,
      "step": 5016
    },
    {
      "epoch": 0.8094546627944498,
      "grad_norm": 1.011986494064331,
      "learning_rate": 0.0004973934818941679,
      "loss": 2.6070265769958496,
      "step": 5017
    },
    {
      "epoch": 0.8096160051629558,
      "grad_norm": 1.0950143337249756,
      "learning_rate": 0.0004973013892002047,
      "loss": 1.8840980529785156,
      "step": 5018
    },
    {
      "epoch": 0.8097773475314618,
      "grad_norm": 1.2908555269241333,
      "learning_rate": 0.0004972077065562821,
      "loss": 2.2252113819122314,
      "step": 5019
    },
    {
      "epoch": 0.8099386898999678,
      "grad_norm": 1.0480070114135742,
      "learning_rate": 0.0004971124345647046,
      "loss": 2.4522228240966797,
      "step": 5020
    },
    {
      "epoch": 0.8101000322684737,
      "grad_norm": 1.9526662826538086,
      "learning_rate": 0.0004970155738379948,
      "loss": 2.370626211166382,
      "step": 5021
    },
    {
      "epoch": 0.8102613746369797,
      "grad_norm": 1.0358330011367798,
      "learning_rate": 0.0004969171249988895,
      "loss": 2.4052045345306396,
      "step": 5022
    },
    {
      "epoch": 0.8104227170054856,
      "grad_norm": 1.6200841665267944,
      "learning_rate": 0.0004968170886803362,
      "loss": 2.3285417556762695,
      "step": 5023
    },
    {
      "epoch": 0.8105840593739916,
      "grad_norm": 1.2142094373703003,
      "learning_rate": 0.0004967154655254881,
      "loss": 2.422004222869873,
      "step": 5024
    },
    {
      "epoch": 0.8107454017424975,
      "grad_norm": 1.4130337238311768,
      "learning_rate": 0.0004966122561877008,
      "loss": 2.5099282264709473,
      "step": 5025
    },
    {
      "epoch": 0.8109067441110035,
      "grad_norm": 1.202347993850708,
      "learning_rate": 0.0004965074613305277,
      "loss": 2.003131866455078,
      "step": 5026
    },
    {
      "epoch": 0.8110680864795096,
      "grad_norm": 1.2566475868225098,
      "learning_rate": 0.0004964010816277159,
      "loss": 2.7875280380249023,
      "step": 5027
    },
    {
      "epoch": 0.8112294288480155,
      "grad_norm": 1.3079787492752075,
      "learning_rate": 0.0004962931177632017,
      "loss": 2.2087044715881348,
      "step": 5028
    },
    {
      "epoch": 0.8113907712165215,
      "grad_norm": 1.1615170240402222,
      "learning_rate": 0.0004961835704311064,
      "loss": 2.0380053520202637,
      "step": 5029
    },
    {
      "epoch": 0.8115521135850274,
      "grad_norm": 1.027773380279541,
      "learning_rate": 0.0004960724403357315,
      "loss": 2.699863910675049,
      "step": 5030
    },
    {
      "epoch": 0.8117134559535334,
      "grad_norm": 1.066609501838684,
      "learning_rate": 0.0004959597281915546,
      "loss": 2.056427001953125,
      "step": 5031
    },
    {
      "epoch": 0.8118747983220393,
      "grad_norm": 1.1516634225845337,
      "learning_rate": 0.0004958454347232246,
      "loss": 2.575050115585327,
      "step": 5032
    },
    {
      "epoch": 0.8120361406905453,
      "grad_norm": 1.1390775442123413,
      "learning_rate": 0.000495729560665557,
      "loss": 2.705580711364746,
      "step": 5033
    },
    {
      "epoch": 0.8121974830590513,
      "grad_norm": 0.8676235675811768,
      "learning_rate": 0.0004956121067635292,
      "loss": 2.2208733558654785,
      "step": 5034
    },
    {
      "epoch": 0.8123588254275573,
      "grad_norm": 1.2907614707946777,
      "learning_rate": 0.0004954930737722758,
      "loss": 2.377892017364502,
      "step": 5035
    },
    {
      "epoch": 0.8125201677960633,
      "grad_norm": 1.451414704322815,
      "learning_rate": 0.0004953724624570837,
      "loss": 2.317471504211426,
      "step": 5036
    },
    {
      "epoch": 0.8126815101645692,
      "grad_norm": 1.1648547649383545,
      "learning_rate": 0.0004952502735933869,
      "loss": 2.557710647583008,
      "step": 5037
    },
    {
      "epoch": 0.8128428525330752,
      "grad_norm": 1.1856048107147217,
      "learning_rate": 0.0004951265079667622,
      "loss": 2.373631477355957,
      "step": 5038
    },
    {
      "epoch": 0.8130041949015812,
      "grad_norm": 1.470803141593933,
      "learning_rate": 0.0004950011663729233,
      "loss": 2.396287441253662,
      "step": 5039
    },
    {
      "epoch": 0.8131655372700871,
      "grad_norm": 1.0837141275405884,
      "learning_rate": 0.0004948742496177164,
      "loss": 2.540599822998047,
      "step": 5040
    },
    {
      "epoch": 0.8133268796385931,
      "grad_norm": 0.8727065324783325,
      "learning_rate": 0.0004947457585171148,
      "loss": 2.3657498359680176,
      "step": 5041
    },
    {
      "epoch": 0.813488222007099,
      "grad_norm": 1.0586928129196167,
      "learning_rate": 0.0004946156938972131,
      "loss": 2.380120038986206,
      "step": 5042
    },
    {
      "epoch": 0.813649564375605,
      "grad_norm": 1.1427127122879028,
      "learning_rate": 0.0004944840565942229,
      "loss": 2.7598395347595215,
      "step": 5043
    },
    {
      "epoch": 0.813810906744111,
      "grad_norm": 1.0714024305343628,
      "learning_rate": 0.0004943508474544667,
      "loss": 2.3996968269348145,
      "step": 5044
    },
    {
      "epoch": 0.813972249112617,
      "grad_norm": 1.3819575309753418,
      "learning_rate": 0.0004942160673343725,
      "loss": 2.4638471603393555,
      "step": 5045
    },
    {
      "epoch": 0.814133591481123,
      "grad_norm": 1.1048247814178467,
      "learning_rate": 0.0004940797171004687,
      "loss": 2.137380838394165,
      "step": 5046
    },
    {
      "epoch": 0.8142949338496289,
      "grad_norm": 1.4358699321746826,
      "learning_rate": 0.0004939417976293779,
      "loss": 2.3508684635162354,
      "step": 5047
    },
    {
      "epoch": 0.8144562762181349,
      "grad_norm": 0.9086236357688904,
      "learning_rate": 0.0004938023098078121,
      "loss": 2.103971481323242,
      "step": 5048
    },
    {
      "epoch": 0.8146176185866408,
      "grad_norm": 1.178072452545166,
      "learning_rate": 0.0004936612545325665,
      "loss": 2.160902976989746,
      "step": 5049
    },
    {
      "epoch": 0.8147789609551468,
      "grad_norm": 1.0027061700820923,
      "learning_rate": 0.0004935186327105132,
      "loss": 2.3973276615142822,
      "step": 5050
    },
    {
      "epoch": 0.8149403033236527,
      "grad_norm": 2.5972952842712402,
      "learning_rate": 0.0004933744452585967,
      "loss": 2.57777738571167,
      "step": 5051
    },
    {
      "epoch": 0.8151016456921588,
      "grad_norm": 1.1378356218338013,
      "learning_rate": 0.0004932286931038266,
      "loss": 2.606384754180908,
      "step": 5052
    },
    {
      "epoch": 0.8152629880606648,
      "grad_norm": 1.452210783958435,
      "learning_rate": 0.0004930813771832728,
      "loss": 2.6424927711486816,
      "step": 5053
    },
    {
      "epoch": 0.8154243304291707,
      "grad_norm": 1.388061285018921,
      "learning_rate": 0.0004929324984440586,
      "loss": 2.6585159301757812,
      "step": 5054
    },
    {
      "epoch": 0.8155856727976767,
      "grad_norm": 1.2199944257736206,
      "learning_rate": 0.0004927820578433552,
      "loss": 2.2588069438934326,
      "step": 5055
    },
    {
      "epoch": 0.8157470151661826,
      "grad_norm": 1.081552267074585,
      "learning_rate": 0.000492630056348375,
      "loss": 2.593531370162964,
      "step": 5056
    },
    {
      "epoch": 0.8159083575346886,
      "grad_norm": 1.3624427318572998,
      "learning_rate": 0.0004924764949363664,
      "loss": 2.6339173316955566,
      "step": 5057
    },
    {
      "epoch": 0.8160696999031946,
      "grad_norm": 1.5309697389602661,
      "learning_rate": 0.0004923213745946059,
      "loss": 2.5291078090667725,
      "step": 5058
    },
    {
      "epoch": 0.8162310422717005,
      "grad_norm": 1.4418823719024658,
      "learning_rate": 0.0004921646963203933,
      "loss": 2.4674954414367676,
      "step": 5059
    },
    {
      "epoch": 0.8163923846402065,
      "grad_norm": 0.9723765850067139,
      "learning_rate": 0.0004920064611210444,
      "loss": 2.601851224899292,
      "step": 5060
    },
    {
      "epoch": 0.8165537270087125,
      "grad_norm": 0.9899846911430359,
      "learning_rate": 0.000491846670013885,
      "loss": 2.622255802154541,
      "step": 5061
    },
    {
      "epoch": 0.8167150693772185,
      "grad_norm": 1.595611333847046,
      "learning_rate": 0.0004916853240262439,
      "loss": 2.568796396255493,
      "step": 5062
    },
    {
      "epoch": 0.8168764117457245,
      "grad_norm": 1.1788462400436401,
      "learning_rate": 0.0004915224241954467,
      "loss": 2.2101399898529053,
      "step": 5063
    },
    {
      "epoch": 0.8170377541142304,
      "grad_norm": 1.1868129968643188,
      "learning_rate": 0.0004913579715688088,
      "loss": 2.3218026161193848,
      "step": 5064
    },
    {
      "epoch": 0.8171990964827364,
      "grad_norm": 1.0728408098220825,
      "learning_rate": 0.000491191967203629,
      "loss": 2.4594974517822266,
      "step": 5065
    },
    {
      "epoch": 0.8173604388512423,
      "grad_norm": 1.1917089223861694,
      "learning_rate": 0.0004910244121671825,
      "loss": 2.6770761013031006,
      "step": 5066
    },
    {
      "epoch": 0.8175217812197483,
      "grad_norm": 0.8368050456047058,
      "learning_rate": 0.0004908553075367141,
      "loss": 2.428867816925049,
      "step": 5067
    },
    {
      "epoch": 0.8176831235882542,
      "grad_norm": 1.1881709098815918,
      "learning_rate": 0.000490684654399431,
      "loss": 2.436087131500244,
      "step": 5068
    },
    {
      "epoch": 0.8178444659567602,
      "grad_norm": 1.0137403011322021,
      "learning_rate": 0.0004905124538524964,
      "loss": 2.1032369136810303,
      "step": 5069
    },
    {
      "epoch": 0.8180058083252663,
      "grad_norm": 1.4139846563339233,
      "learning_rate": 0.000490338707003022,
      "loss": 2.1770987510681152,
      "step": 5070
    },
    {
      "epoch": 0.8181671506937722,
      "grad_norm": 1.1989284753799438,
      "learning_rate": 0.0004901634149680608,
      "loss": 3.0986862182617188,
      "step": 5071
    },
    {
      "epoch": 0.8183284930622782,
      "grad_norm": 1.1556740999221802,
      "learning_rate": 0.0004899865788746005,
      "loss": 2.501673936843872,
      "step": 5072
    },
    {
      "epoch": 0.8184898354307841,
      "grad_norm": 1.908079743385315,
      "learning_rate": 0.0004898081998595554,
      "loss": 2.9694862365722656,
      "step": 5073
    },
    {
      "epoch": 0.8186511777992901,
      "grad_norm": 1.231934666633606,
      "learning_rate": 0.0004896282790697599,
      "loss": 2.213773012161255,
      "step": 5074
    },
    {
      "epoch": 0.818812520167796,
      "grad_norm": 1.276483178138733,
      "learning_rate": 0.0004894468176619605,
      "loss": 2.157528877258301,
      "step": 5075
    },
    {
      "epoch": 0.818973862536302,
      "grad_norm": 1.2059271335601807,
      "learning_rate": 0.000489263816802809,
      "loss": 2.457543134689331,
      "step": 5076
    },
    {
      "epoch": 0.819135204904808,
      "grad_norm": 1.2823854684829712,
      "learning_rate": 0.0004890792776688543,
      "loss": 2.1579341888427734,
      "step": 5077
    },
    {
      "epoch": 0.819296547273314,
      "grad_norm": 0.9937081933021545,
      "learning_rate": 0.0004888932014465352,
      "loss": 1.9926068782806396,
      "step": 5078
    },
    {
      "epoch": 0.81945788964182,
      "grad_norm": 1.3487948179244995,
      "learning_rate": 0.000488705589332173,
      "loss": 2.4476847648620605,
      "step": 5079
    },
    {
      "epoch": 0.8196192320103259,
      "grad_norm": 0.901314377784729,
      "learning_rate": 0.0004885164425319634,
      "loss": 2.3350300788879395,
      "step": 5080
    },
    {
      "epoch": 0.8197805743788319,
      "grad_norm": 1.0377498865127563,
      "learning_rate": 0.0004883257622619688,
      "loss": 2.0428085327148438,
      "step": 5081
    },
    {
      "epoch": 0.8199419167473379,
      "grad_norm": 0.9137289524078369,
      "learning_rate": 0.0004881335497481108,
      "loss": 2.1821892261505127,
      "step": 5082
    },
    {
      "epoch": 0.8201032591158438,
      "grad_norm": 1.1913241147994995,
      "learning_rate": 0.0004879398062261619,
      "loss": 2.4045979976654053,
      "step": 5083
    },
    {
      "epoch": 0.8202646014843498,
      "grad_norm": 1.0582728385925293,
      "learning_rate": 0.0004877445329417379,
      "loss": 2.2318997383117676,
      "step": 5084
    },
    {
      "epoch": 0.8204259438528557,
      "grad_norm": 1.219377875328064,
      "learning_rate": 0.00048754773115028974,
      "loss": 2.676797866821289,
      "step": 5085
    },
    {
      "epoch": 0.8205872862213617,
      "grad_norm": 1.10997474193573,
      "learning_rate": 0.0004873494021170953,
      "loss": 2.714531421661377,
      "step": 5086
    },
    {
      "epoch": 0.8207486285898677,
      "grad_norm": 1.296379804611206,
      "learning_rate": 0.00048714954711725155,
      "loss": 2.138277292251587,
      "step": 5087
    },
    {
      "epoch": 0.8209099709583737,
      "grad_norm": 0.9446429014205933,
      "learning_rate": 0.00048694816743566615,
      "loss": 2.4706063270568848,
      "step": 5088
    },
    {
      "epoch": 0.8210713133268797,
      "grad_norm": 1.133078694343567,
      "learning_rate": 0.00048674526436704926,
      "loss": 2.178187131881714,
      "step": 5089
    },
    {
      "epoch": 0.8212326556953856,
      "grad_norm": 1.1192845106124878,
      "learning_rate": 0.00048654083921590514,
      "loss": 2.180962562561035,
      "step": 5090
    },
    {
      "epoch": 0.8213939980638916,
      "grad_norm": 1.2598427534103394,
      "learning_rate": 0.0004863348932965238,
      "loss": 2.509986639022827,
      "step": 5091
    },
    {
      "epoch": 0.8215553404323975,
      "grad_norm": 1.1375079154968262,
      "learning_rate": 0.00048612742793297285,
      "loss": 2.434051513671875,
      "step": 5092
    },
    {
      "epoch": 0.8217166828009035,
      "grad_norm": 1.1824203729629517,
      "learning_rate": 0.0004859184444590882,
      "loss": 2.343477487564087,
      "step": 5093
    },
    {
      "epoch": 0.8218780251694094,
      "grad_norm": 1.1517412662506104,
      "learning_rate": 0.0004857079442184665,
      "loss": 2.34618878364563,
      "step": 5094
    },
    {
      "epoch": 0.8220393675379155,
      "grad_norm": 1.525386929512024,
      "learning_rate": 0.0004854959285644557,
      "loss": 2.9125022888183594,
      "step": 5095
    },
    {
      "epoch": 0.8222007099064215,
      "grad_norm": 1.0244569778442383,
      "learning_rate": 0.0004852823988601468,
      "loss": 1.8624000549316406,
      "step": 5096
    },
    {
      "epoch": 0.8223620522749274,
      "grad_norm": 1.1666356325149536,
      "learning_rate": 0.0004850673564783647,
      "loss": 2.654102325439453,
      "step": 5097
    },
    {
      "epoch": 0.8225233946434334,
      "grad_norm": 0.9665104746818542,
      "learning_rate": 0.00048485080280166005,
      "loss": 2.3364205360412598,
      "step": 5098
    },
    {
      "epoch": 0.8226847370119393,
      "grad_norm": 1.495004415512085,
      "learning_rate": 0.00048463273922229934,
      "loss": 2.745492935180664,
      "step": 5099
    },
    {
      "epoch": 0.8228460793804453,
      "grad_norm": 1.0685641765594482,
      "learning_rate": 0.00048441316714225696,
      "loss": 2.4743432998657227,
      "step": 5100
    },
    {
      "epoch": 0.8230074217489513,
      "grad_norm": 0.851396918296814,
      "learning_rate": 0.00048419208797320564,
      "loss": 2.2518296241760254,
      "step": 5101
    },
    {
      "epoch": 0.8231687641174572,
      "grad_norm": 0.9652995467185974,
      "learning_rate": 0.00048396950313650734,
      "loss": 2.9258310794830322,
      "step": 5102
    },
    {
      "epoch": 0.8233301064859632,
      "grad_norm": 1.2983580827713013,
      "learning_rate": 0.0004837454140632045,
      "loss": 2.572739839553833,
      "step": 5103
    },
    {
      "epoch": 0.8234914488544692,
      "grad_norm": 1.023155689239502,
      "learning_rate": 0.0004835198221940104,
      "loss": 2.4504756927490234,
      "step": 5104
    },
    {
      "epoch": 0.8236527912229752,
      "grad_norm": 0.9192457795143127,
      "learning_rate": 0.0004832927289793001,
      "loss": 2.832460880279541,
      "step": 5105
    },
    {
      "epoch": 0.8238141335914811,
      "grad_norm": 0.888702929019928,
      "learning_rate": 0.0004830641358791014,
      "loss": 2.3750181198120117,
      "step": 5106
    },
    {
      "epoch": 0.8239754759599871,
      "grad_norm": 0.9891437888145447,
      "learning_rate": 0.0004828340443630846,
      "loss": 1.9034643173217773,
      "step": 5107
    },
    {
      "epoch": 0.8241368183284931,
      "grad_norm": 1.3014942407608032,
      "learning_rate": 0.00048260245591055426,
      "loss": 2.7205922603607178,
      "step": 5108
    },
    {
      "epoch": 0.824298160696999,
      "grad_norm": 1.112025499343872,
      "learning_rate": 0.0004823693720104386,
      "loss": 2.627366781234741,
      "step": 5109
    },
    {
      "epoch": 0.824459503065505,
      "grad_norm": 1.5216879844665527,
      "learning_rate": 0.00048213479416128047,
      "loss": 1.7685296535491943,
      "step": 5110
    },
    {
      "epoch": 0.8246208454340109,
      "grad_norm": 1.2303310632705688,
      "learning_rate": 0.0004818987238712278,
      "loss": 2.273158073425293,
      "step": 5111
    },
    {
      "epoch": 0.8247821878025169,
      "grad_norm": 1.2195072174072266,
      "learning_rate": 0.0004816611626580234,
      "loss": 2.366044521331787,
      "step": 5112
    },
    {
      "epoch": 0.824943530171023,
      "grad_norm": 1.335150957107544,
      "learning_rate": 0.0004814221120489958,
      "loss": 2.580305814743042,
      "step": 5113
    },
    {
      "epoch": 0.8251048725395289,
      "grad_norm": 1.0730060338974,
      "learning_rate": 0.000481181573581049,
      "loss": 2.6909799575805664,
      "step": 5114
    },
    {
      "epoch": 0.8252662149080349,
      "grad_norm": 1.1072883605957031,
      "learning_rate": 0.00048093954880065276,
      "loss": 2.1189513206481934,
      "step": 5115
    },
    {
      "epoch": 0.8254275572765408,
      "grad_norm": 1.0026805400848389,
      "learning_rate": 0.00048069603926383277,
      "loss": 2.6631789207458496,
      "step": 5116
    },
    {
      "epoch": 0.8255888996450468,
      "grad_norm": 0.9673439264297485,
      "learning_rate": 0.00048045104653616035,
      "loss": 2.7505767345428467,
      "step": 5117
    },
    {
      "epoch": 0.8257502420135527,
      "grad_norm": 1.400573492050171,
      "learning_rate": 0.00048020457219274265,
      "loss": 2.350564956665039,
      "step": 5118
    },
    {
      "epoch": 0.8259115843820587,
      "grad_norm": 1.1340628862380981,
      "learning_rate": 0.0004799566178182123,
      "loss": 2.2941596508026123,
      "step": 5119
    },
    {
      "epoch": 0.8260729267505647,
      "grad_norm": 0.8868634104728699,
      "learning_rate": 0.00047970718500671763,
      "loss": 2.113722324371338,
      "step": 5120
    },
    {
      "epoch": 0.8262342691190707,
      "grad_norm": 1.0581873655319214,
      "learning_rate": 0.00047945627536191167,
      "loss": 2.3935623168945312,
      "step": 5121
    },
    {
      "epoch": 0.8263956114875767,
      "grad_norm": 0.9025821685791016,
      "learning_rate": 0.0004792038904969427,
      "loss": 2.3645358085632324,
      "step": 5122
    },
    {
      "epoch": 0.8265569538560826,
      "grad_norm": 0.9127691984176636,
      "learning_rate": 0.0004789500320344434,
      "loss": 2.4707834720611572,
      "step": 5123
    },
    {
      "epoch": 0.8267182962245886,
      "grad_norm": 1.266373872756958,
      "learning_rate": 0.0004786947016065205,
      "loss": 2.3156003952026367,
      "step": 5124
    },
    {
      "epoch": 0.8268796385930945,
      "grad_norm": 0.8202945590019226,
      "learning_rate": 0.00047843790085474417,
      "loss": 2.4486117362976074,
      "step": 5125
    },
    {
      "epoch": 0.8270409809616005,
      "grad_norm": 1.0947177410125732,
      "learning_rate": 0.0004781796314301375,
      "loss": 2.0960376262664795,
      "step": 5126
    },
    {
      "epoch": 0.8272023233301065,
      "grad_norm": 0.8692250847816467,
      "learning_rate": 0.00047791989499316654,
      "loss": 2.4433813095092773,
      "step": 5127
    },
    {
      "epoch": 0.8273636656986124,
      "grad_norm": 1.166074514389038,
      "learning_rate": 0.00047765869321372834,
      "loss": 2.536611557006836,
      "step": 5128
    },
    {
      "epoch": 0.8275250080671184,
      "grad_norm": 1.224955677986145,
      "learning_rate": 0.0004773960277711413,
      "loss": 2.1840620040893555,
      "step": 5129
    },
    {
      "epoch": 0.8276863504356244,
      "grad_norm": 1.0184613466262817,
      "learning_rate": 0.00047713190035413396,
      "loss": 1.9402122497558594,
      "step": 5130
    },
    {
      "epoch": 0.8278476928041304,
      "grad_norm": 1.7287570238113403,
      "learning_rate": 0.0004768663126608342,
      "loss": 2.194319486618042,
      "step": 5131
    },
    {
      "epoch": 0.8280090351726364,
      "grad_norm": 0.9470556378364563,
      "learning_rate": 0.0004765992663987582,
      "loss": 2.4825377464294434,
      "step": 5132
    },
    {
      "epoch": 0.8281703775411423,
      "grad_norm": 1.0997997522354126,
      "learning_rate": 0.00047633076328479975,
      "loss": 2.5832080841064453,
      "step": 5133
    },
    {
      "epoch": 0.8283317199096483,
      "grad_norm": 1.283823847770691,
      "learning_rate": 0.00047606080504521885,
      "loss": 2.0650501251220703,
      "step": 5134
    },
    {
      "epoch": 0.8284930622781542,
      "grad_norm": 0.971722424030304,
      "learning_rate": 0.0004757893934156309,
      "loss": 2.315830707550049,
      "step": 5135
    },
    {
      "epoch": 0.8286544046466602,
      "grad_norm": 1.0511229038238525,
      "learning_rate": 0.0004755165301409954,
      "loss": 2.1404473781585693,
      "step": 5136
    },
    {
      "epoch": 0.8288157470151661,
      "grad_norm": 1.0069866180419922,
      "learning_rate": 0.00047524221697560476,
      "loss": 2.3322458267211914,
      "step": 5137
    },
    {
      "epoch": 0.8289770893836722,
      "grad_norm": 1.4307712316513062,
      "learning_rate": 0.00047496645568307304,
      "loss": 2.543541193008423,
      "step": 5138
    },
    {
      "epoch": 0.8291384317521782,
      "grad_norm": 1.044577717781067,
      "learning_rate": 0.00047468924803632464,
      "loss": 2.1599879264831543,
      "step": 5139
    },
    {
      "epoch": 0.8292997741206841,
      "grad_norm": 1.2216347455978394,
      "learning_rate": 0.0004744105958175828,
      "loss": 2.589568614959717,
      "step": 5140
    },
    {
      "epoch": 0.8294611164891901,
      "grad_norm": 1.105404019355774,
      "learning_rate": 0.00047413050081835805,
      "loss": 2.57688570022583,
      "step": 5141
    },
    {
      "epoch": 0.829622458857696,
      "grad_norm": 1.1535156965255737,
      "learning_rate": 0.0004738489648394373,
      "loss": 2.384568214416504,
      "step": 5142
    },
    {
      "epoch": 0.829783801226202,
      "grad_norm": 1.1998635530471802,
      "learning_rate": 0.00047356598969087126,
      "loss": 2.171971321105957,
      "step": 5143
    },
    {
      "epoch": 0.829945143594708,
      "grad_norm": 1.154705286026001,
      "learning_rate": 0.00047328157719196375,
      "loss": 2.680100679397583,
      "step": 5144
    },
    {
      "epoch": 0.8301064859632139,
      "grad_norm": 1.4067167043685913,
      "learning_rate": 0.00047299572917125943,
      "loss": 2.150846481323242,
      "step": 5145
    },
    {
      "epoch": 0.8302678283317199,
      "grad_norm": 0.9425761103630066,
      "learning_rate": 0.0004727084474665322,
      "loss": 2.544008493423462,
      "step": 5146
    },
    {
      "epoch": 0.8304291707002259,
      "grad_norm": 1.2151744365692139,
      "learning_rate": 0.00047241973392477346,
      "loss": 2.5010550022125244,
      "step": 5147
    },
    {
      "epoch": 0.8305905130687319,
      "grad_norm": 0.9480879306793213,
      "learning_rate": 0.00047212959040218017,
      "loss": 2.505800724029541,
      "step": 5148
    },
    {
      "epoch": 0.8307518554372378,
      "grad_norm": 1.02858567237854,
      "learning_rate": 0.0004718380187641429,
      "loss": 1.683302640914917,
      "step": 5149
    },
    {
      "epoch": 0.8309131978057438,
      "grad_norm": 1.1435925960540771,
      "learning_rate": 0.0004715450208852339,
      "loss": 2.344679355621338,
      "step": 5150
    },
    {
      "epoch": 0.8310745401742498,
      "grad_norm": 1.0560892820358276,
      "learning_rate": 0.00047125059864919493,
      "loss": 2.400266647338867,
      "step": 5151
    },
    {
      "epoch": 0.8312358825427557,
      "grad_norm": 1.189399242401123,
      "learning_rate": 0.0004709547539489253,
      "loss": 2.367532253265381,
      "step": 5152
    },
    {
      "epoch": 0.8313972249112617,
      "grad_norm": 1.6582543849945068,
      "learning_rate": 0.00047065748868646963,
      "loss": 2.550902843475342,
      "step": 5153
    },
    {
      "epoch": 0.8315585672797676,
      "grad_norm": 2.2350993156433105,
      "learning_rate": 0.00047035880477300555,
      "loss": 2.434438943862915,
      "step": 5154
    },
    {
      "epoch": 0.8317199096482737,
      "grad_norm": 1.7724615335464478,
      "learning_rate": 0.00047005870412883145,
      "loss": 2.618867874145508,
      "step": 5155
    },
    {
      "epoch": 0.8318812520167796,
      "grad_norm": 0.8152908682823181,
      "learning_rate": 0.0004697571886833544,
      "loss": 2.224036931991577,
      "step": 5156
    },
    {
      "epoch": 0.8320425943852856,
      "grad_norm": 1.1750779151916504,
      "learning_rate": 0.0004694542603750771,
      "loss": 2.061168670654297,
      "step": 5157
    },
    {
      "epoch": 0.8322039367537916,
      "grad_norm": 2.0319864749908447,
      "learning_rate": 0.00046914992115158635,
      "loss": 2.74300479888916,
      "step": 5158
    },
    {
      "epoch": 0.8323652791222975,
      "grad_norm": 0.9910510182380676,
      "learning_rate": 0.0004688441729695395,
      "loss": 2.59712815284729,
      "step": 5159
    },
    {
      "epoch": 0.8325266214908035,
      "grad_norm": 1.2086787223815918,
      "learning_rate": 0.0004685370177946527,
      "loss": 2.2070322036743164,
      "step": 5160
    },
    {
      "epoch": 0.8326879638593094,
      "grad_norm": 0.9746235013008118,
      "learning_rate": 0.00046822845760168783,
      "loss": 2.306373119354248,
      "step": 5161
    },
    {
      "epoch": 0.8328493062278154,
      "grad_norm": 1.5562478303909302,
      "learning_rate": 0.0004679184943744399,
      "loss": 2.27384614944458,
      "step": 5162
    },
    {
      "epoch": 0.8330106485963213,
      "grad_norm": 0.9972090125083923,
      "learning_rate": 0.0004676071301057243,
      "loss": 2.2720558643341064,
      "step": 5163
    },
    {
      "epoch": 0.8331719909648274,
      "grad_norm": 1.4613219499588013,
      "learning_rate": 0.000467294366797364,
      "loss": 2.4663565158843994,
      "step": 5164
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 1.3858436346054077,
      "learning_rate": 0.0004669802064601769,
      "loss": 2.4971890449523926,
      "step": 5165
    },
    {
      "epoch": 0.8334946757018393,
      "grad_norm": 1.3889429569244385,
      "learning_rate": 0.0004666646511139622,
      "loss": 2.8093738555908203,
      "step": 5166
    },
    {
      "epoch": 0.8336560180703453,
      "grad_norm": 1.306401252746582,
      "learning_rate": 0.00046634770278748835,
      "loss": 2.320141315460205,
      "step": 5167
    },
    {
      "epoch": 0.8338173604388512,
      "grad_norm": 1.6481866836547852,
      "learning_rate": 0.00046602936351847923,
      "loss": 2.4871268272399902,
      "step": 5168
    },
    {
      "epoch": 0.8339787028073572,
      "grad_norm": 1.1662371158599854,
      "learning_rate": 0.00046570963535360157,
      "loss": 2.1828315258026123,
      "step": 5169
    },
    {
      "epoch": 0.8341400451758632,
      "grad_norm": 1.2280431985855103,
      "learning_rate": 0.0004653885203484515,
      "loss": 2.5593161582946777,
      "step": 5170
    },
    {
      "epoch": 0.8343013875443691,
      "grad_norm": 1.065619945526123,
      "learning_rate": 0.00046506602056754147,
      "loss": 2.168026924133301,
      "step": 5171
    },
    {
      "epoch": 0.8344627299128751,
      "grad_norm": 1.649698257446289,
      "learning_rate": 0.0004647421380842869,
      "loss": 2.087904453277588,
      "step": 5172
    },
    {
      "epoch": 0.8346240722813811,
      "grad_norm": 1.046409010887146,
      "learning_rate": 0.0004644168749809928,
      "loss": 2.5595614910125732,
      "step": 5173
    },
    {
      "epoch": 0.8347854146498871,
      "grad_norm": 1.3791979551315308,
      "learning_rate": 0.00046409023334884064,
      "loss": 2.7682628631591797,
      "step": 5174
    },
    {
      "epoch": 0.834946757018393,
      "grad_norm": 1.167371153831482,
      "learning_rate": 0.00046376221528787455,
      "loss": 2.5288991928100586,
      "step": 5175
    },
    {
      "epoch": 0.835108099386899,
      "grad_norm": 1.0076711177825928,
      "learning_rate": 0.0004634328229069881,
      "loss": 2.1299781799316406,
      "step": 5176
    },
    {
      "epoch": 0.835269441755405,
      "grad_norm": 1.7026221752166748,
      "learning_rate": 0.00046310205832391065,
      "loss": 2.6850497722625732,
      "step": 5177
    },
    {
      "epoch": 0.8354307841239109,
      "grad_norm": 1.0765031576156616,
      "learning_rate": 0.0004627699236651937,
      "loss": 2.332723379135132,
      "step": 5178
    },
    {
      "epoch": 0.8355921264924169,
      "grad_norm": 1.120027780532837,
      "learning_rate": 0.000462436421066197,
      "loss": 2.1882526874542236,
      "step": 5179
    },
    {
      "epoch": 0.8357534688609228,
      "grad_norm": 1.2487913370132446,
      "learning_rate": 0.00046210155267107554,
      "loss": 2.708014726638794,
      "step": 5180
    },
    {
      "epoch": 0.8359148112294289,
      "grad_norm": 0.9942679405212402,
      "learning_rate": 0.0004617653206327649,
      "loss": 2.3289413452148438,
      "step": 5181
    },
    {
      "epoch": 0.8360761535979349,
      "grad_norm": 1.0948874950408936,
      "learning_rate": 0.00046142772711296797,
      "loss": 2.2180516719818115,
      "step": 5182
    },
    {
      "epoch": 0.8362374959664408,
      "grad_norm": 1.2540360689163208,
      "learning_rate": 0.0004610887742821408,
      "loss": 2.4162771701812744,
      "step": 5183
    },
    {
      "epoch": 0.8363988383349468,
      "grad_norm": 1.273309588432312,
      "learning_rate": 0.00046074846431947884,
      "loss": 2.2287182807922363,
      "step": 5184
    },
    {
      "epoch": 0.8365601807034527,
      "grad_norm": 0.9263597726821899,
      "learning_rate": 0.00046040679941290253,
      "loss": 1.9338229894638062,
      "step": 5185
    },
    {
      "epoch": 0.8367215230719587,
      "grad_norm": 1.2779085636138916,
      "learning_rate": 0.00046006378175904387,
      "loss": 2.2040255069732666,
      "step": 5186
    },
    {
      "epoch": 0.8368828654404646,
      "grad_norm": 0.9203891158103943,
      "learning_rate": 0.00045971941356323166,
      "loss": 2.1741418838500977,
      "step": 5187
    },
    {
      "epoch": 0.8370442078089706,
      "grad_norm": 1.072078824043274,
      "learning_rate": 0.00045937369703947785,
      "loss": 2.175294876098633,
      "step": 5188
    },
    {
      "epoch": 0.8372055501774766,
      "grad_norm": 1.1466950178146362,
      "learning_rate": 0.00045902663441046277,
      "loss": 2.4804680347442627,
      "step": 5189
    },
    {
      "epoch": 0.8373668925459826,
      "grad_norm": 1.1804866790771484,
      "learning_rate": 0.00045867822790752143,
      "loss": 2.5263800621032715,
      "step": 5190
    },
    {
      "epoch": 0.8375282349144886,
      "grad_norm": 1.2930938005447388,
      "learning_rate": 0.00045832847977062875,
      "loss": 2.4742591381073,
      "step": 5191
    },
    {
      "epoch": 0.8376895772829945,
      "grad_norm": 1.0239123106002808,
      "learning_rate": 0.0004579773922483851,
      "loss": 2.403987407684326,
      "step": 5192
    },
    {
      "epoch": 0.8378509196515005,
      "grad_norm": 0.7956540584564209,
      "learning_rate": 0.00045762496759800244,
      "loss": 2.01952862739563,
      "step": 5193
    },
    {
      "epoch": 0.8380122620200064,
      "grad_norm": 1.0331907272338867,
      "learning_rate": 0.00045727120808528883,
      "loss": 2.2455477714538574,
      "step": 5194
    },
    {
      "epoch": 0.8381736043885124,
      "grad_norm": 1.2703229188919067,
      "learning_rate": 0.000456916115984635,
      "loss": 2.347841262817383,
      "step": 5195
    },
    {
      "epoch": 0.8383349467570184,
      "grad_norm": 0.9422970414161682,
      "learning_rate": 0.00045655969357899875,
      "loss": 2.4911985397338867,
      "step": 5196
    },
    {
      "epoch": 0.8384962891255243,
      "grad_norm": 1.523508906364441,
      "learning_rate": 0.00045620194315989084,
      "loss": 2.2970635890960693,
      "step": 5197
    },
    {
      "epoch": 0.8386576314940304,
      "grad_norm": 1.8015466928482056,
      "learning_rate": 0.0004558428670273601,
      "loss": 2.345972776412964,
      "step": 5198
    },
    {
      "epoch": 0.8388189738625363,
      "grad_norm": 0.9171695113182068,
      "learning_rate": 0.0004554824674899787,
      "loss": 1.988501787185669,
      "step": 5199
    },
    {
      "epoch": 0.8389803162310423,
      "grad_norm": 1.1843875646591187,
      "learning_rate": 0.00045512074686482703,
      "loss": 2.0937039852142334,
      "step": 5200
    },
    {
      "epoch": 0.8391416585995483,
      "grad_norm": 1.0328408479690552,
      "learning_rate": 0.00045475770747747913,
      "loss": 1.960404872894287,
      "step": 5201
    },
    {
      "epoch": 0.8393030009680542,
      "grad_norm": 1.0652954578399658,
      "learning_rate": 0.0004543933516619877,
      "loss": 2.2219653129577637,
      "step": 5202
    },
    {
      "epoch": 0.8394643433365602,
      "grad_norm": 1.0301769971847534,
      "learning_rate": 0.00045402768176086897,
      "loss": 1.9796643257141113,
      "step": 5203
    },
    {
      "epoch": 0.8396256857050661,
      "grad_norm": 0.8798727989196777,
      "learning_rate": 0.0004536607001250875,
      "loss": 2.4738857746124268,
      "step": 5204
    },
    {
      "epoch": 0.8397870280735721,
      "grad_norm": 1.0412054061889648,
      "learning_rate": 0.0004532924091140417,
      "loss": 2.3364105224609375,
      "step": 5205
    },
    {
      "epoch": 0.839948370442078,
      "grad_norm": 0.9184571504592896,
      "learning_rate": 0.0004529228110955478,
      "loss": 2.1877527236938477,
      "step": 5206
    },
    {
      "epoch": 0.8401097128105841,
      "grad_norm": 1.2229902744293213,
      "learning_rate": 0.0004525519084458251,
      "loss": 2.8416006565093994,
      "step": 5207
    },
    {
      "epoch": 0.8402710551790901,
      "grad_norm": 1.3364211320877075,
      "learning_rate": 0.0004521797035494809,
      "loss": 2.440688133239746,
      "step": 5208
    },
    {
      "epoch": 0.840432397547596,
      "grad_norm": 1.1229363679885864,
      "learning_rate": 0.00045180619879949457,
      "loss": 2.573380947113037,
      "step": 5209
    },
    {
      "epoch": 0.840593739916102,
      "grad_norm": 1.246124029159546,
      "learning_rate": 0.0004514313965972027,
      "loss": 2.3281259536743164,
      "step": 5210
    },
    {
      "epoch": 0.8407550822846079,
      "grad_norm": 1.1275758743286133,
      "learning_rate": 0.0004510552993522834,
      "loss": 2.465841770172119,
      "step": 5211
    },
    {
      "epoch": 0.8409164246531139,
      "grad_norm": 3.1939072608947754,
      "learning_rate": 0.0004506779094827409,
      "loss": 2.6319737434387207,
      "step": 5212
    },
    {
      "epoch": 0.8410777670216198,
      "grad_norm": 1.1843822002410889,
      "learning_rate": 0.00045029922941489,
      "loss": 2.5723764896392822,
      "step": 5213
    },
    {
      "epoch": 0.8412391093901258,
      "grad_norm": 2.5785515308380127,
      "learning_rate": 0.0004499192615833403,
      "loss": 2.438117504119873,
      "step": 5214
    },
    {
      "epoch": 0.8414004517586318,
      "grad_norm": 0.9929623603820801,
      "learning_rate": 0.00044953800843098077,
      "loss": 2.27166485786438,
      "step": 5215
    },
    {
      "epoch": 0.8415617941271378,
      "grad_norm": 1.3152884244918823,
      "learning_rate": 0.000449155472408964,
      "loss": 2.149567127227783,
      "step": 5216
    },
    {
      "epoch": 0.8417231364956438,
      "grad_norm": 1.1448163986206055,
      "learning_rate": 0.0004487716559766904,
      "loss": 1.8198356628417969,
      "step": 5217
    },
    {
      "epoch": 0.8418844788641497,
      "grad_norm": 1.213577151298523,
      "learning_rate": 0.00044838656160179236,
      "loss": 1.9725821018218994,
      "step": 5218
    },
    {
      "epoch": 0.8420458212326557,
      "grad_norm": 0.9966157078742981,
      "learning_rate": 0.00044800019176011847,
      "loss": 2.594693660736084,
      "step": 5219
    },
    {
      "epoch": 0.8422071636011617,
      "grad_norm": 0.9730197191238403,
      "learning_rate": 0.0004476125489357175,
      "loss": 1.976178526878357,
      "step": 5220
    },
    {
      "epoch": 0.8423685059696676,
      "grad_norm": 1.1910456418991089,
      "learning_rate": 0.00044722363562082237,
      "loss": 2.478442907333374,
      "step": 5221
    },
    {
      "epoch": 0.8425298483381736,
      "grad_norm": 0.9634907245635986,
      "learning_rate": 0.00044683345431583446,
      "loss": 2.214474678039551,
      "step": 5222
    },
    {
      "epoch": 0.8426911907066795,
      "grad_norm": 1.1345068216323853,
      "learning_rate": 0.00044644200752930717,
      "loss": 2.0066545009613037,
      "step": 5223
    },
    {
      "epoch": 0.8428525330751856,
      "grad_norm": 0.8493626713752747,
      "learning_rate": 0.00044604929777793004,
      "loss": 2.368685722351074,
      "step": 5224
    },
    {
      "epoch": 0.8430138754436916,
      "grad_norm": 0.9452282190322876,
      "learning_rate": 0.0004456553275865123,
      "loss": 2.248295545578003,
      "step": 5225
    },
    {
      "epoch": 0.8431752178121975,
      "grad_norm": 1.7015666961669922,
      "learning_rate": 0.000445260099487967,
      "loss": 2.3754563331604004,
      "step": 5226
    },
    {
      "epoch": 0.8433365601807035,
      "grad_norm": 2.865062713623047,
      "learning_rate": 0.0004448636160232943,
      "loss": 2.005652904510498,
      "step": 5227
    },
    {
      "epoch": 0.8434979025492094,
      "grad_norm": 1.216199517250061,
      "learning_rate": 0.00044446587974156563,
      "loss": 2.4944496154785156,
      "step": 5228
    },
    {
      "epoch": 0.8436592449177154,
      "grad_norm": 1.292547345161438,
      "learning_rate": 0.0004440668931999068,
      "loss": 2.2070345878601074,
      "step": 5229
    },
    {
      "epoch": 0.8438205872862213,
      "grad_norm": 0.9084379076957703,
      "learning_rate": 0.0004436666589634817,
      "loss": 2.642152786254883,
      "step": 5230
    },
    {
      "epoch": 0.8439819296547273,
      "grad_norm": 0.9125267267227173,
      "learning_rate": 0.0004432651796054764,
      "loss": 1.9970961809158325,
      "step": 5231
    },
    {
      "epoch": 0.8441432720232332,
      "grad_norm": 1.0261834859848022,
      "learning_rate": 0.0004428624577070816,
      "loss": 2.3134522438049316,
      "step": 5232
    },
    {
      "epoch": 0.8443046143917393,
      "grad_norm": 1.0283441543579102,
      "learning_rate": 0.00044245849585747656,
      "loss": 2.1414990425109863,
      "step": 5233
    },
    {
      "epoch": 0.8444659567602453,
      "grad_norm": 1.1574618816375732,
      "learning_rate": 0.0004420532966538129,
      "loss": 2.3050220012664795,
      "step": 5234
    },
    {
      "epoch": 0.8446272991287512,
      "grad_norm": 1.115075707435608,
      "learning_rate": 0.00044164686270119685,
      "loss": 2.1845083236694336,
      "step": 5235
    },
    {
      "epoch": 0.8447886414972572,
      "grad_norm": 1.1309186220169067,
      "learning_rate": 0.0004412391966126735,
      "loss": 2.4767403602600098,
      "step": 5236
    },
    {
      "epoch": 0.8449499838657631,
      "grad_norm": 0.8907322287559509,
      "learning_rate": 0.0004408303010092095,
      "loss": 2.4172942638397217,
      "step": 5237
    },
    {
      "epoch": 0.8451113262342691,
      "grad_norm": 1.2281781435012817,
      "learning_rate": 0.0004404201785196762,
      "loss": 2.295319080352783,
      "step": 5238
    },
    {
      "epoch": 0.8452726686027751,
      "grad_norm": 1.093697190284729,
      "learning_rate": 0.00044000883178083283,
      "loss": 2.1987476348876953,
      "step": 5239
    },
    {
      "epoch": 0.845434010971281,
      "grad_norm": 1.6352763175964355,
      "learning_rate": 0.0004395962634373096,
      "loss": 2.3733632564544678,
      "step": 5240
    },
    {
      "epoch": 0.8455953533397871,
      "grad_norm": 1.4859668016433716,
      "learning_rate": 0.00043918247614159077,
      "loss": 2.03570294380188,
      "step": 5241
    },
    {
      "epoch": 0.845756695708293,
      "grad_norm": 0.9094337821006775,
      "learning_rate": 0.0004387674725539974,
      "loss": 2.4987144470214844,
      "step": 5242
    },
    {
      "epoch": 0.845918038076799,
      "grad_norm": 0.9232254028320312,
      "learning_rate": 0.00043835125534267037,
      "loss": 2.286665439605713,
      "step": 5243
    },
    {
      "epoch": 0.846079380445305,
      "grad_norm": 1.0036927461624146,
      "learning_rate": 0.00043793382718355293,
      "loss": 2.5896897315979004,
      "step": 5244
    },
    {
      "epoch": 0.8462407228138109,
      "grad_norm": 0.9928473234176636,
      "learning_rate": 0.00043751519076037427,
      "loss": 2.287721633911133,
      "step": 5245
    },
    {
      "epoch": 0.8464020651823169,
      "grad_norm": 1.1348650455474854,
      "learning_rate": 0.0004370953487646313,
      "loss": 2.3592140674591064,
      "step": 5246
    },
    {
      "epoch": 0.8465634075508228,
      "grad_norm": 0.995826005935669,
      "learning_rate": 0.00043667430389557197,
      "loss": 2.5048861503601074,
      "step": 5247
    },
    {
      "epoch": 0.8467247499193288,
      "grad_norm": 1.0643786191940308,
      "learning_rate": 0.00043625205886017773,
      "loss": 2.7157740592956543,
      "step": 5248
    },
    {
      "epoch": 0.8468860922878347,
      "grad_norm": 1.2655984163284302,
      "learning_rate": 0.00043582861637314613,
      "loss": 2.246293306350708,
      "step": 5249
    },
    {
      "epoch": 0.8470474346563408,
      "grad_norm": 1.144315242767334,
      "learning_rate": 0.0004354039791568734,
      "loss": 1.9845905303955078,
      "step": 5250
    },
    {
      "epoch": 0.8472087770248468,
      "grad_norm": 1.9438573122024536,
      "learning_rate": 0.0004349781499414369,
      "loss": 3.0342578887939453,
      "step": 5251
    },
    {
      "epoch": 0.8473701193933527,
      "grad_norm": 1.2941166162490845,
      "learning_rate": 0.00043455113146457765,
      "loss": 2.4322736263275146,
      "step": 5252
    },
    {
      "epoch": 0.8475314617618587,
      "grad_norm": 1.4356337785720825,
      "learning_rate": 0.0004341229264716825,
      "loss": 2.8968000411987305,
      "step": 5253
    },
    {
      "epoch": 0.8476928041303646,
      "grad_norm": 0.9394021034240723,
      "learning_rate": 0.0004336935377157668,
      "loss": 2.35625958442688,
      "step": 5254
    },
    {
      "epoch": 0.8478541464988706,
      "grad_norm": 1.083450198173523,
      "learning_rate": 0.00043326296795745655,
      "loss": 2.5158863067626953,
      "step": 5255
    },
    {
      "epoch": 0.8480154888673765,
      "grad_norm": 1.1762282848358154,
      "learning_rate": 0.00043283121996497067,
      "loss": 2.1617650985717773,
      "step": 5256
    },
    {
      "epoch": 0.8481768312358825,
      "grad_norm": 1.0420430898666382,
      "learning_rate": 0.00043239829651410295,
      "loss": 2.3609323501586914,
      "step": 5257
    },
    {
      "epoch": 0.8483381736043886,
      "grad_norm": 1.1458582878112793,
      "learning_rate": 0.00043196420038820474,
      "loss": 2.2683935165405273,
      "step": 5258
    },
    {
      "epoch": 0.8484995159728945,
      "grad_norm": 1.2088394165039062,
      "learning_rate": 0.00043152893437816656,
      "loss": 2.843104839324951,
      "step": 5259
    },
    {
      "epoch": 0.8486608583414005,
      "grad_norm": 1.063546895980835,
      "learning_rate": 0.0004310925012824004,
      "loss": 2.406360626220703,
      "step": 5260
    },
    {
      "epoch": 0.8488222007099064,
      "grad_norm": 1.1172617673873901,
      "learning_rate": 0.00043065490390682183,
      "loss": 2.4699149131774902,
      "step": 5261
    },
    {
      "epoch": 0.8489835430784124,
      "grad_norm": 1.1116464138031006,
      "learning_rate": 0.0004302161450648315,
      "loss": 2.17457914352417,
      "step": 5262
    },
    {
      "epoch": 0.8491448854469184,
      "grad_norm": 1.1614876985549927,
      "learning_rate": 0.0004297762275772975,
      "loss": 2.3673815727233887,
      "step": 5263
    },
    {
      "epoch": 0.8493062278154243,
      "grad_norm": 1.4165802001953125,
      "learning_rate": 0.0004293351542725372,
      "loss": 2.7323853969573975,
      "step": 5264
    },
    {
      "epoch": 0.8494675701839303,
      "grad_norm": 1.0016204118728638,
      "learning_rate": 0.0004288929279862989,
      "loss": 2.0308420658111572,
      "step": 5265
    },
    {
      "epoch": 0.8496289125524362,
      "grad_norm": 1.061154842376709,
      "learning_rate": 0.00042844955156174345,
      "loss": 2.2582077980041504,
      "step": 5266
    },
    {
      "epoch": 0.8497902549209423,
      "grad_norm": 1.3986892700195312,
      "learning_rate": 0.0004280050278494265,
      "loss": 2.618410587310791,
      "step": 5267
    },
    {
      "epoch": 0.8499515972894482,
      "grad_norm": 1.6353174448013306,
      "learning_rate": 0.0004275593597072795,
      "loss": 2.698467254638672,
      "step": 5268
    },
    {
      "epoch": 0.8501129396579542,
      "grad_norm": 1.1574816703796387,
      "learning_rate": 0.000427112550000592,
      "loss": 2.17616605758667,
      "step": 5269
    },
    {
      "epoch": 0.8502742820264602,
      "grad_norm": 1.1658220291137695,
      "learning_rate": 0.0004266646016019926,
      "loss": 2.2899129390716553,
      "step": 5270
    },
    {
      "epoch": 0.8504356243949661,
      "grad_norm": 1.6123028993606567,
      "learning_rate": 0.00042621551739143094,
      "loss": 2.8310086727142334,
      "step": 5271
    },
    {
      "epoch": 0.8505969667634721,
      "grad_norm": 1.4425289630889893,
      "learning_rate": 0.00042576530025615895,
      "loss": 2.415170192718506,
      "step": 5272
    },
    {
      "epoch": 0.850758309131978,
      "grad_norm": 1.1501363515853882,
      "learning_rate": 0.0004253139530907124,
      "loss": 2.4433531761169434,
      "step": 5273
    },
    {
      "epoch": 0.850919651500484,
      "grad_norm": 1.040900707244873,
      "learning_rate": 0.0004248614787968924,
      "loss": 1.9826043844223022,
      "step": 5274
    },
    {
      "epoch": 0.8510809938689899,
      "grad_norm": 1.0285804271697998,
      "learning_rate": 0.0004244078802837462,
      "loss": 2.23799729347229,
      "step": 5275
    },
    {
      "epoch": 0.851242336237496,
      "grad_norm": 1.0574653148651123,
      "learning_rate": 0.00042395316046754927,
      "loss": 2.5759201049804688,
      "step": 5276
    },
    {
      "epoch": 0.851403678606002,
      "grad_norm": 0.9507919549942017,
      "learning_rate": 0.0004234973222717859,
      "loss": 2.0793962478637695,
      "step": 5277
    },
    {
      "epoch": 0.8515650209745079,
      "grad_norm": 1.2798717021942139,
      "learning_rate": 0.0004230403686271309,
      "loss": 2.1708579063415527,
      "step": 5278
    },
    {
      "epoch": 0.8517263633430139,
      "grad_norm": 0.9582514762878418,
      "learning_rate": 0.00042258230247143035,
      "loss": 2.61698055267334,
      "step": 5279
    },
    {
      "epoch": 0.8518877057115198,
      "grad_norm": 1.9851090908050537,
      "learning_rate": 0.000422123126749683,
      "loss": 1.9355547428131104,
      "step": 5280
    },
    {
      "epoch": 0.8520490480800258,
      "grad_norm": 1.8089245557785034,
      "learning_rate": 0.000421662844414021,
      "loss": 2.5995967388153076,
      "step": 5281
    },
    {
      "epoch": 0.8522103904485318,
      "grad_norm": 2.0715363025665283,
      "learning_rate": 0.0004212014584236914,
      "loss": 2.3969926834106445,
      "step": 5282
    },
    {
      "epoch": 0.8523717328170377,
      "grad_norm": 1.4834296703338623,
      "learning_rate": 0.00042073897174503674,
      "loss": 2.708540916442871,
      "step": 5283
    },
    {
      "epoch": 0.8525330751855438,
      "grad_norm": 0.9792858362197876,
      "learning_rate": 0.00042027538735147615,
      "loss": 2.3314733505249023,
      "step": 5284
    },
    {
      "epoch": 0.8526944175540497,
      "grad_norm": 1.0086363554000854,
      "learning_rate": 0.0004198107082234861,
      "loss": 2.686985492706299,
      "step": 5285
    },
    {
      "epoch": 0.8528557599225557,
      "grad_norm": 0.9877910017967224,
      "learning_rate": 0.0004193449373485815,
      "loss": 2.104372501373291,
      "step": 5286
    },
    {
      "epoch": 0.8530171022910616,
      "grad_norm": 1.195796012878418,
      "learning_rate": 0.0004188780777212961,
      "loss": 2.64156174659729,
      "step": 5287
    },
    {
      "epoch": 0.8531784446595676,
      "grad_norm": 1.1903440952301025,
      "learning_rate": 0.0004184101323431636,
      "loss": 2.4556732177734375,
      "step": 5288
    },
    {
      "epoch": 0.8533397870280736,
      "grad_norm": 0.9650347828865051,
      "learning_rate": 0.0004179411042226982,
      "loss": 2.3178529739379883,
      "step": 5289
    },
    {
      "epoch": 0.8535011293965795,
      "grad_norm": 1.0310077667236328,
      "learning_rate": 0.00041747099637537524,
      "loss": 2.133742570877075,
      "step": 5290
    },
    {
      "epoch": 0.8536624717650855,
      "grad_norm": 0.9793081879615784,
      "learning_rate": 0.0004169998118236118,
      "loss": 2.2055861949920654,
      "step": 5291
    },
    {
      "epoch": 0.8538238141335914,
      "grad_norm": 2.784398317337036,
      "learning_rate": 0.0004165275535967474,
      "loss": 2.3668832778930664,
      "step": 5292
    },
    {
      "epoch": 0.8539851565020975,
      "grad_norm": 1.6301900148391724,
      "learning_rate": 0.0004160542247310244,
      "loss": 2.0584065914154053,
      "step": 5293
    },
    {
      "epoch": 0.8541464988706035,
      "grad_norm": 0.9331153631210327,
      "learning_rate": 0.0004155798282695684,
      "loss": 2.2995779514312744,
      "step": 5294
    },
    {
      "epoch": 0.8543078412391094,
      "grad_norm": 0.9524773359298706,
      "learning_rate": 0.00041510436726236897,
      "loss": 2.305027961730957,
      "step": 5295
    },
    {
      "epoch": 0.8544691836076154,
      "grad_norm": 1.0908808708190918,
      "learning_rate": 0.0004146278447662597,
      "loss": 2.454967737197876,
      "step": 5296
    },
    {
      "epoch": 0.8546305259761213,
      "grad_norm": 0.9242610335350037,
      "learning_rate": 0.00041415026384489874,
      "loss": 2.4803428649902344,
      "step": 5297
    },
    {
      "epoch": 0.8547918683446273,
      "grad_norm": 0.8644957542419434,
      "learning_rate": 0.0004136716275687492,
      "loss": 1.781579613685608,
      "step": 5298
    },
    {
      "epoch": 0.8549532107131332,
      "grad_norm": 0.8835524320602417,
      "learning_rate": 0.0004131919390150591,
      "loss": 2.1268348693847656,
      "step": 5299
    },
    {
      "epoch": 0.8551145530816392,
      "grad_norm": 1.151764154434204,
      "learning_rate": 0.0004127112012678418,
      "loss": 2.5245859622955322,
      "step": 5300
    },
    {
      "epoch": 0.8552758954501453,
      "grad_norm": 0.8095287680625916,
      "learning_rate": 0.0004122294174178563,
      "loss": 2.1851320266723633,
      "step": 5301
    },
    {
      "epoch": 0.8554372378186512,
      "grad_norm": 1.305848240852356,
      "learning_rate": 0.0004117465905625869,
      "loss": 2.762420177459717,
      "step": 5302
    },
    {
      "epoch": 0.8555985801871572,
      "grad_norm": 1.3632022142410278,
      "learning_rate": 0.00041126272380622385,
      "loss": 2.277350902557373,
      "step": 5303
    },
    {
      "epoch": 0.8557599225556631,
      "grad_norm": 1.447457194328308,
      "learning_rate": 0.00041077782025964317,
      "loss": 2.223546028137207,
      "step": 5304
    },
    {
      "epoch": 0.8559212649241691,
      "grad_norm": 1.369831919670105,
      "learning_rate": 0.0004102918830403862,
      "loss": 2.7456717491149902,
      "step": 5305
    },
    {
      "epoch": 0.856082607292675,
      "grad_norm": 1.2317644357681274,
      "learning_rate": 0.00040980491527264044,
      "loss": 2.827317237854004,
      "step": 5306
    },
    {
      "epoch": 0.856243949661181,
      "grad_norm": 0.9732476472854614,
      "learning_rate": 0.00040931692008721877,
      "loss": 2.0193958282470703,
      "step": 5307
    },
    {
      "epoch": 0.856405292029687,
      "grad_norm": 0.9825813174247742,
      "learning_rate": 0.0004088279006215396,
      "loss": 2.385957717895508,
      "step": 5308
    },
    {
      "epoch": 0.8565666343981929,
      "grad_norm": 1.0766546726226807,
      "learning_rate": 0.00040833786001960646,
      "loss": 2.5534253120422363,
      "step": 5309
    },
    {
      "epoch": 0.856727976766699,
      "grad_norm": 1.2786555290222168,
      "learning_rate": 0.0004078468014319884,
      "loss": 2.3300015926361084,
      "step": 5310
    },
    {
      "epoch": 0.8568893191352049,
      "grad_norm": 1.0196902751922607,
      "learning_rate": 0.00040735472801579887,
      "loss": 2.6409897804260254,
      "step": 5311
    },
    {
      "epoch": 0.8570506615037109,
      "grad_norm": 0.9804675579071045,
      "learning_rate": 0.00040686164293467594,
      "loss": 2.112515449523926,
      "step": 5312
    },
    {
      "epoch": 0.8572120038722169,
      "grad_norm": 0.9312787055969238,
      "learning_rate": 0.0004063675493587621,
      "loss": 2.3853185176849365,
      "step": 5313
    },
    {
      "epoch": 0.8573733462407228,
      "grad_norm": 1.4907327890396118,
      "learning_rate": 0.0004058724504646834,
      "loss": 2.6759610176086426,
      "step": 5314
    },
    {
      "epoch": 0.8575346886092288,
      "grad_norm": 1.6978976726531982,
      "learning_rate": 0.0004053763494355295,
      "loss": 2.203360080718994,
      "step": 5315
    },
    {
      "epoch": 0.8576960309777347,
      "grad_norm": 1.033203125,
      "learning_rate": 0.00040487924946083266,
      "loss": 2.2210254669189453,
      "step": 5316
    },
    {
      "epoch": 0.8578573733462407,
      "grad_norm": 0.9735785722732544,
      "learning_rate": 0.000404381153736548,
      "loss": 2.0561156272888184,
      "step": 5317
    },
    {
      "epoch": 0.8580187157147466,
      "grad_norm": 0.9839133024215698,
      "learning_rate": 0.00040388206546503215,
      "loss": 2.6938936710357666,
      "step": 5318
    },
    {
      "epoch": 0.8581800580832527,
      "grad_norm": 2.091325521469116,
      "learning_rate": 0.0004033819878550232,
      "loss": 2.430863857269287,
      "step": 5319
    },
    {
      "epoch": 0.8583414004517587,
      "grad_norm": 1.2266234159469604,
      "learning_rate": 0.00040288092412161993,
      "loss": 2.2996511459350586,
      "step": 5320
    },
    {
      "epoch": 0.8585027428202646,
      "grad_norm": 1.1569278240203857,
      "learning_rate": 0.0004023788774862609,
      "loss": 2.2867374420166016,
      "step": 5321
    },
    {
      "epoch": 0.8586640851887706,
      "grad_norm": 1.2195546627044678,
      "learning_rate": 0.0004018758511767041,
      "loss": 2.1889944076538086,
      "step": 5322
    },
    {
      "epoch": 0.8588254275572765,
      "grad_norm": 1.8876703977584839,
      "learning_rate": 0.00040137184842700614,
      "loss": 2.1926674842834473,
      "step": 5323
    },
    {
      "epoch": 0.8589867699257825,
      "grad_norm": 1.1708883047103882,
      "learning_rate": 0.000400866872477501,
      "loss": 2.378904342651367,
      "step": 5324
    },
    {
      "epoch": 0.8591481122942884,
      "grad_norm": 1.0236085653305054,
      "learning_rate": 0.00040036092657477995,
      "loss": 2.4301161766052246,
      "step": 5325
    },
    {
      "epoch": 0.8593094546627944,
      "grad_norm": 1.0863163471221924,
      "learning_rate": 0.0003998540139716701,
      "loss": 2.0382919311523438,
      "step": 5326
    },
    {
      "epoch": 0.8594707970313005,
      "grad_norm": 0.9510501623153687,
      "learning_rate": 0.00039934613792721356,
      "loss": 2.4916677474975586,
      "step": 5327
    },
    {
      "epoch": 0.8596321393998064,
      "grad_norm": 1.2003557682037354,
      "learning_rate": 0.0003988373017066469,
      "loss": 2.482771396636963,
      "step": 5328
    },
    {
      "epoch": 0.8597934817683124,
      "grad_norm": 0.9131209850311279,
      "learning_rate": 0.0003983275085813796,
      "loss": 2.467278480529785,
      "step": 5329
    },
    {
      "epoch": 0.8599548241368183,
      "grad_norm": 1.1649816036224365,
      "learning_rate": 0.0003978167618289734,
      "loss": 2.21921706199646,
      "step": 5330
    },
    {
      "epoch": 0.8601161665053243,
      "grad_norm": 1.005751371383667,
      "learning_rate": 0.0003973050647331209,
      "loss": 2.599980354309082,
      "step": 5331
    },
    {
      "epoch": 0.8602775088738303,
      "grad_norm": 1.2479554414749146,
      "learning_rate": 0.000396792420583625,
      "loss": 2.8136630058288574,
      "step": 5332
    },
    {
      "epoch": 0.8604388512423362,
      "grad_norm": 1.0438919067382812,
      "learning_rate": 0.000396278832676377,
      "loss": 2.258626937866211,
      "step": 5333
    },
    {
      "epoch": 0.8606001936108422,
      "grad_norm": 1.2620937824249268,
      "learning_rate": 0.0003957643043133361,
      "loss": 2.737128973007202,
      "step": 5334
    },
    {
      "epoch": 0.8607615359793481,
      "grad_norm": 0.9928492307662964,
      "learning_rate": 0.00039524883880250777,
      "loss": 1.9368873834609985,
      "step": 5335
    },
    {
      "epoch": 0.8609228783478542,
      "grad_norm": 1.233316421508789,
      "learning_rate": 0.0003947324394579226,
      "loss": 2.9229888916015625,
      "step": 5336
    },
    {
      "epoch": 0.8610842207163601,
      "grad_norm": 0.824320912361145,
      "learning_rate": 0.000394215109599615,
      "loss": 2.2716100215911865,
      "step": 5337
    },
    {
      "epoch": 0.8612455630848661,
      "grad_norm": 1.348173975944519,
      "learning_rate": 0.0003936968525536018,
      "loss": 2.732670783996582,
      "step": 5338
    },
    {
      "epoch": 0.8614069054533721,
      "grad_norm": 1.1269475221633911,
      "learning_rate": 0.0003931776716518608,
      "loss": 2.353881359100342,
      "step": 5339
    },
    {
      "epoch": 0.861568247821878,
      "grad_norm": 1.3286341428756714,
      "learning_rate": 0.00039265757023230975,
      "loss": 1.962100863456726,
      "step": 5340
    },
    {
      "epoch": 0.861729590190384,
      "grad_norm": 1.259482502937317,
      "learning_rate": 0.00039213655163878436,
      "loss": 2.1966915130615234,
      "step": 5341
    },
    {
      "epoch": 0.8618909325588899,
      "grad_norm": 0.9978271126747131,
      "learning_rate": 0.0003916146192210171,
      "loss": 2.3640336990356445,
      "step": 5342
    },
    {
      "epoch": 0.8620522749273959,
      "grad_norm": 0.9768631458282471,
      "learning_rate": 0.00039109177633461556,
      "loss": 2.008115291595459,
      "step": 5343
    },
    {
      "epoch": 0.862213617295902,
      "grad_norm": 0.9610313177108765,
      "learning_rate": 0.000390568026341041,
      "loss": 2.3561408519744873,
      "step": 5344
    },
    {
      "epoch": 0.8623749596644079,
      "grad_norm": 0.8536903262138367,
      "learning_rate": 0.0003900433726075865,
      "loss": 2.336329936981201,
      "step": 5345
    },
    {
      "epoch": 0.8625363020329139,
      "grad_norm": 0.912268877029419,
      "learning_rate": 0.00038951781850735573,
      "loss": 2.2147138118743896,
      "step": 5346
    },
    {
      "epoch": 0.8626976444014198,
      "grad_norm": 0.9644731879234314,
      "learning_rate": 0.0003889913674192407,
      "loss": 2.395688772201538,
      "step": 5347
    },
    {
      "epoch": 0.8628589867699258,
      "grad_norm": 0.9929842948913574,
      "learning_rate": 0.00038846402272790044,
      "loss": 2.298990249633789,
      "step": 5348
    },
    {
      "epoch": 0.8630203291384317,
      "grad_norm": 0.9870695471763611,
      "learning_rate": 0.00038793578782373925,
      "loss": 2.51552152633667,
      "step": 5349
    },
    {
      "epoch": 0.8631816715069377,
      "grad_norm": 1.1223781108856201,
      "learning_rate": 0.0003874066661028846,
      "loss": 2.425151824951172,
      "step": 5350
    },
    {
      "epoch": 0.8633430138754437,
      "grad_norm": 0.9422720670700073,
      "learning_rate": 0.0003868766609671657,
      "loss": 2.5044713020324707,
      "step": 5351
    },
    {
      "epoch": 0.8635043562439496,
      "grad_norm": 1.3302990198135376,
      "learning_rate": 0.0003863457758240912,
      "loss": 2.4805023670196533,
      "step": 5352
    },
    {
      "epoch": 0.8636656986124557,
      "grad_norm": 0.93210768699646,
      "learning_rate": 0.0003858140140868276,
      "loss": 2.1563236713409424,
      "step": 5353
    },
    {
      "epoch": 0.8638270409809616,
      "grad_norm": 1.2175259590148926,
      "learning_rate": 0.00038528137917417715,
      "loss": 2.303069591522217,
      "step": 5354
    },
    {
      "epoch": 0.8639883833494676,
      "grad_norm": 0.9085325002670288,
      "learning_rate": 0.000384747874510556,
      "loss": 2.116408348083496,
      "step": 5355
    },
    {
      "epoch": 0.8641497257179735,
      "grad_norm": 0.7440609335899353,
      "learning_rate": 0.00038421350352597195,
      "loss": 2.3581511974334717,
      "step": 5356
    },
    {
      "epoch": 0.8643110680864795,
      "grad_norm": 9.685898780822754,
      "learning_rate": 0.00038367826965600275,
      "loss": 2.5720911026000977,
      "step": 5357
    },
    {
      "epoch": 0.8644724104549855,
      "grad_norm": 1.0623538494110107,
      "learning_rate": 0.00038314217634177376,
      "loss": 2.099398136138916,
      "step": 5358
    },
    {
      "epoch": 0.8646337528234914,
      "grad_norm": 0.9799504280090332,
      "learning_rate": 0.0003826052270299356,
      "loss": 2.4682579040527344,
      "step": 5359
    },
    {
      "epoch": 0.8647950951919974,
      "grad_norm": 1.3057194948196411,
      "learning_rate": 0.00038206742517264274,
      "loss": 2.2375986576080322,
      "step": 5360
    },
    {
      "epoch": 0.8649564375605033,
      "grad_norm": 0.8359051942825317,
      "learning_rate": 0.0003815287742275305,
      "loss": 2.539494276046753,
      "step": 5361
    },
    {
      "epoch": 0.8651177799290094,
      "grad_norm": 1.0578868389129639,
      "learning_rate": 0.0003809892776576932,
      "loss": 2.344245672225952,
      "step": 5362
    },
    {
      "epoch": 0.8652791222975154,
      "grad_norm": 1.0373343229293823,
      "learning_rate": 0.000380448938931662,
      "loss": 2.552609920501709,
      "step": 5363
    },
    {
      "epoch": 0.8654404646660213,
      "grad_norm": 1.4880684614181519,
      "learning_rate": 0.00037990776152338224,
      "loss": 2.4093518257141113,
      "step": 5364
    },
    {
      "epoch": 0.8656018070345273,
      "grad_norm": 0.8788935542106628,
      "learning_rate": 0.00037936574891219143,
      "loss": 2.266730308532715,
      "step": 5365
    },
    {
      "epoch": 0.8657631494030332,
      "grad_norm": 1.0392040014266968,
      "learning_rate": 0.000378822904582797,
      "loss": 1.9939067363739014,
      "step": 5366
    },
    {
      "epoch": 0.8659244917715392,
      "grad_norm": 0.8539532423019409,
      "learning_rate": 0.00037827923202525304,
      "loss": 2.4864509105682373,
      "step": 5367
    },
    {
      "epoch": 0.8660858341400451,
      "grad_norm": 1.0990281105041504,
      "learning_rate": 0.00037773473473493913,
      "loss": 2.0996482372283936,
      "step": 5368
    },
    {
      "epoch": 0.8662471765085511,
      "grad_norm": 1.2721045017242432,
      "learning_rate": 0.00037718941621253687,
      "loss": 2.5603203773498535,
      "step": 5369
    },
    {
      "epoch": 0.8664085188770572,
      "grad_norm": 0.8080335259437561,
      "learning_rate": 0.00037664327996400783,
      "loss": 2.1083872318267822,
      "step": 5370
    },
    {
      "epoch": 0.8665698612455631,
      "grad_norm": 1.0364974737167358,
      "learning_rate": 0.00037609632950057095,
      "loss": 2.3783822059631348,
      "step": 5371
    },
    {
      "epoch": 0.8667312036140691,
      "grad_norm": 0.7705390453338623,
      "learning_rate": 0.00037554856833867974,
      "loss": 1.8831851482391357,
      "step": 5372
    },
    {
      "epoch": 0.866892545982575,
      "grad_norm": 1.5841737985610962,
      "learning_rate": 0.000375,
      "loss": 2.217048168182373,
      "step": 5373
    },
    {
      "epoch": 0.867053888351081,
      "grad_norm": 0.9648942351341248,
      "learning_rate": 0.00037445062801138706,
      "loss": 2.510824203491211,
      "step": 5374
    },
    {
      "epoch": 0.867215230719587,
      "grad_norm": 0.7720943093299866,
      "learning_rate": 0.000373900455904863,
      "loss": 2.1968729496002197,
      "step": 5375
    },
    {
      "epoch": 0.8673765730880929,
      "grad_norm": 1.1203594207763672,
      "learning_rate": 0.000373349487217594,
      "loss": 2.2310714721679688,
      "step": 5376
    },
    {
      "epoch": 0.8675379154565989,
      "grad_norm": 0.8874133229255676,
      "learning_rate": 0.00037279772549186755,
      "loss": 2.370368480682373,
      "step": 5377
    },
    {
      "epoch": 0.8676992578251048,
      "grad_norm": 1.0497950315475464,
      "learning_rate": 0.0003722451742750701,
      "loss": 2.4503045082092285,
      "step": 5378
    },
    {
      "epoch": 0.8678606001936109,
      "grad_norm": 1.1444284915924072,
      "learning_rate": 0.00037169183711966347,
      "loss": 2.2146100997924805,
      "step": 5379
    },
    {
      "epoch": 0.8680219425621168,
      "grad_norm": 1.0839049816131592,
      "learning_rate": 0.00037113771758316255,
      "loss": 2.037330389022827,
      "step": 5380
    },
    {
      "epoch": 0.8681832849306228,
      "grad_norm": 0.9370010495185852,
      "learning_rate": 0.00037058281922811254,
      "loss": 2.4425196647644043,
      "step": 5381
    },
    {
      "epoch": 0.8683446272991288,
      "grad_norm": 1.0633999109268188,
      "learning_rate": 0.0003700271456220654,
      "loss": 2.4235239028930664,
      "step": 5382
    },
    {
      "epoch": 0.8685059696676347,
      "grad_norm": 0.9254845976829529,
      "learning_rate": 0.0003694707003375579,
      "loss": 2.0742039680480957,
      "step": 5383
    },
    {
      "epoch": 0.8686673120361407,
      "grad_norm": 1.1596310138702393,
      "learning_rate": 0.00036891348695208755,
      "loss": 2.6381378173828125,
      "step": 5384
    },
    {
      "epoch": 0.8688286544046466,
      "grad_norm": 0.8912410140037537,
      "learning_rate": 0.00036835550904809036,
      "loss": 2.1977906227111816,
      "step": 5385
    },
    {
      "epoch": 0.8689899967731526,
      "grad_norm": 0.9813294410705566,
      "learning_rate": 0.0003677967702129177,
      "loss": 2.4800662994384766,
      "step": 5386
    },
    {
      "epoch": 0.8691513391416587,
      "grad_norm": 1.2324402332305908,
      "learning_rate": 0.0003672372740388128,
      "loss": 2.7386045455932617,
      "step": 5387
    },
    {
      "epoch": 0.8693126815101646,
      "grad_norm": 9.116289138793945,
      "learning_rate": 0.0003666770241228883,
      "loss": 2.096754789352417,
      "step": 5388
    },
    {
      "epoch": 0.8694740238786706,
      "grad_norm": 2.303964138031006,
      "learning_rate": 0.00036611602406710263,
      "loss": 2.2152328491210938,
      "step": 5389
    },
    {
      "epoch": 0.8696353662471765,
      "grad_norm": 1.0503143072128296,
      "learning_rate": 0.00036555427747823703,
      "loss": 2.1274209022521973,
      "step": 5390
    },
    {
      "epoch": 0.8697967086156825,
      "grad_norm": 1.1574742794036865,
      "learning_rate": 0.00036499178796787243,
      "loss": 2.565647602081299,
      "step": 5391
    },
    {
      "epoch": 0.8699580509841884,
      "grad_norm": 1.3982572555541992,
      "learning_rate": 0.00036442855915236604,
      "loss": 2.4033455848693848,
      "step": 5392
    },
    {
      "epoch": 0.8701193933526944,
      "grad_norm": 1.0735805034637451,
      "learning_rate": 0.0003638645946528282,
      "loss": 2.2982029914855957,
      "step": 5393
    },
    {
      "epoch": 0.8702807357212003,
      "grad_norm": 1.3609952926635742,
      "learning_rate": 0.00036329989809509935,
      "loss": 2.57481050491333,
      "step": 5394
    },
    {
      "epoch": 0.8704420780897063,
      "grad_norm": 0.9021140336990356,
      "learning_rate": 0.00036273447310972607,
      "loss": 2.0064308643341064,
      "step": 5395
    },
    {
      "epoch": 0.8706034204582124,
      "grad_norm": 0.8844737410545349,
      "learning_rate": 0.00036216832333193854,
      "loss": 2.284518241882324,
      "step": 5396
    },
    {
      "epoch": 0.8707647628267183,
      "grad_norm": 1.134842038154602,
      "learning_rate": 0.00036160145240162655,
      "loss": 2.728844165802002,
      "step": 5397
    },
    {
      "epoch": 0.8709261051952243,
      "grad_norm": 1.10336434841156,
      "learning_rate": 0.00036103386396331633,
      "loss": 2.078549861907959,
      "step": 5398
    },
    {
      "epoch": 0.8710874475637302,
      "grad_norm": 0.9118515849113464,
      "learning_rate": 0.0003604655616661474,
      "loss": 2.3777151107788086,
      "step": 5399
    },
    {
      "epoch": 0.8712487899322362,
      "grad_norm": 1.3479657173156738,
      "learning_rate": 0.0003598965491638482,
      "loss": 2.597032308578491,
      "step": 5400
    },
    {
      "epoch": 0.8714101323007422,
      "grad_norm": 1.0240119695663452,
      "learning_rate": 0.0003593268301147139,
      "loss": 2.736823558807373,
      "step": 5401
    },
    {
      "epoch": 0.8715714746692481,
      "grad_norm": 0.9633117318153381,
      "learning_rate": 0.0003587564081815818,
      "loss": 2.3905653953552246,
      "step": 5402
    },
    {
      "epoch": 0.8717328170377541,
      "grad_norm": 0.8074313402175903,
      "learning_rate": 0.0003581852870318083,
      "loss": 2.263998508453369,
      "step": 5403
    },
    {
      "epoch": 0.8718941594062601,
      "grad_norm": 1.2047594785690308,
      "learning_rate": 0.0003576134703372452,
      "loss": 2.7280662059783936,
      "step": 5404
    },
    {
      "epoch": 0.8720555017747661,
      "grad_norm": 1.0416475534439087,
      "learning_rate": 0.0003570409617742161,
      "loss": 2.412477970123291,
      "step": 5405
    },
    {
      "epoch": 0.872216844143272,
      "grad_norm": 2.077237844467163,
      "learning_rate": 0.0003564677650234927,
      "loss": 2.127513885498047,
      "step": 5406
    },
    {
      "epoch": 0.872378186511778,
      "grad_norm": 1.1192626953125,
      "learning_rate": 0.0003558938837702713,
      "loss": 2.403250217437744,
      "step": 5407
    },
    {
      "epoch": 0.872539528880284,
      "grad_norm": 0.9526357054710388,
      "learning_rate": 0.00035531932170414893,
      "loss": 2.765874147415161,
      "step": 5408
    },
    {
      "epoch": 0.8727008712487899,
      "grad_norm": 2.0018200874328613,
      "learning_rate": 0.0003547440825190997,
      "loss": 2.6736810207366943,
      "step": 5409
    },
    {
      "epoch": 0.8728622136172959,
      "grad_norm": 1.0923051834106445,
      "learning_rate": 0.00035416816991345107,
      "loss": 2.677669048309326,
      "step": 5410
    },
    {
      "epoch": 0.8730235559858018,
      "grad_norm": 1.0337135791778564,
      "learning_rate": 0.0003535915875898601,
      "loss": 2.116558790206909,
      "step": 5411
    },
    {
      "epoch": 0.8731848983543078,
      "grad_norm": 1.2511157989501953,
      "learning_rate": 0.00035301433925528957,
      "loss": 2.2430834770202637,
      "step": 5412
    },
    {
      "epoch": 0.8733462407228139,
      "grad_norm": 1.987660527229309,
      "learning_rate": 0.000352436428620984,
      "loss": 2.4697375297546387,
      "step": 5413
    },
    {
      "epoch": 0.8735075830913198,
      "grad_norm": 1.1424839496612549,
      "learning_rate": 0.00035185785940244636,
      "loss": 2.1941380500793457,
      "step": 5414
    },
    {
      "epoch": 0.8736689254598258,
      "grad_norm": 1.1961791515350342,
      "learning_rate": 0.0003512786353194134,
      "loss": 2.4224088191986084,
      "step": 5415
    },
    {
      "epoch": 0.8738302678283317,
      "grad_norm": 1.1810996532440186,
      "learning_rate": 0.00035069876009583234,
      "loss": 2.2677884101867676,
      "step": 5416
    },
    {
      "epoch": 0.8739916101968377,
      "grad_norm": 1.6977691650390625,
      "learning_rate": 0.0003501182374598367,
      "loss": 2.1337881088256836,
      "step": 5417
    },
    {
      "epoch": 0.8741529525653436,
      "grad_norm": 0.8266363739967346,
      "learning_rate": 0.0003495370711437221,
      "loss": 2.230032205581665,
      "step": 5418
    },
    {
      "epoch": 0.8743142949338496,
      "grad_norm": 1.153975009918213,
      "learning_rate": 0.0003489552648839231,
      "loss": 2.082284927368164,
      "step": 5419
    },
    {
      "epoch": 0.8744756373023556,
      "grad_norm": 0.9443587064743042,
      "learning_rate": 0.00034837282242098776,
      "loss": 2.0291178226470947,
      "step": 5420
    },
    {
      "epoch": 0.8746369796708615,
      "grad_norm": 1.026201605796814,
      "learning_rate": 0.0003477897474995552,
      "loss": 2.4238595962524414,
      "step": 5421
    },
    {
      "epoch": 0.8747983220393676,
      "grad_norm": 1.096345067024231,
      "learning_rate": 0.0003472060438683302,
      "loss": 2.5789341926574707,
      "step": 5422
    },
    {
      "epoch": 0.8749596644078735,
      "grad_norm": 1.0405179262161255,
      "learning_rate": 0.0003466217152800598,
      "loss": 2.4720911979675293,
      "step": 5423
    },
    {
      "epoch": 0.8751210067763795,
      "grad_norm": 1.1705354452133179,
      "learning_rate": 0.0003460367654915092,
      "loss": 3.0034594535827637,
      "step": 5424
    },
    {
      "epoch": 0.8752823491448855,
      "grad_norm": 1.0822592973709106,
      "learning_rate": 0.0003454511982634372,
      "loss": 2.394007682800293,
      "step": 5425
    },
    {
      "epoch": 0.8754436915133914,
      "grad_norm": 1.0198845863342285,
      "learning_rate": 0.0003448650173605723,
      "loss": 2.3081841468811035,
      "step": 5426
    },
    {
      "epoch": 0.8756050338818974,
      "grad_norm": 1.3670470714569092,
      "learning_rate": 0.00034427822655158847,
      "loss": 1.83518385887146,
      "step": 5427
    },
    {
      "epoch": 0.8757663762504033,
      "grad_norm": 1.0741620063781738,
      "learning_rate": 0.0003436908296090808,
      "loss": 2.554466724395752,
      "step": 5428
    },
    {
      "epoch": 0.8759277186189093,
      "grad_norm": 1.2584068775177002,
      "learning_rate": 0.00034310283030954147,
      "loss": 2.4312798976898193,
      "step": 5429
    },
    {
      "epoch": 0.8760890609874153,
      "grad_norm": 1.0747883319854736,
      "learning_rate": 0.0003425142324333353,
      "loss": 2.149176597595215,
      "step": 5430
    },
    {
      "epoch": 0.8762504033559213,
      "grad_norm": 0.9623032212257385,
      "learning_rate": 0.00034192503976467525,
      "loss": 2.245181083679199,
      "step": 5431
    },
    {
      "epoch": 0.8764117457244273,
      "grad_norm": 0.9377937316894531,
      "learning_rate": 0.00034133525609159884,
      "loss": 2.611821413040161,
      "step": 5432
    },
    {
      "epoch": 0.8765730880929332,
      "grad_norm": 1.0926390886306763,
      "learning_rate": 0.00034074488520594254,
      "loss": 2.1547722816467285,
      "step": 5433
    },
    {
      "epoch": 0.8767344304614392,
      "grad_norm": 1.0985065698623657,
      "learning_rate": 0.0003401539309033189,
      "loss": 2.245410919189453,
      "step": 5434
    },
    {
      "epoch": 0.8768957728299451,
      "grad_norm": 1.1862330436706543,
      "learning_rate": 0.0003395623969830908,
      "loss": 2.1721606254577637,
      "step": 5435
    },
    {
      "epoch": 0.8770571151984511,
      "grad_norm": 1.019092082977295,
      "learning_rate": 0.00033897028724834774,
      "loss": 2.2867043018341064,
      "step": 5436
    },
    {
      "epoch": 0.877218457566957,
      "grad_norm": 1.025211215019226,
      "learning_rate": 0.00033837760550588155,
      "loss": 2.3788414001464844,
      "step": 5437
    },
    {
      "epoch": 0.877379799935463,
      "grad_norm": 1.0469999313354492,
      "learning_rate": 0.00033778435556616114,
      "loss": 2.221471071243286,
      "step": 5438
    },
    {
      "epoch": 0.8775411423039691,
      "grad_norm": 1.2523677349090576,
      "learning_rate": 0.0003371905412433089,
      "loss": 2.519897937774658,
      "step": 5439
    },
    {
      "epoch": 0.877702484672475,
      "grad_norm": 0.9623591303825378,
      "learning_rate": 0.0003365961663550755,
      "loss": 1.9421168565750122,
      "step": 5440
    },
    {
      "epoch": 0.877863827040981,
      "grad_norm": 1.2778462171554565,
      "learning_rate": 0.0003360012347228155,
      "loss": 1.633262038230896,
      "step": 5441
    },
    {
      "epoch": 0.8780251694094869,
      "grad_norm": 1.0199251174926758,
      "learning_rate": 0.00033540575017146324,
      "loss": 2.470047950744629,
      "step": 5442
    },
    {
      "epoch": 0.8781865117779929,
      "grad_norm": 1.199858546257019,
      "learning_rate": 0.00033480971652950753,
      "loss": 2.675017833709717,
      "step": 5443
    },
    {
      "epoch": 0.8783478541464989,
      "grad_norm": 0.9630619883537292,
      "learning_rate": 0.00033421313762896766,
      "loss": 2.102978229522705,
      "step": 5444
    },
    {
      "epoch": 0.8785091965150048,
      "grad_norm": 0.8255236148834229,
      "learning_rate": 0.00033361601730536836,
      "loss": 1.9399268627166748,
      "step": 5445
    },
    {
      "epoch": 0.8786705388835108,
      "grad_norm": 1.2265862226486206,
      "learning_rate": 0.0003330183593977152,
      "loss": 2.171901226043701,
      "step": 5446
    },
    {
      "epoch": 0.8788318812520168,
      "grad_norm": 1.0704628229141235,
      "learning_rate": 0.00033242016774847016,
      "loss": 2.1281468868255615,
      "step": 5447
    },
    {
      "epoch": 0.8789932236205228,
      "grad_norm": 1.9990161657333374,
      "learning_rate": 0.0003318214462035266,
      "loss": 2.6370697021484375,
      "step": 5448
    },
    {
      "epoch": 0.8791545659890287,
      "grad_norm": 1.2482987642288208,
      "learning_rate": 0.0003312221986121849,
      "loss": 2.799142599105835,
      "step": 5449
    },
    {
      "epoch": 0.8793159083575347,
      "grad_norm": 0.842339813709259,
      "learning_rate": 0.0003306224288271272,
      "loss": 2.391494035720825,
      "step": 5450
    },
    {
      "epoch": 0.8794772507260407,
      "grad_norm": 1.0890954732894897,
      "learning_rate": 0.00033002214070439325,
      "loss": 2.4471704959869385,
      "step": 5451
    },
    {
      "epoch": 0.8796385930945466,
      "grad_norm": 1.143375277519226,
      "learning_rate": 0.0003294213381033549,
      "loss": 2.0103678703308105,
      "step": 5452
    },
    {
      "epoch": 0.8797999354630526,
      "grad_norm": 1.3342050313949585,
      "learning_rate": 0.00032882002488669206,
      "loss": 2.731510639190674,
      "step": 5453
    },
    {
      "epoch": 0.8799612778315585,
      "grad_norm": 0.8464125990867615,
      "learning_rate": 0.00032821820492036737,
      "loss": 2.176246166229248,
      "step": 5454
    },
    {
      "epoch": 0.8801226202000645,
      "grad_norm": 1.0551097393035889,
      "learning_rate": 0.00032761588207360135,
      "loss": 2.2243056297302246,
      "step": 5455
    },
    {
      "epoch": 0.8802839625685706,
      "grad_norm": 1.2038723230361938,
      "learning_rate": 0.00032701306021884767,
      "loss": 2.186704635620117,
      "step": 5456
    },
    {
      "epoch": 0.8804453049370765,
      "grad_norm": 0.818015456199646,
      "learning_rate": 0.0003264097432317685,
      "loss": 2.038635730743408,
      "step": 5457
    },
    {
      "epoch": 0.8806066473055825,
      "grad_norm": 1.1468853950500488,
      "learning_rate": 0.00032580593499120894,
      "loss": 2.575333595275879,
      "step": 5458
    },
    {
      "epoch": 0.8807679896740884,
      "grad_norm": 1.2028151750564575,
      "learning_rate": 0.00032520163937917255,
      "loss": 2.274503707885742,
      "step": 5459
    },
    {
      "epoch": 0.8809293320425944,
      "grad_norm": 1.115159511566162,
      "learning_rate": 0.0003245968602807965,
      "loss": 2.758232593536377,
      "step": 5460
    },
    {
      "epoch": 0.8810906744111003,
      "grad_norm": 0.8592069745063782,
      "learning_rate": 0.00032399160158432606,
      "loss": 2.224698781967163,
      "step": 5461
    },
    {
      "epoch": 0.8812520167796063,
      "grad_norm": 1.0026355981826782,
      "learning_rate": 0.0003233858671810903,
      "loss": 2.279146671295166,
      "step": 5462
    },
    {
      "epoch": 0.8814133591481123,
      "grad_norm": 0.9892069101333618,
      "learning_rate": 0.0003227796609654765,
      "loss": 2.188797950744629,
      "step": 5463
    },
    {
      "epoch": 0.8815747015166182,
      "grad_norm": 1.2399487495422363,
      "learning_rate": 0.00032217298683490526,
      "loss": 2.2948663234710693,
      "step": 5464
    },
    {
      "epoch": 0.8817360438851243,
      "grad_norm": 0.8714036345481873,
      "learning_rate": 0.00032156584868980566,
      "loss": 2.1032204627990723,
      "step": 5465
    },
    {
      "epoch": 0.8818973862536302,
      "grad_norm": 1.3020520210266113,
      "learning_rate": 0.00032095825043358976,
      "loss": 2.3110601902008057,
      "step": 5466
    },
    {
      "epoch": 0.8820587286221362,
      "grad_norm": 1.0576112270355225,
      "learning_rate": 0.0003203501959726282,
      "loss": 2.433696746826172,
      "step": 5467
    },
    {
      "epoch": 0.8822200709906421,
      "grad_norm": 0.9047051072120667,
      "learning_rate": 0.0003197416892162242,
      "loss": 1.8283288478851318,
      "step": 5468
    },
    {
      "epoch": 0.8823814133591481,
      "grad_norm": 1.0048341751098633,
      "learning_rate": 0.00031913273407658914,
      "loss": 2.315234899520874,
      "step": 5469
    },
    {
      "epoch": 0.8825427557276541,
      "grad_norm": 1.0097500085830688,
      "learning_rate": 0.0003185233344688171,
      "loss": 2.1138017177581787,
      "step": 5470
    },
    {
      "epoch": 0.88270409809616,
      "grad_norm": 1.1329950094223022,
      "learning_rate": 0.0003179134943108597,
      "loss": 2.352112293243408,
      "step": 5471
    },
    {
      "epoch": 0.882865440464666,
      "grad_norm": 1.1909912824630737,
      "learning_rate": 0.00031730321752350096,
      "loss": 2.4709911346435547,
      "step": 5472
    },
    {
      "epoch": 0.883026782833172,
      "grad_norm": 0.9942952394485474,
      "learning_rate": 0.000316692508030332,
      "loss": 2.3382256031036377,
      "step": 5473
    },
    {
      "epoch": 0.883188125201678,
      "grad_norm": 1.2789387702941895,
      "learning_rate": 0.0003160813697577261,
      "loss": 2.109579563140869,
      "step": 5474
    },
    {
      "epoch": 0.883349467570184,
      "grad_norm": 1.203315258026123,
      "learning_rate": 0.000315469806634813,
      "loss": 2.3128223419189453,
      "step": 5475
    },
    {
      "epoch": 0.8835108099386899,
      "grad_norm": 1.328345537185669,
      "learning_rate": 0.00031485782259345406,
      "loss": 2.4394655227661133,
      "step": 5476
    },
    {
      "epoch": 0.8836721523071959,
      "grad_norm": 1.0513962507247925,
      "learning_rate": 0.0003142454215682168,
      "loss": 2.0579347610473633,
      "step": 5477
    },
    {
      "epoch": 0.8838334946757018,
      "grad_norm": 0.8532996773719788,
      "learning_rate": 0.00031363260749634944,
      "loss": 2.2327117919921875,
      "step": 5478
    },
    {
      "epoch": 0.8839948370442078,
      "grad_norm": 0.9654008150100708,
      "learning_rate": 0.000313019384317756,
      "loss": 1.9791479110717773,
      "step": 5479
    },
    {
      "epoch": 0.8841561794127137,
      "grad_norm": 1.1997852325439453,
      "learning_rate": 0.00031240575597497053,
      "loss": 2.041602373123169,
      "step": 5480
    },
    {
      "epoch": 0.8843175217812197,
      "grad_norm": 0.9347178936004639,
      "learning_rate": 0.000311791726413132,
      "loss": 2.2005434036254883,
      "step": 5481
    },
    {
      "epoch": 0.8844788641497258,
      "grad_norm": 1.083848476409912,
      "learning_rate": 0.00031117729957995914,
      "loss": 2.685269832611084,
      "step": 5482
    },
    {
      "epoch": 0.8846402065182317,
      "grad_norm": 1.3145090341567993,
      "learning_rate": 0.00031056247942572445,
      "loss": 2.3546481132507324,
      "step": 5483
    },
    {
      "epoch": 0.8848015488867377,
      "grad_norm": 0.9224599003791809,
      "learning_rate": 0.0003099472699032294,
      "loss": 2.411229133605957,
      "step": 5484
    },
    {
      "epoch": 0.8849628912552436,
      "grad_norm": 0.9115111827850342,
      "learning_rate": 0.00030933167496777875,
      "loss": 2.377643585205078,
      "step": 5485
    },
    {
      "epoch": 0.8851242336237496,
      "grad_norm": 0.903043270111084,
      "learning_rate": 0.000308715698577155,
      "loss": 2.264206647872925,
      "step": 5486
    },
    {
      "epoch": 0.8852855759922555,
      "grad_norm": 0.9932525157928467,
      "learning_rate": 0.00030809934469159335,
      "loss": 1.8604127168655396,
      "step": 5487
    },
    {
      "epoch": 0.8854469183607615,
      "grad_norm": 0.933732807636261,
      "learning_rate": 0.00030748261727375585,
      "loss": 2.286543846130371,
      "step": 5488
    },
    {
      "epoch": 0.8856082607292675,
      "grad_norm": 1.1272261142730713,
      "learning_rate": 0.0003068655202887061,
      "loss": 2.0907790660858154,
      "step": 5489
    },
    {
      "epoch": 0.8857696030977735,
      "grad_norm": 0.925712525844574,
      "learning_rate": 0.00030624805770388357,
      "loss": 2.247488260269165,
      "step": 5490
    },
    {
      "epoch": 0.8859309454662795,
      "grad_norm": 0.7671411037445068,
      "learning_rate": 0.0003056302334890786,
      "loss": 1.7113149166107178,
      "step": 5491
    },
    {
      "epoch": 0.8860922878347854,
      "grad_norm": 0.9603378176689148,
      "learning_rate": 0.0003050120516164062,
      "loss": 2.1404199600219727,
      "step": 5492
    },
    {
      "epoch": 0.8862536302032914,
      "grad_norm": 1.0907031297683716,
      "learning_rate": 0.0003043935160602809,
      "loss": 2.012381076812744,
      "step": 5493
    },
    {
      "epoch": 0.8864149725717974,
      "grad_norm": 1.0749303102493286,
      "learning_rate": 0.00030377463079739114,
      "loss": 2.10270619392395,
      "step": 5494
    },
    {
      "epoch": 0.8865763149403033,
      "grad_norm": 0.9620113372802734,
      "learning_rate": 0.0003031553998066738,
      "loss": 2.1399781703948975,
      "step": 5495
    },
    {
      "epoch": 0.8867376573088093,
      "grad_norm": 0.9252629280090332,
      "learning_rate": 0.0003025358270692885,
      "loss": 1.8650249242782593,
      "step": 5496
    },
    {
      "epoch": 0.8868989996773152,
      "grad_norm": 1.1514756679534912,
      "learning_rate": 0.0003019159165685918,
      "loss": 2.0872693061828613,
      "step": 5497
    },
    {
      "epoch": 0.8870603420458212,
      "grad_norm": 0.925308108329773,
      "learning_rate": 0.0003012956722901121,
      "loss": 2.3153293132781982,
      "step": 5498
    },
    {
      "epoch": 0.8872216844143272,
      "grad_norm": 0.8361868858337402,
      "learning_rate": 0.0003006750982215234,
      "loss": 2.0597071647644043,
      "step": 5499
    },
    {
      "epoch": 0.8873830267828332,
      "grad_norm": 1.1494344472885132,
      "learning_rate": 0.00030005419835262036,
      "loss": 2.45347261428833,
      "step": 5500
    },
    {
      "epoch": 0.8875443691513392,
      "grad_norm": 1.0866483449935913,
      "learning_rate": 0.00029943297667529194,
      "loss": 2.592707633972168,
      "step": 5501
    },
    {
      "epoch": 0.8877057115198451,
      "grad_norm": 0.841417670249939,
      "learning_rate": 0.0002988114371834963,
      "loss": 2.2121355533599854,
      "step": 5502
    },
    {
      "epoch": 0.8878670538883511,
      "grad_norm": 1.3615397214889526,
      "learning_rate": 0.0002981895838732348,
      "loss": 2.5173051357269287,
      "step": 5503
    },
    {
      "epoch": 0.888028396256857,
      "grad_norm": 1.1753545999526978,
      "learning_rate": 0.0002975674207425265,
      "loss": 2.279388427734375,
      "step": 5504
    },
    {
      "epoch": 0.888189738625363,
      "grad_norm": 1.223082184791565,
      "learning_rate": 0.00029694495179138203,
      "loss": 2.491241693496704,
      "step": 5505
    },
    {
      "epoch": 0.888351080993869,
      "grad_norm": 1.2049272060394287,
      "learning_rate": 0.0002963221810217786,
      "loss": 2.768693208694458,
      "step": 5506
    },
    {
      "epoch": 0.888512423362375,
      "grad_norm": 0.912580132484436,
      "learning_rate": 0.0002956991124376337,
      "loss": 2.35557222366333,
      "step": 5507
    },
    {
      "epoch": 0.888673765730881,
      "grad_norm": 1.0428227186203003,
      "learning_rate": 0.00029507575004477954,
      "loss": 2.2137556076049805,
      "step": 5508
    },
    {
      "epoch": 0.8888351080993869,
      "grad_norm": 0.9748162627220154,
      "learning_rate": 0.00029445209785093727,
      "loss": 2.0386886596679688,
      "step": 5509
    },
    {
      "epoch": 0.8889964504678929,
      "grad_norm": 1.1758781671524048,
      "learning_rate": 0.00029382815986569115,
      "loss": 2.681055784225464,
      "step": 5510
    },
    {
      "epoch": 0.8891577928363988,
      "grad_norm": 0.973530113697052,
      "learning_rate": 0.00029320394010046315,
      "loss": 2.513822555541992,
      "step": 5511
    },
    {
      "epoch": 0.8893191352049048,
      "grad_norm": 2.256565570831299,
      "learning_rate": 0.00029257944256848653,
      "loss": 1.8556597232818604,
      "step": 5512
    },
    {
      "epoch": 0.8894804775734108,
      "grad_norm": 1.1046249866485596,
      "learning_rate": 0.00029195467128478045,
      "loss": 2.0318174362182617,
      "step": 5513
    },
    {
      "epoch": 0.8896418199419167,
      "grad_norm": 1.2170305252075195,
      "learning_rate": 0.00029132963026612407,
      "loss": 2.3336691856384277,
      "step": 5514
    },
    {
      "epoch": 0.8898031623104227,
      "grad_norm": 1.0631428956985474,
      "learning_rate": 0.000290704323531031,
      "loss": 2.4909167289733887,
      "step": 5515
    },
    {
      "epoch": 0.8899645046789287,
      "grad_norm": 1.4659702777862549,
      "learning_rate": 0.00029007875509972275,
      "loss": 2.536280870437622,
      "step": 5516
    },
    {
      "epoch": 0.8901258470474347,
      "grad_norm": 1.0957876443862915,
      "learning_rate": 0.0002894529289941037,
      "loss": 2.4672603607177734,
      "step": 5517
    },
    {
      "epoch": 0.8902871894159406,
      "grad_norm": 0.9742821455001831,
      "learning_rate": 0.00028882684923773455,
      "loss": 2.1576948165893555,
      "step": 5518
    },
    {
      "epoch": 0.8904485317844466,
      "grad_norm": 0.8738653063774109,
      "learning_rate": 0.00028820051985580696,
      "loss": 2.3969826698303223,
      "step": 5519
    },
    {
      "epoch": 0.8906098741529526,
      "grad_norm": 0.7905071973800659,
      "learning_rate": 0.0002875739448751176,
      "loss": 1.992425799369812,
      "step": 5520
    },
    {
      "epoch": 0.8907712165214585,
      "grad_norm": 1.705178141593933,
      "learning_rate": 0.00028694712832404195,
      "loss": 1.9626140594482422,
      "step": 5521
    },
    {
      "epoch": 0.8909325588899645,
      "grad_norm": 1.0456724166870117,
      "learning_rate": 0.00028632007423250855,
      "loss": 2.4279277324676514,
      "step": 5522
    },
    {
      "epoch": 0.8910939012584704,
      "grad_norm": 1.1809219121932983,
      "learning_rate": 0.0002856927866319733,
      "loss": 2.3450093269348145,
      "step": 5523
    },
    {
      "epoch": 0.8912552436269764,
      "grad_norm": 0.8299116492271423,
      "learning_rate": 0.00028506526955539334,
      "loss": 2.3694238662719727,
      "step": 5524
    },
    {
      "epoch": 0.8914165859954825,
      "grad_norm": 1.219104290008545,
      "learning_rate": 0.00028443752703720106,
      "loss": 2.582994222640991,
      "step": 5525
    },
    {
      "epoch": 0.8915779283639884,
      "grad_norm": 1.0817569494247437,
      "learning_rate": 0.0002838095631132784,
      "loss": 2.2098190784454346,
      "step": 5526
    },
    {
      "epoch": 0.8917392707324944,
      "grad_norm": 1.0800528526306152,
      "learning_rate": 0.00028318138182093053,
      "loss": 1.9450052976608276,
      "step": 5527
    },
    {
      "epoch": 0.8919006131010003,
      "grad_norm": 1.0676987171173096,
      "learning_rate": 0.0002825529871988604,
      "loss": 2.1025755405426025,
      "step": 5528
    },
    {
      "epoch": 0.8920619554695063,
      "grad_norm": 1.0162729024887085,
      "learning_rate": 0.00028192438328714244,
      "loss": 2.6638894081115723,
      "step": 5529
    },
    {
      "epoch": 0.8922232978380122,
      "grad_norm": 1.1688464879989624,
      "learning_rate": 0.00028129557412719634,
      "loss": 2.518289089202881,
      "step": 5530
    },
    {
      "epoch": 0.8923846402065182,
      "grad_norm": 1.0499393939971924,
      "learning_rate": 0.000280666563761762,
      "loss": 2.7730281352996826,
      "step": 5531
    },
    {
      "epoch": 0.8925459825750242,
      "grad_norm": 21.3468017578125,
      "learning_rate": 0.0002800373562348721,
      "loss": 2.4861483573913574,
      "step": 5532
    },
    {
      "epoch": 0.8927073249435302,
      "grad_norm": 1.1834911108016968,
      "learning_rate": 0.00027940795559182764,
      "loss": 2.246339797973633,
      "step": 5533
    },
    {
      "epoch": 0.8928686673120362,
      "grad_norm": 1.3173191547393799,
      "learning_rate": 0.0002787783658791707,
      "loss": 2.2453296184539795,
      "step": 5534
    },
    {
      "epoch": 0.8930300096805421,
      "grad_norm": 0.9010804295539856,
      "learning_rate": 0.00027814859114465935,
      "loss": 2.5663070678710938,
      "step": 5535
    },
    {
      "epoch": 0.8931913520490481,
      "grad_norm": 1.481069803237915,
      "learning_rate": 0.0002775186354372408,
      "loss": 2.541877269744873,
      "step": 5536
    },
    {
      "epoch": 0.893352694417554,
      "grad_norm": 1.1491708755493164,
      "learning_rate": 0.000276888502807026,
      "loss": 2.0463311672210693,
      "step": 5537
    },
    {
      "epoch": 0.89351403678606,
      "grad_norm": 0.8745516538619995,
      "learning_rate": 0.0002762581973052633,
      "loss": 2.186627149581909,
      "step": 5538
    },
    {
      "epoch": 0.893675379154566,
      "grad_norm": 1.0035120248794556,
      "learning_rate": 0.00027562772298431247,
      "loss": 2.1206374168395996,
      "step": 5539
    },
    {
      "epoch": 0.8938367215230719,
      "grad_norm": 1.9908696413040161,
      "learning_rate": 0.0002749970838976188,
      "loss": 2.3241844177246094,
      "step": 5540
    },
    {
      "epoch": 0.8939980638915779,
      "grad_norm": 0.7891056537628174,
      "learning_rate": 0.0002743662840996866,
      "loss": 1.941540241241455,
      "step": 5541
    },
    {
      "epoch": 0.8941594062600839,
      "grad_norm": 1.2804313898086548,
      "learning_rate": 0.0002737353276460537,
      "loss": 2.0683705806732178,
      "step": 5542
    },
    {
      "epoch": 0.8943207486285899,
      "grad_norm": 1.0032345056533813,
      "learning_rate": 0.000273104218593265,
      "loss": 2.5284571647644043,
      "step": 5543
    },
    {
      "epoch": 0.8944820909970959,
      "grad_norm": 0.97925865650177,
      "learning_rate": 0.0002724729609988465,
      "loss": 2.5666911602020264,
      "step": 5544
    },
    {
      "epoch": 0.8946434333656018,
      "grad_norm": 1.1389484405517578,
      "learning_rate": 0.0002718415589212791,
      "loss": 2.244483232498169,
      "step": 5545
    },
    {
      "epoch": 0.8948047757341078,
      "grad_norm": 1.0301704406738281,
      "learning_rate": 0.0002712100164199728,
      "loss": 2.31177020072937,
      "step": 5546
    },
    {
      "epoch": 0.8949661181026137,
      "grad_norm": 0.9342659115791321,
      "learning_rate": 0.0002705783375552402,
      "loss": 2.055555582046509,
      "step": 5547
    },
    {
      "epoch": 0.8951274604711197,
      "grad_norm": 0.9784713387489319,
      "learning_rate": 0.0002699465263882708,
      "loss": 2.7634172439575195,
      "step": 5548
    },
    {
      "epoch": 0.8952888028396256,
      "grad_norm": 0.8515415787696838,
      "learning_rate": 0.0002693145869811046,
      "loss": 2.4464259147644043,
      "step": 5549
    },
    {
      "epoch": 0.8954501452081317,
      "grad_norm": 1.1819885969161987,
      "learning_rate": 0.00026868252339660607,
      "loss": 2.550548791885376,
      "step": 5550
    },
    {
      "epoch": 0.8956114875766377,
      "grad_norm": 1.0188645124435425,
      "learning_rate": 0.0002680503396984382,
      "loss": 2.395747184753418,
      "step": 5551
    },
    {
      "epoch": 0.8957728299451436,
      "grad_norm": 1.304548978805542,
      "learning_rate": 0.0002674180399510358,
      "loss": 1.7890019416809082,
      "step": 5552
    },
    {
      "epoch": 0.8959341723136496,
      "grad_norm": 1.0857487916946411,
      "learning_rate": 0.00026678562821958043,
      "loss": 2.346419334411621,
      "step": 5553
    },
    {
      "epoch": 0.8960955146821555,
      "grad_norm": 0.9614415764808655,
      "learning_rate": 0.000266153108569973,
      "loss": 1.8583314418792725,
      "step": 5554
    },
    {
      "epoch": 0.8962568570506615,
      "grad_norm": 1.0183161497116089,
      "learning_rate": 0.0002655204850688085,
      "loss": 2.333779811859131,
      "step": 5555
    },
    {
      "epoch": 0.8964181994191674,
      "grad_norm": 0.9637329578399658,
      "learning_rate": 0.0002648877617833498,
      "loss": 1.9922456741333008,
      "step": 5556
    },
    {
      "epoch": 0.8965795417876734,
      "grad_norm": 1.4243407249450684,
      "learning_rate": 0.000264254942781501,
      "loss": 2.521465301513672,
      "step": 5557
    },
    {
      "epoch": 0.8967408841561794,
      "grad_norm": 7.819278240203857,
      "learning_rate": 0.0002636220321317816,
      "loss": 1.9440966844558716,
      "step": 5558
    },
    {
      "epoch": 0.8969022265246854,
      "grad_norm": 1.0657241344451904,
      "learning_rate": 0.0002629890339033006,
      "loss": 2.2348828315734863,
      "step": 5559
    },
    {
      "epoch": 0.8970635688931914,
      "grad_norm": 1.0522104501724243,
      "learning_rate": 0.0002623559521657296,
      "loss": 2.3155903816223145,
      "step": 5560
    },
    {
      "epoch": 0.8972249112616973,
      "grad_norm": 2.0060789585113525,
      "learning_rate": 0.0002617227909892777,
      "loss": 2.9205985069274902,
      "step": 5561
    },
    {
      "epoch": 0.8973862536302033,
      "grad_norm": 1.0605460405349731,
      "learning_rate": 0.0002610895544446641,
      "loss": 2.354929208755493,
      "step": 5562
    },
    {
      "epoch": 0.8975475959987093,
      "grad_norm": 1.026746153831482,
      "learning_rate": 0.0002604562466030931,
      "loss": 2.2781243324279785,
      "step": 5563
    },
    {
      "epoch": 0.8977089383672152,
      "grad_norm": 1.6673402786254883,
      "learning_rate": 0.000259822871536227,
      "loss": 2.8787918090820312,
      "step": 5564
    },
    {
      "epoch": 0.8978702807357212,
      "grad_norm": 0.9401122331619263,
      "learning_rate": 0.0002591894333161604,
      "loss": 2.045189619064331,
      "step": 5565
    },
    {
      "epoch": 0.8980316231042271,
      "grad_norm": 0.903829038143158,
      "learning_rate": 0.00025855593601539415,
      "loss": 2.3431050777435303,
      "step": 5566
    },
    {
      "epoch": 0.8981929654727331,
      "grad_norm": 1.15309739112854,
      "learning_rate": 0.00025792238370680846,
      "loss": 2.155367851257324,
      "step": 5567
    },
    {
      "epoch": 0.8983543078412392,
      "grad_norm": 1.2626910209655762,
      "learning_rate": 0.00025728878046363783,
      "loss": 2.7530760765075684,
      "step": 5568
    },
    {
      "epoch": 0.8985156502097451,
      "grad_norm": 1.1928093433380127,
      "learning_rate": 0.00025665513035944373,
      "loss": 2.1316771507263184,
      "step": 5569
    },
    {
      "epoch": 0.8986769925782511,
      "grad_norm": 1.1469203233718872,
      "learning_rate": 0.00025602143746808904,
      "loss": 2.1358132362365723,
      "step": 5570
    },
    {
      "epoch": 0.898838334946757,
      "grad_norm": 1.154435396194458,
      "learning_rate": 0.0002553877058637118,
      "loss": 2.631593942642212,
      "step": 5571
    },
    {
      "epoch": 0.898999677315263,
      "grad_norm": 3.9243016242980957,
      "learning_rate": 0.0002547539396206988,
      "loss": 1.9876923561096191,
      "step": 5572
    },
    {
      "epoch": 0.8991610196837689,
      "grad_norm": 1.476759910583496,
      "learning_rate": 0.00025412014281365986,
      "loss": 2.3736090660095215,
      "step": 5573
    },
    {
      "epoch": 0.8993223620522749,
      "grad_norm": 1.2678405046463013,
      "learning_rate": 0.0002534863195174009,
      "loss": 2.1474173069000244,
      "step": 5574
    },
    {
      "epoch": 0.8994837044207808,
      "grad_norm": 1.1806312799453735,
      "learning_rate": 0.00025285247380689836,
      "loss": 2.5393474102020264,
      "step": 5575
    },
    {
      "epoch": 0.8996450467892869,
      "grad_norm": 0.8399229645729065,
      "learning_rate": 0.00025221860975727274,
      "loss": 2.035167694091797,
      "step": 5576
    },
    {
      "epoch": 0.8998063891577929,
      "grad_norm": 1.0311306715011597,
      "learning_rate": 0.0002515847314437625,
      "loss": 2.297898292541504,
      "step": 5577
    },
    {
      "epoch": 0.8999677315262988,
      "grad_norm": 0.9334678053855896,
      "learning_rate": 0.00025095084294169767,
      "loss": 1.7730481624603271,
      "step": 5578
    },
    {
      "epoch": 0.9001290738948048,
      "grad_norm": 1.269159197807312,
      "learning_rate": 0.00025031694832647393,
      "loss": 2.1968119144439697,
      "step": 5579
    },
    {
      "epoch": 0.9002904162633107,
      "grad_norm": 1.0508685111999512,
      "learning_rate": 0.00024968305167352614,
      "loss": 2.404365062713623,
      "step": 5580
    },
    {
      "epoch": 0.9004517586318167,
      "grad_norm": 1.0357694625854492,
      "learning_rate": 0.00024904915705830234,
      "loss": 1.9692527055740356,
      "step": 5581
    },
    {
      "epoch": 0.9006131010003227,
      "grad_norm": 0.9053937196731567,
      "learning_rate": 0.00024841526855623756,
      "loss": 2.1119418144226074,
      "step": 5582
    },
    {
      "epoch": 0.9007744433688286,
      "grad_norm": 0.8492647409439087,
      "learning_rate": 0.0002477813902427272,
      "loss": 2.451328754425049,
      "step": 5583
    },
    {
      "epoch": 0.9009357857373346,
      "grad_norm": 1.1277567148208618,
      "learning_rate": 0.00024714752619310165,
      "loss": 2.3911166191101074,
      "step": 5584
    },
    {
      "epoch": 0.9010971281058406,
      "grad_norm": 0.9501830339431763,
      "learning_rate": 0.0002465136804825991,
      "loss": 2.438875198364258,
      "step": 5585
    },
    {
      "epoch": 0.9012584704743466,
      "grad_norm": 0.7962259650230408,
      "learning_rate": 0.0002458798571863401,
      "loss": 2.201268196105957,
      "step": 5586
    },
    {
      "epoch": 0.9014198128428526,
      "grad_norm": 1.0014179944992065,
      "learning_rate": 0.0002452460603793012,
      "loss": 2.5079338550567627,
      "step": 5587
    },
    {
      "epoch": 0.9015811552113585,
      "grad_norm": 1.4457917213439941,
      "learning_rate": 0.0002446122941362883,
      "loss": 2.2670791149139404,
      "step": 5588
    },
    {
      "epoch": 0.9017424975798645,
      "grad_norm": 0.9832919239997864,
      "learning_rate": 0.00024397856253191105,
      "loss": 2.177170991897583,
      "step": 5589
    },
    {
      "epoch": 0.9019038399483704,
      "grad_norm": 0.9594550132751465,
      "learning_rate": 0.00024334486964055634,
      "loss": 2.546726703643799,
      "step": 5590
    },
    {
      "epoch": 0.9020651823168764,
      "grad_norm": 1.11093270778656,
      "learning_rate": 0.0002427112195363622,
      "loss": 2.76936674118042,
      "step": 5591
    },
    {
      "epoch": 0.9022265246853823,
      "grad_norm": 1.4842475652694702,
      "learning_rate": 0.00024207761629319152,
      "loss": 2.609065055847168,
      "step": 5592
    },
    {
      "epoch": 0.9023878670538884,
      "grad_norm": 0.8593600392341614,
      "learning_rate": 0.00024144406398460592,
      "loss": 2.3500537872314453,
      "step": 5593
    },
    {
      "epoch": 0.9025492094223944,
      "grad_norm": 1.249864101409912,
      "learning_rate": 0.0002408105666838396,
      "loss": 2.1923446655273438,
      "step": 5594
    },
    {
      "epoch": 0.9027105517909003,
      "grad_norm": 1.149857997894287,
      "learning_rate": 0.000240177128463773,
      "loss": 2.0239179134368896,
      "step": 5595
    },
    {
      "epoch": 0.9028718941594063,
      "grad_norm": 1.0417709350585938,
      "learning_rate": 0.0002395437533969069,
      "loss": 1.9601045846939087,
      "step": 5596
    },
    {
      "epoch": 0.9030332365279122,
      "grad_norm": 1.0456228256225586,
      "learning_rate": 0.00023891044555533587,
      "loss": 2.664292812347412,
      "step": 5597
    },
    {
      "epoch": 0.9031945788964182,
      "grad_norm": 0.9664387702941895,
      "learning_rate": 0.0002382772090107223,
      "loss": 2.2449212074279785,
      "step": 5598
    },
    {
      "epoch": 0.9033559212649241,
      "grad_norm": 0.8192553520202637,
      "learning_rate": 0.00023764404783427036,
      "loss": 2.1228692531585693,
      "step": 5599
    },
    {
      "epoch": 0.9035172636334301,
      "grad_norm": 1.783605933189392,
      "learning_rate": 0.00023701096609669953,
      "loss": 2.3340611457824707,
      "step": 5600
    },
    {
      "epoch": 0.9036786060019361,
      "grad_norm": 1.3521244525909424,
      "learning_rate": 0.00023637796786821848,
      "loss": 2.1697378158569336,
      "step": 5601
    },
    {
      "epoch": 0.9038399483704421,
      "grad_norm": 1.2357661724090576,
      "learning_rate": 0.00023574505721849908,
      "loss": 2.2503700256347656,
      "step": 5602
    },
    {
      "epoch": 0.9040012907389481,
      "grad_norm": 1.6084375381469727,
      "learning_rate": 0.00023511223821665026,
      "loss": 2.4573006629943848,
      "step": 5603
    },
    {
      "epoch": 0.904162633107454,
      "grad_norm": 1.2173985242843628,
      "learning_rate": 0.00023447951493119152,
      "loss": 1.9610623121261597,
      "step": 5604
    },
    {
      "epoch": 0.90432397547596,
      "grad_norm": 1.4300898313522339,
      "learning_rate": 0.00023384689143002708,
      "loss": 2.0255255699157715,
      "step": 5605
    },
    {
      "epoch": 0.904485317844466,
      "grad_norm": 1.2413133382797241,
      "learning_rate": 0.00023321437178041966,
      "loss": 3.2995736598968506,
      "step": 5606
    },
    {
      "epoch": 0.9046466602129719,
      "grad_norm": 1.0818594694137573,
      "learning_rate": 0.0002325819600489642,
      "loss": 2.61163330078125,
      "step": 5607
    },
    {
      "epoch": 0.9048080025814779,
      "grad_norm": 1.5663981437683105,
      "learning_rate": 0.00023194966030156187,
      "loss": 2.281949758529663,
      "step": 5608
    },
    {
      "epoch": 0.9049693449499838,
      "grad_norm": 0.8972164392471313,
      "learning_rate": 0.00023131747660339394,
      "loss": 2.306976556777954,
      "step": 5609
    },
    {
      "epoch": 0.9051306873184899,
      "grad_norm": 1.080259919166565,
      "learning_rate": 0.00023068541301889546,
      "loss": 2.1744298934936523,
      "step": 5610
    },
    {
      "epoch": 0.9052920296869958,
      "grad_norm": 1.0184978246688843,
      "learning_rate": 0.0002300534736117292,
      "loss": 1.9990007877349854,
      "step": 5611
    },
    {
      "epoch": 0.9054533720555018,
      "grad_norm": 1.0374751091003418,
      "learning_rate": 0.00022942166244475982,
      "loss": 2.6646275520324707,
      "step": 5612
    },
    {
      "epoch": 0.9056147144240078,
      "grad_norm": 1.0376170873641968,
      "learning_rate": 0.00022878998358002732,
      "loss": 2.1612555980682373,
      "step": 5613
    },
    {
      "epoch": 0.9057760567925137,
      "grad_norm": 0.7809903025627136,
      "learning_rate": 0.00022815844107872097,
      "loss": 2.1976866722106934,
      "step": 5614
    },
    {
      "epoch": 0.9059373991610197,
      "grad_norm": 1.2374649047851562,
      "learning_rate": 0.00022752703900115354,
      "loss": 2.432645797729492,
      "step": 5615
    },
    {
      "epoch": 0.9060987415295256,
      "grad_norm": 1.0441992282867432,
      "learning_rate": 0.00022689578140673501,
      "loss": 2.261195182800293,
      "step": 5616
    },
    {
      "epoch": 0.9062600838980316,
      "grad_norm": 0.9078009128570557,
      "learning_rate": 0.0002262646723539463,
      "loss": 2.2012081146240234,
      "step": 5617
    },
    {
      "epoch": 0.9064214262665375,
      "grad_norm": 1.0305813550949097,
      "learning_rate": 0.0002256337159003134,
      "loss": 2.2967395782470703,
      "step": 5618
    },
    {
      "epoch": 0.9065827686350436,
      "grad_norm": 1.3026955127716064,
      "learning_rate": 0.00022500291610238126,
      "loss": 1.7690856456756592,
      "step": 5619
    },
    {
      "epoch": 0.9067441110035496,
      "grad_norm": 0.8762756586074829,
      "learning_rate": 0.00022437227701568757,
      "loss": 2.211549758911133,
      "step": 5620
    },
    {
      "epoch": 0.9069054533720555,
      "grad_norm": 1.110281229019165,
      "learning_rate": 0.00022374180269473673,
      "loss": 2.1956539154052734,
      "step": 5621
    },
    {
      "epoch": 0.9070667957405615,
      "grad_norm": 2.4370925426483154,
      "learning_rate": 0.00022311149719297406,
      "loss": 2.1619129180908203,
      "step": 5622
    },
    {
      "epoch": 0.9072281381090674,
      "grad_norm": 1.6734799146652222,
      "learning_rate": 0.0002224813645627592,
      "loss": 2.523777723312378,
      "step": 5623
    },
    {
      "epoch": 0.9073894804775734,
      "grad_norm": 1.525912880897522,
      "learning_rate": 0.00022185140885534066,
      "loss": 2.325906276702881,
      "step": 5624
    },
    {
      "epoch": 0.9075508228460794,
      "grad_norm": 1.5625128746032715,
      "learning_rate": 0.00022122163412082927,
      "loss": 2.292111396789551,
      "step": 5625
    },
    {
      "epoch": 0.9077121652145853,
      "grad_norm": 0.9477306604385376,
      "learning_rate": 0.00022059204440817246,
      "loss": 2.0324654579162598,
      "step": 5626
    },
    {
      "epoch": 0.9078735075830913,
      "grad_norm": 1.517592430114746,
      "learning_rate": 0.000219962643765128,
      "loss": 2.1086173057556152,
      "step": 5627
    },
    {
      "epoch": 0.9080348499515973,
      "grad_norm": 1.0354100465774536,
      "learning_rate": 0.00021933343623823813,
      "loss": 1.9451625347137451,
      "step": 5628
    },
    {
      "epoch": 0.9081961923201033,
      "grad_norm": 1.14425790309906,
      "learning_rate": 0.00021870442587280367,
      "loss": 2.5043044090270996,
      "step": 5629
    },
    {
      "epoch": 0.9083575346886092,
      "grad_norm": 1.0156573057174683,
      "learning_rate": 0.00021807561671285765,
      "loss": 2.4250478744506836,
      "step": 5630
    },
    {
      "epoch": 0.9085188770571152,
      "grad_norm": 1.0794063806533813,
      "learning_rate": 0.00021744701280113963,
      "loss": 2.1237707138061523,
      "step": 5631
    },
    {
      "epoch": 0.9086802194256212,
      "grad_norm": 0.8015317320823669,
      "learning_rate": 0.00021681861817906954,
      "loss": 2.592623233795166,
      "step": 5632
    },
    {
      "epoch": 0.9088415617941271,
      "grad_norm": 2.0520336627960205,
      "learning_rate": 0.00021619043688672168,
      "loss": 2.2118802070617676,
      "step": 5633
    },
    {
      "epoch": 0.9090029041626331,
      "grad_norm": 1.2111726999282837,
      "learning_rate": 0.00021556247296279895,
      "loss": 2.130692481994629,
      "step": 5634
    },
    {
      "epoch": 0.909164246531139,
      "grad_norm": 1.3981926441192627,
      "learning_rate": 0.0002149347304446067,
      "loss": 2.3939852714538574,
      "step": 5635
    },
    {
      "epoch": 0.9093255888996451,
      "grad_norm": 1.3451578617095947,
      "learning_rate": 0.00021430721336802668,
      "loss": 1.9365887641906738,
      "step": 5636
    },
    {
      "epoch": 0.909486931268151,
      "grad_norm": 0.9779455661773682,
      "learning_rate": 0.00021367992576749146,
      "loss": 2.459991693496704,
      "step": 5637
    },
    {
      "epoch": 0.909648273636657,
      "grad_norm": 1.073395848274231,
      "learning_rate": 0.00021305287167595806,
      "loss": 1.8851871490478516,
      "step": 5638
    },
    {
      "epoch": 0.909809616005163,
      "grad_norm": 1.463918924331665,
      "learning_rate": 0.00021242605512488248,
      "loss": 2.2075464725494385,
      "step": 5639
    },
    {
      "epoch": 0.9099709583736689,
      "grad_norm": 0.9680071473121643,
      "learning_rate": 0.00021179948014419305,
      "loss": 2.0222461223602295,
      "step": 5640
    },
    {
      "epoch": 0.9101323007421749,
      "grad_norm": 0.8582352995872498,
      "learning_rate": 0.00021117315076226557,
      "loss": 2.075312614440918,
      "step": 5641
    },
    {
      "epoch": 0.9102936431106808,
      "grad_norm": 0.8289976716041565,
      "learning_rate": 0.00021054707100589642,
      "loss": 2.4101686477661133,
      "step": 5642
    },
    {
      "epoch": 0.9104549854791868,
      "grad_norm": 1.3771947622299194,
      "learning_rate": 0.0002099212449002773,
      "loss": 2.2478904724121094,
      "step": 5643
    },
    {
      "epoch": 0.9106163278476928,
      "grad_norm": 0.8733595013618469,
      "learning_rate": 0.00020929567646896905,
      "loss": 2.3248801231384277,
      "step": 5644
    },
    {
      "epoch": 0.9107776702161988,
      "grad_norm": 1.3174593448638916,
      "learning_rate": 0.00020867036973387594,
      "loss": 2.0144424438476562,
      "step": 5645
    },
    {
      "epoch": 0.9109390125847048,
      "grad_norm": 1.4341001510620117,
      "learning_rate": 0.00020804532871521959,
      "loss": 2.579725742340088,
      "step": 5646
    },
    {
      "epoch": 0.9111003549532107,
      "grad_norm": 1.082398772239685,
      "learning_rate": 0.00020742055743151354,
      "loss": 2.3726625442504883,
      "step": 5647
    },
    {
      "epoch": 0.9112616973217167,
      "grad_norm": 0.942253589630127,
      "learning_rate": 0.00020679605989953686,
      "loss": 2.1107029914855957,
      "step": 5648
    },
    {
      "epoch": 0.9114230396902226,
      "grad_norm": 1.5137345790863037,
      "learning_rate": 0.00020617184013430878,
      "loss": 2.2617173194885254,
      "step": 5649
    },
    {
      "epoch": 0.9115843820587286,
      "grad_norm": 0.8263640999794006,
      "learning_rate": 0.00020554790214906277,
      "loss": 2.0456149578094482,
      "step": 5650
    },
    {
      "epoch": 0.9117457244272346,
      "grad_norm": 0.7803864479064941,
      "learning_rate": 0.00020492424995522044,
      "loss": 1.9958741664886475,
      "step": 5651
    },
    {
      "epoch": 0.9119070667957405,
      "grad_norm": 1.5325697660446167,
      "learning_rate": 0.00020430088756236638,
      "loss": 2.498399496078491,
      "step": 5652
    },
    {
      "epoch": 0.9120684091642466,
      "grad_norm": 1.1071646213531494,
      "learning_rate": 0.00020367781897822146,
      "loss": 2.242870807647705,
      "step": 5653
    },
    {
      "epoch": 0.9122297515327525,
      "grad_norm": 1.3090145587921143,
      "learning_rate": 0.00020305504820861806,
      "loss": 2.2114641666412354,
      "step": 5654
    },
    {
      "epoch": 0.9123910939012585,
      "grad_norm": 1.009934663772583,
      "learning_rate": 0.00020243257925747367,
      "loss": 2.073507785797119,
      "step": 5655
    },
    {
      "epoch": 0.9125524362697645,
      "grad_norm": 1.9098128080368042,
      "learning_rate": 0.0002018104161267652,
      "loss": 2.2828454971313477,
      "step": 5656
    },
    {
      "epoch": 0.9127137786382704,
      "grad_norm": 1.0433177947998047,
      "learning_rate": 0.00020118856281650373,
      "loss": 1.9577864408493042,
      "step": 5657
    },
    {
      "epoch": 0.9128751210067764,
      "grad_norm": 1.2063168287277222,
      "learning_rate": 0.00020056702332470804,
      "loss": 2.7299699783325195,
      "step": 5658
    },
    {
      "epoch": 0.9130364633752823,
      "grad_norm": 1.1384243965148926,
      "learning_rate": 0.0001999458016473797,
      "loss": 2.382052421569824,
      "step": 5659
    },
    {
      "epoch": 0.9131978057437883,
      "grad_norm": 0.85708087682724,
      "learning_rate": 0.00019932490177847662,
      "loss": 2.3409743309020996,
      "step": 5660
    },
    {
      "epoch": 0.9133591481122942,
      "grad_norm": 0.9734636545181274,
      "learning_rate": 0.00019870432770988794,
      "loss": 2.3223252296447754,
      "step": 5661
    },
    {
      "epoch": 0.9135204904808003,
      "grad_norm": 0.6780524253845215,
      "learning_rate": 0.0001980840834314082,
      "loss": 1.8385412693023682,
      "step": 5662
    },
    {
      "epoch": 0.9136818328493063,
      "grad_norm": 1.0881025791168213,
      "learning_rate": 0.0001974641729307115,
      "loss": 2.1691417694091797,
      "step": 5663
    },
    {
      "epoch": 0.9138431752178122,
      "grad_norm": 0.9308658838272095,
      "learning_rate": 0.0001968446001933261,
      "loss": 1.6834222078323364,
      "step": 5664
    },
    {
      "epoch": 0.9140045175863182,
      "grad_norm": 0.9290453791618347,
      "learning_rate": 0.00019622536920260885,
      "loss": 2.3611340522766113,
      "step": 5665
    },
    {
      "epoch": 0.9141658599548241,
      "grad_norm": 0.9201470613479614,
      "learning_rate": 0.0001956064839397192,
      "loss": 2.2178330421447754,
      "step": 5666
    },
    {
      "epoch": 0.9143272023233301,
      "grad_norm": 0.9525240659713745,
      "learning_rate": 0.00019498794838359392,
      "loss": 2.4155526161193848,
      "step": 5667
    },
    {
      "epoch": 0.914488544691836,
      "grad_norm": 1.2954165935516357,
      "learning_rate": 0.00019436976651092142,
      "loss": 2.260127544403076,
      "step": 5668
    },
    {
      "epoch": 0.914649887060342,
      "grad_norm": 1.0543497800827026,
      "learning_rate": 0.00019375194229611642,
      "loss": 2.190403938293457,
      "step": 5669
    },
    {
      "epoch": 0.914811229428848,
      "grad_norm": 1.1604104042053223,
      "learning_rate": 0.000193134479711294,
      "loss": 2.2523810863494873,
      "step": 5670
    },
    {
      "epoch": 0.914972571797354,
      "grad_norm": 0.9469138979911804,
      "learning_rate": 0.00019251738272624416,
      "loss": 2.278099536895752,
      "step": 5671
    },
    {
      "epoch": 0.91513391416586,
      "grad_norm": 1.4200997352600098,
      "learning_rate": 0.00019190065530840666,
      "loss": 2.371854543685913,
      "step": 5672
    },
    {
      "epoch": 0.9152952565343659,
      "grad_norm": 0.8571293950080872,
      "learning_rate": 0.00019128430142284502,
      "loss": 2.3401167392730713,
      "step": 5673
    },
    {
      "epoch": 0.9154565989028719,
      "grad_norm": 0.9352531433105469,
      "learning_rate": 0.00019066832503222128,
      "loss": 2.335583448410034,
      "step": 5674
    },
    {
      "epoch": 0.9156179412713779,
      "grad_norm": 0.9115647673606873,
      "learning_rate": 0.0001900527300967706,
      "loss": 2.232111692428589,
      "step": 5675
    },
    {
      "epoch": 0.9157792836398838,
      "grad_norm": 1.1489750146865845,
      "learning_rate": 0.00018943752057427554,
      "loss": 2.274637460708618,
      "step": 5676
    },
    {
      "epoch": 0.9159406260083898,
      "grad_norm": 0.8413456678390503,
      "learning_rate": 0.00018882270042004082,
      "loss": 2.0175604820251465,
      "step": 5677
    },
    {
      "epoch": 0.9161019683768957,
      "grad_norm": 1.5204249620437622,
      "learning_rate": 0.00018820827358686793,
      "loss": 1.9409769773483276,
      "step": 5678
    },
    {
      "epoch": 0.9162633107454018,
      "grad_norm": 0.9241288900375366,
      "learning_rate": 0.00018759424402502956,
      "loss": 2.4738121032714844,
      "step": 5679
    },
    {
      "epoch": 0.9164246531139078,
      "grad_norm": 0.9196511507034302,
      "learning_rate": 0.00018698061568224412,
      "loss": 2.0937139987945557,
      "step": 5680
    },
    {
      "epoch": 0.9165859954824137,
      "grad_norm": 1.0198520421981812,
      "learning_rate": 0.0001863673925036506,
      "loss": 2.3421413898468018,
      "step": 5681
    },
    {
      "epoch": 0.9167473378509197,
      "grad_norm": 1.5263456106185913,
      "learning_rate": 0.0001857545784317833,
      "loss": 2.5949254035949707,
      "step": 5682
    },
    {
      "epoch": 0.9169086802194256,
      "grad_norm": 1.1771920919418335,
      "learning_rate": 0.00018514217740654598,
      "loss": 2.247544288635254,
      "step": 5683
    },
    {
      "epoch": 0.9170700225879316,
      "grad_norm": 0.9807739853858948,
      "learning_rate": 0.00018453019336518703,
      "loss": 2.4328885078430176,
      "step": 5684
    },
    {
      "epoch": 0.9172313649564375,
      "grad_norm": 1.1545618772506714,
      "learning_rate": 0.00018391863024227396,
      "loss": 2.2388992309570312,
      "step": 5685
    },
    {
      "epoch": 0.9173927073249435,
      "grad_norm": 1.5281574726104736,
      "learning_rate": 0.00018330749196966806,
      "loss": 1.9819209575653076,
      "step": 5686
    },
    {
      "epoch": 0.9175540496934494,
      "grad_norm": 1.1162554025650024,
      "learning_rate": 0.00018269678247649905,
      "loss": 2.119269371032715,
      "step": 5687
    },
    {
      "epoch": 0.9177153920619555,
      "grad_norm": 1.1812044382095337,
      "learning_rate": 0.00018208650568914033,
      "loss": 2.632534980773926,
      "step": 5688
    },
    {
      "epoch": 0.9178767344304615,
      "grad_norm": 0.961361825466156,
      "learning_rate": 0.0001814766655311829,
      "loss": 1.9656484127044678,
      "step": 5689
    },
    {
      "epoch": 0.9180380767989674,
      "grad_norm": 0.9282949566841125,
      "learning_rate": 0.00018086726592341084,
      "loss": 2.2909812927246094,
      "step": 5690
    },
    {
      "epoch": 0.9181994191674734,
      "grad_norm": 1.1095192432403564,
      "learning_rate": 0.0001802583107837758,
      "loss": 2.220144271850586,
      "step": 5691
    },
    {
      "epoch": 0.9183607615359793,
      "grad_norm": 0.963973343372345,
      "learning_rate": 0.0001796498040273719,
      "loss": 2.198232650756836,
      "step": 5692
    },
    {
      "epoch": 0.9185221039044853,
      "grad_norm": 0.9558708071708679,
      "learning_rate": 0.00017904174956641028,
      "loss": 2.2990760803222656,
      "step": 5693
    },
    {
      "epoch": 0.9186834462729913,
      "grad_norm": 0.8627705574035645,
      "learning_rate": 0.00017843415131019446,
      "loss": 2.24932861328125,
      "step": 5694
    },
    {
      "epoch": 0.9188447886414972,
      "grad_norm": 0.7364809513092041,
      "learning_rate": 0.0001778270131650948,
      "loss": 2.189875364303589,
      "step": 5695
    },
    {
      "epoch": 0.9190061310100033,
      "grad_norm": 0.8412521481513977,
      "learning_rate": 0.0001772203390345235,
      "loss": 1.7247960567474365,
      "step": 5696
    },
    {
      "epoch": 0.9191674733785092,
      "grad_norm": 0.9373334646224976,
      "learning_rate": 0.0001766141328189097,
      "loss": 2.1843528747558594,
      "step": 5697
    },
    {
      "epoch": 0.9193288157470152,
      "grad_norm": 0.978742778301239,
      "learning_rate": 0.00017600839841567395,
      "loss": 2.03572678565979,
      "step": 5698
    },
    {
      "epoch": 0.9194901581155212,
      "grad_norm": 1.1155085563659668,
      "learning_rate": 0.00017540313971920357,
      "loss": 2.299128293991089,
      "step": 5699
    },
    {
      "epoch": 0.9196515004840271,
      "grad_norm": 1.0188113451004028,
      "learning_rate": 0.0001747983606208275,
      "loss": 1.67110013961792,
      "step": 5700
    },
    {
      "epoch": 0.9198128428525331,
      "grad_norm": 1.0147382020950317,
      "learning_rate": 0.00017419406500879115,
      "loss": 1.9262471199035645,
      "step": 5701
    },
    {
      "epoch": 0.919974185221039,
      "grad_norm": 0.8550005555152893,
      "learning_rate": 0.0001735902567682315,
      "loss": 2.643092393875122,
      "step": 5702
    },
    {
      "epoch": 0.920135527589545,
      "grad_norm": 0.7654330134391785,
      "learning_rate": 0.00017298693978115229,
      "loss": 1.6195316314697266,
      "step": 5703
    },
    {
      "epoch": 0.9202968699580509,
      "grad_norm": 1.3314532041549683,
      "learning_rate": 0.00017238411792639872,
      "loss": 2.208174228668213,
      "step": 5704
    },
    {
      "epoch": 0.920458212326557,
      "grad_norm": 0.8904825448989868,
      "learning_rate": 0.00017178179507963277,
      "loss": 2.1135120391845703,
      "step": 5705
    },
    {
      "epoch": 0.920619554695063,
      "grad_norm": 0.8916188478469849,
      "learning_rate": 0.00017117997511330798,
      "loss": 2.3429718017578125,
      "step": 5706
    },
    {
      "epoch": 0.9207808970635689,
      "grad_norm": 0.8817134499549866,
      "learning_rate": 0.00017057866189664516,
      "loss": 2.1472666263580322,
      "step": 5707
    },
    {
      "epoch": 0.9209422394320749,
      "grad_norm": 1.0025923252105713,
      "learning_rate": 0.0001699778592956069,
      "loss": 2.439089298248291,
      "step": 5708
    },
    {
      "epoch": 0.9211035818005808,
      "grad_norm": 1.0987021923065186,
      "learning_rate": 0.00016937757117287277,
      "loss": 2.541459560394287,
      "step": 5709
    },
    {
      "epoch": 0.9212649241690868,
      "grad_norm": 1.2858078479766846,
      "learning_rate": 0.00016877780138781513,
      "loss": 2.7884340286254883,
      "step": 5710
    },
    {
      "epoch": 0.9214262665375927,
      "grad_norm": 1.0848864316940308,
      "learning_rate": 0.0001681785537964734,
      "loss": 2.14947509765625,
      "step": 5711
    },
    {
      "epoch": 0.9215876089060987,
      "grad_norm": 0.9196638464927673,
      "learning_rate": 0.00016757983225152985,
      "loss": 1.9761983156204224,
      "step": 5712
    },
    {
      "epoch": 0.9217489512746048,
      "grad_norm": 0.9764648079872131,
      "learning_rate": 0.00016698164060228483,
      "loss": 2.1239700317382812,
      "step": 5713
    },
    {
      "epoch": 0.9219102936431107,
      "grad_norm": 1.0174559354782104,
      "learning_rate": 0.00016638398269463168,
      "loss": 2.2300827503204346,
      "step": 5714
    },
    {
      "epoch": 0.9220716360116167,
      "grad_norm": 0.8550124764442444,
      "learning_rate": 0.0001657868623710323,
      "loss": 2.166463851928711,
      "step": 5715
    },
    {
      "epoch": 0.9222329783801226,
      "grad_norm": 1.0011323690414429,
      "learning_rate": 0.00016519028347049242,
      "loss": 2.2792763710021973,
      "step": 5716
    },
    {
      "epoch": 0.9223943207486286,
      "grad_norm": 1.1774698495864868,
      "learning_rate": 0.0001645942498285367,
      "loss": 2.388126850128174,
      "step": 5717
    },
    {
      "epoch": 0.9225556631171346,
      "grad_norm": 1.3098596334457397,
      "learning_rate": 0.00016399876527718456,
      "loss": 2.026521921157837,
      "step": 5718
    },
    {
      "epoch": 0.9227170054856405,
      "grad_norm": 0.9199739694595337,
      "learning_rate": 0.00016340383364492458,
      "loss": 1.9724229574203491,
      "step": 5719
    },
    {
      "epoch": 0.9228783478541465,
      "grad_norm": 1.1430251598358154,
      "learning_rate": 0.00016280945875669114,
      "loss": 2.6337239742279053,
      "step": 5720
    },
    {
      "epoch": 0.9230396902226524,
      "grad_norm": 1.7001283168792725,
      "learning_rate": 0.0001622156444338389,
      "loss": 2.7091288566589355,
      "step": 5721
    },
    {
      "epoch": 0.9232010325911585,
      "grad_norm": 1.6020210981369019,
      "learning_rate": 0.00016162239449411854,
      "loss": 2.231804847717285,
      "step": 5722
    },
    {
      "epoch": 0.9233623749596644,
      "grad_norm": 1.1688969135284424,
      "learning_rate": 0.0001610297127516523,
      "loss": 2.2509193420410156,
      "step": 5723
    },
    {
      "epoch": 0.9235237173281704,
      "grad_norm": 1.1931941509246826,
      "learning_rate": 0.00016043760301690928,
      "loss": 2.4386777877807617,
      "step": 5724
    },
    {
      "epoch": 0.9236850596966764,
      "grad_norm": 1.1874035596847534,
      "learning_rate": 0.0001598460690966812,
      "loss": 2.7518625259399414,
      "step": 5725
    },
    {
      "epoch": 0.9238464020651823,
      "grad_norm": 0.9364150166511536,
      "learning_rate": 0.00015925511479405747,
      "loss": 2.0448825359344482,
      "step": 5726
    },
    {
      "epoch": 0.9240077444336883,
      "grad_norm": 1.0809872150421143,
      "learning_rate": 0.00015866474390840125,
      "loss": 2.5006179809570312,
      "step": 5727
    },
    {
      "epoch": 0.9241690868021942,
      "grad_norm": 0.8605847954750061,
      "learning_rate": 0.00015807496023532473,
      "loss": 2.116743564605713,
      "step": 5728
    },
    {
      "epoch": 0.9243304291707002,
      "grad_norm": 1.4062340259552002,
      "learning_rate": 0.00015748576756666478,
      "loss": 2.7260944843292236,
      "step": 5729
    },
    {
      "epoch": 0.9244917715392061,
      "grad_norm": 1.074617862701416,
      "learning_rate": 0.00015689716969045846,
      "loss": 2.3268566131591797,
      "step": 5730
    },
    {
      "epoch": 0.9246531139077122,
      "grad_norm": 2.0925838947296143,
      "learning_rate": 0.00015630917039091919,
      "loss": 2.3293159008026123,
      "step": 5731
    },
    {
      "epoch": 0.9248144562762182,
      "grad_norm": 10.231657028198242,
      "learning_rate": 0.0001557217734484116,
      "loss": 2.6329805850982666,
      "step": 5732
    },
    {
      "epoch": 0.9249757986447241,
      "grad_norm": 1.128654956817627,
      "learning_rate": 0.00015513498263942778,
      "loss": 2.3465933799743652,
      "step": 5733
    },
    {
      "epoch": 0.9251371410132301,
      "grad_norm": 1.4090533256530762,
      "learning_rate": 0.00015454880173656284,
      "loss": 2.3717846870422363,
      "step": 5734
    },
    {
      "epoch": 0.925298483381736,
      "grad_norm": 0.9626041054725647,
      "learning_rate": 0.00015396323450849082,
      "loss": 2.598062515258789,
      "step": 5735
    },
    {
      "epoch": 0.925459825750242,
      "grad_norm": 1.134602665901184,
      "learning_rate": 0.00015337828471994022,
      "loss": 2.0947327613830566,
      "step": 5736
    },
    {
      "epoch": 0.925621168118748,
      "grad_norm": 1.6202428340911865,
      "learning_rate": 0.00015279395613166985,
      "loss": 2.030475378036499,
      "step": 5737
    },
    {
      "epoch": 0.9257825104872539,
      "grad_norm": 1.3391255140304565,
      "learning_rate": 0.00015221025250044485,
      "loss": 2.2924482822418213,
      "step": 5738
    },
    {
      "epoch": 0.92594385285576,
      "grad_norm": 1.4494906663894653,
      "learning_rate": 0.00015162717757901222,
      "loss": 2.014472484588623,
      "step": 5739
    },
    {
      "epoch": 0.9261051952242659,
      "grad_norm": 1.1024408340454102,
      "learning_rate": 0.00015104473511607695,
      "loss": 2.272946834564209,
      "step": 5740
    },
    {
      "epoch": 0.9262665375927719,
      "grad_norm": 1.190788984298706,
      "learning_rate": 0.00015046292885627783,
      "loss": 2.3431482315063477,
      "step": 5741
    },
    {
      "epoch": 0.9264278799612778,
      "grad_norm": 1.1225368976593018,
      "learning_rate": 0.0001498817625401634,
      "loss": 2.32305908203125,
      "step": 5742
    },
    {
      "epoch": 0.9265892223297838,
      "grad_norm": 1.1828852891921997,
      "learning_rate": 0.00014930123990416767,
      "loss": 1.997127890586853,
      "step": 5743
    },
    {
      "epoch": 0.9267505646982898,
      "grad_norm": 0.8229861259460449,
      "learning_rate": 0.0001487213646805866,
      "loss": 2.2215347290039062,
      "step": 5744
    },
    {
      "epoch": 0.9269119070667957,
      "grad_norm": 1.0199021100997925,
      "learning_rate": 0.0001481421405975537,
      "loss": 1.9619497060775757,
      "step": 5745
    },
    {
      "epoch": 0.9270732494353017,
      "grad_norm": 1.202834963798523,
      "learning_rate": 0.00014756357137901604,
      "loss": 2.374830722808838,
      "step": 5746
    },
    {
      "epoch": 0.9272345918038076,
      "grad_norm": 0.9438436627388,
      "learning_rate": 0.00014698566074471052,
      "loss": 2.487265110015869,
      "step": 5747
    },
    {
      "epoch": 0.9273959341723137,
      "grad_norm": 0.9873560070991516,
      "learning_rate": 0.00014640841241013994,
      "loss": 2.2362451553344727,
      "step": 5748
    },
    {
      "epoch": 0.9275572765408197,
      "grad_norm": 0.9455001354217529,
      "learning_rate": 0.000145831830086549,
      "loss": 1.9854199886322021,
      "step": 5749
    },
    {
      "epoch": 0.9277186189093256,
      "grad_norm": 1.198499321937561,
      "learning_rate": 0.00014525591748090034,
      "loss": 2.325623035430908,
      "step": 5750
    },
    {
      "epoch": 0.9278799612778316,
      "grad_norm": 1.0320814847946167,
      "learning_rate": 0.00014468067829585108,
      "loss": 1.9949684143066406,
      "step": 5751
    },
    {
      "epoch": 0.9280413036463375,
      "grad_norm": 1.0316804647445679,
      "learning_rate": 0.00014410611622972874,
      "loss": 2.3974976539611816,
      "step": 5752
    },
    {
      "epoch": 0.9282026460148435,
      "grad_norm": 0.8272411823272705,
      "learning_rate": 0.0001435322349765073,
      "loss": 1.9988446235656738,
      "step": 5753
    },
    {
      "epoch": 0.9283639883833494,
      "grad_norm": 0.9859471917152405,
      "learning_rate": 0.0001429590382257839,
      "loss": 2.399923801422119,
      "step": 5754
    },
    {
      "epoch": 0.9285253307518554,
      "grad_norm": 0.9639829993247986,
      "learning_rate": 0.00014238652966275483,
      "loss": 2.356937885284424,
      "step": 5755
    },
    {
      "epoch": 0.9286866731203615,
      "grad_norm": 2.1713809967041016,
      "learning_rate": 0.0001418147129681917,
      "loss": 2.759000778198242,
      "step": 5756
    },
    {
      "epoch": 0.9288480154888674,
      "grad_norm": 0.7791919112205505,
      "learning_rate": 0.0001412435918184182,
      "loss": 2.0949127674102783,
      "step": 5757
    },
    {
      "epoch": 0.9290093578573734,
      "grad_norm": 1.2896572351455688,
      "learning_rate": 0.00014067316988528616,
      "loss": 2.510648012161255,
      "step": 5758
    },
    {
      "epoch": 0.9291707002258793,
      "grad_norm": 2.5631215572357178,
      "learning_rate": 0.00014010345083615192,
      "loss": 2.687011241912842,
      "step": 5759
    },
    {
      "epoch": 0.9293320425943853,
      "grad_norm": 1.2925641536712646,
      "learning_rate": 0.00013953443833385277,
      "loss": 2.3485093116760254,
      "step": 5760
    },
    {
      "epoch": 0.9294933849628912,
      "grad_norm": 1.6731349229812622,
      "learning_rate": 0.00013896613603668365,
      "loss": 2.376511335372925,
      "step": 5761
    },
    {
      "epoch": 0.9296547273313972,
      "grad_norm": 0.9365823864936829,
      "learning_rate": 0.00013839854759837355,
      "loss": 2.334284782409668,
      "step": 5762
    },
    {
      "epoch": 0.9298160696999032,
      "grad_norm": 0.8431598544120789,
      "learning_rate": 0.00013783167666806152,
      "loss": 2.0719306468963623,
      "step": 5763
    },
    {
      "epoch": 0.9299774120684091,
      "grad_norm": 0.9973412156105042,
      "learning_rate": 0.00013726552689027395,
      "loss": 2.2162020206451416,
      "step": 5764
    },
    {
      "epoch": 0.9301387544369152,
      "grad_norm": 0.811614990234375,
      "learning_rate": 0.00013670010190490074,
      "loss": 2.17056941986084,
      "step": 5765
    },
    {
      "epoch": 0.9303000968054211,
      "grad_norm": 1.1558232307434082,
      "learning_rate": 0.00013613540534717177,
      "loss": 2.136967897415161,
      "step": 5766
    },
    {
      "epoch": 0.9304614391739271,
      "grad_norm": 1.340582251548767,
      "learning_rate": 0.00013557144084763394,
      "loss": 2.134878158569336,
      "step": 5767
    },
    {
      "epoch": 0.930622781542433,
      "grad_norm": 0.9968165159225464,
      "learning_rate": 0.0001350082120321276,
      "loss": 2.234431266784668,
      "step": 5768
    },
    {
      "epoch": 0.930784123910939,
      "grad_norm": 1.1550400257110596,
      "learning_rate": 0.00013444572252176295,
      "loss": 2.2513997554779053,
      "step": 5769
    },
    {
      "epoch": 0.930945466279445,
      "grad_norm": 1.0462074279785156,
      "learning_rate": 0.00013388397593289733,
      "loss": 2.2571821212768555,
      "step": 5770
    },
    {
      "epoch": 0.9311068086479509,
      "grad_norm": 0.816447913646698,
      "learning_rate": 0.00013332297587711174,
      "loss": 1.9479453563690186,
      "step": 5771
    },
    {
      "epoch": 0.9312681510164569,
      "grad_norm": 1.1999518871307373,
      "learning_rate": 0.0001327627259611873,
      "loss": 2.4054574966430664,
      "step": 5772
    },
    {
      "epoch": 0.9314294933849628,
      "grad_norm": 0.9040449857711792,
      "learning_rate": 0.0001322032297870824,
      "loss": 2.228677272796631,
      "step": 5773
    },
    {
      "epoch": 0.9315908357534689,
      "grad_norm": 0.980999231338501,
      "learning_rate": 0.00013164449095190968,
      "loss": 2.244044303894043,
      "step": 5774
    },
    {
      "epoch": 0.9317521781219749,
      "grad_norm": 1.4590102434158325,
      "learning_rate": 0.00013108651304791254,
      "loss": 2.212855815887451,
      "step": 5775
    },
    {
      "epoch": 0.9319135204904808,
      "grad_norm": 0.888088047504425,
      "learning_rate": 0.00013052929966244216,
      "loss": 2.2108330726623535,
      "step": 5776
    },
    {
      "epoch": 0.9320748628589868,
      "grad_norm": 0.9256157875061035,
      "learning_rate": 0.00012997285437793454,
      "loss": 2.28646183013916,
      "step": 5777
    },
    {
      "epoch": 0.9322362052274927,
      "grad_norm": 0.7651255130767822,
      "learning_rate": 0.00012941718077188758,
      "loss": 1.8822304010391235,
      "step": 5778
    },
    {
      "epoch": 0.9323975475959987,
      "grad_norm": 0.8786900043487549,
      "learning_rate": 0.0001288622824168375,
      "loss": 2.2620320320129395,
      "step": 5779
    },
    {
      "epoch": 0.9325588899645046,
      "grad_norm": 0.8944407105445862,
      "learning_rate": 0.00012830816288033657,
      "loss": 2.4667811393737793,
      "step": 5780
    },
    {
      "epoch": 0.9327202323330106,
      "grad_norm": 0.9121222496032715,
      "learning_rate": 0.00012775482572492997,
      "loss": 2.1965079307556152,
      "step": 5781
    },
    {
      "epoch": 0.9328815747015167,
      "grad_norm": 2.085125207901001,
      "learning_rate": 0.00012720227450813243,
      "loss": 2.0059309005737305,
      "step": 5782
    },
    {
      "epoch": 0.9330429170700226,
      "grad_norm": 1.0551033020019531,
      "learning_rate": 0.00012665051278240602,
      "loss": 2.407332420349121,
      "step": 5783
    },
    {
      "epoch": 0.9332042594385286,
      "grad_norm": 0.8057535290718079,
      "learning_rate": 0.00012609954409513707,
      "loss": 2.1973180770874023,
      "step": 5784
    },
    {
      "epoch": 0.9333656018070345,
      "grad_norm": 1.1009727716445923,
      "learning_rate": 0.00012554937198861304,
      "loss": 2.1213200092315674,
      "step": 5785
    },
    {
      "epoch": 0.9335269441755405,
      "grad_norm": 0.854183554649353,
      "learning_rate": 0.00012500000000000006,
      "loss": 1.9396086931228638,
      "step": 5786
    },
    {
      "epoch": 0.9336882865440465,
      "grad_norm": 0.7397076487541199,
      "learning_rate": 0.0001244514316613203,
      "loss": 1.9826385974884033,
      "step": 5787
    },
    {
      "epoch": 0.9338496289125524,
      "grad_norm": 0.9201609492301941,
      "learning_rate": 0.00012390367049942917,
      "loss": 2.260211229324341,
      "step": 5788
    },
    {
      "epoch": 0.9340109712810584,
      "grad_norm": 0.7862886190414429,
      "learning_rate": 0.00012335672003599218,
      "loss": 2.1105246543884277,
      "step": 5789
    },
    {
      "epoch": 0.9341723136495643,
      "grad_norm": 0.8741709589958191,
      "learning_rate": 0.00012281058378746314,
      "loss": 2.0341718196868896,
      "step": 5790
    },
    {
      "epoch": 0.9343336560180704,
      "grad_norm": 1.1192446947097778,
      "learning_rate": 0.00012226526526506093,
      "loss": 2.411837577819824,
      "step": 5791
    },
    {
      "epoch": 0.9344949983865763,
      "grad_norm": 1.1306326389312744,
      "learning_rate": 0.00012172076797474698,
      "loss": 2.200761556625366,
      "step": 5792
    },
    {
      "epoch": 0.9346563407550823,
      "grad_norm": 1.132523536682129,
      "learning_rate": 0.00012117709541720306,
      "loss": 2.574486255645752,
      "step": 5793
    },
    {
      "epoch": 0.9348176831235883,
      "grad_norm": 1.245619535446167,
      "learning_rate": 0.00012063425108780849,
      "loss": 2.232335090637207,
      "step": 5794
    },
    {
      "epoch": 0.9349790254920942,
      "grad_norm": 1.126914143562317,
      "learning_rate": 0.0001200922384766178,
      "loss": 2.217407703399658,
      "step": 5795
    },
    {
      "epoch": 0.9351403678606002,
      "grad_norm": 0.857172429561615,
      "learning_rate": 0.00011955106106833802,
      "loss": 1.995792031288147,
      "step": 5796
    },
    {
      "epoch": 0.9353017102291061,
      "grad_norm": 0.9818625450134277,
      "learning_rate": 0.00011901072234230683,
      "loss": 2.222123146057129,
      "step": 5797
    },
    {
      "epoch": 0.9354630525976121,
      "grad_norm": 1.263013482093811,
      "learning_rate": 0.00011847122577246963,
      "loss": 1.9350614547729492,
      "step": 5798
    },
    {
      "epoch": 0.9356243949661182,
      "grad_norm": 0.8567526340484619,
      "learning_rate": 0.00011793257482735731,
      "loss": 2.100519895553589,
      "step": 5799
    },
    {
      "epoch": 0.9357857373346241,
      "grad_norm": 0.891952633857727,
      "learning_rate": 0.0001173947729700644,
      "loss": 1.8653315305709839,
      "step": 5800
    },
    {
      "epoch": 0.9359470797031301,
      "grad_norm": 0.799082338809967,
      "learning_rate": 0.00011685782365822628,
      "loss": 2.1541996002197266,
      "step": 5801
    },
    {
      "epoch": 0.936108422071636,
      "grad_norm": 1.1924705505371094,
      "learning_rate": 0.00011632173034399726,
      "loss": 2.1359715461730957,
      "step": 5802
    },
    {
      "epoch": 0.936269764440142,
      "grad_norm": 0.8257532715797424,
      "learning_rate": 0.00011578649647402806,
      "loss": 2.1776938438415527,
      "step": 5803
    },
    {
      "epoch": 0.9364311068086479,
      "grad_norm": 0.7926017642021179,
      "learning_rate": 0.000115252125489444,
      "loss": 2.2717952728271484,
      "step": 5804
    },
    {
      "epoch": 0.9365924491771539,
      "grad_norm": 1.0362426042556763,
      "learning_rate": 0.0001147186208258229,
      "loss": 2.0576000213623047,
      "step": 5805
    },
    {
      "epoch": 0.9367537915456599,
      "grad_norm": 1.666021704673767,
      "learning_rate": 0.00011418598591317242,
      "loss": 1.9140390157699585,
      "step": 5806
    },
    {
      "epoch": 0.9369151339141658,
      "grad_norm": 1.0915095806121826,
      "learning_rate": 0.00011365422417590879,
      "loss": 2.109468698501587,
      "step": 5807
    },
    {
      "epoch": 0.9370764762826719,
      "grad_norm": 1.0251480340957642,
      "learning_rate": 0.00011312333903283434,
      "loss": 2.4994559288024902,
      "step": 5808
    },
    {
      "epoch": 0.9372378186511778,
      "grad_norm": 0.8288384675979614,
      "learning_rate": 0.00011259333389711537,
      "loss": 2.150571346282959,
      "step": 5809
    },
    {
      "epoch": 0.9373991610196838,
      "grad_norm": 0.9817918539047241,
      "learning_rate": 0.00011206421217626072,
      "loss": 1.8046014308929443,
      "step": 5810
    },
    {
      "epoch": 0.9375605033881897,
      "grad_norm": 0.7718875408172607,
      "learning_rate": 0.0001115359772720996,
      "loss": 2.1635971069335938,
      "step": 5811
    },
    {
      "epoch": 0.9377218457566957,
      "grad_norm": 1.0525822639465332,
      "learning_rate": 0.00011100863258075941,
      "loss": 1.7250771522521973,
      "step": 5812
    },
    {
      "epoch": 0.9378831881252017,
      "grad_norm": 1.025900959968567,
      "learning_rate": 0.00011048218149264434,
      "loss": 1.9454516172409058,
      "step": 5813
    },
    {
      "epoch": 0.9380445304937076,
      "grad_norm": 0.9799865484237671,
      "learning_rate": 0.00010995662739241347,
      "loss": 1.9819130897521973,
      "step": 5814
    },
    {
      "epoch": 0.9382058728622136,
      "grad_norm": 0.9883753061294556,
      "learning_rate": 0.00010943197365895907,
      "loss": 2.6872100830078125,
      "step": 5815
    },
    {
      "epoch": 0.9383672152307196,
      "grad_norm": 0.9758572578430176,
      "learning_rate": 0.00010890822366538447,
      "loss": 2.2791924476623535,
      "step": 5816
    },
    {
      "epoch": 0.9385285575992256,
      "grad_norm": 0.9894323348999023,
      "learning_rate": 0.00010838538077898292,
      "loss": 2.0613722801208496,
      "step": 5817
    },
    {
      "epoch": 0.9386898999677316,
      "grad_norm": 1.1425235271453857,
      "learning_rate": 0.0001078634483612157,
      "loss": 2.1487534046173096,
      "step": 5818
    },
    {
      "epoch": 0.9388512423362375,
      "grad_norm": 0.9491897225379944,
      "learning_rate": 0.00010734242976769027,
      "loss": 2.026165008544922,
      "step": 5819
    },
    {
      "epoch": 0.9390125847047435,
      "grad_norm": 0.957035481929779,
      "learning_rate": 0.00010682232834813918,
      "loss": 2.447134494781494,
      "step": 5820
    },
    {
      "epoch": 0.9391739270732494,
      "grad_norm": 1.0022966861724854,
      "learning_rate": 0.0001063031474463983,
      "loss": 2.034090042114258,
      "step": 5821
    },
    {
      "epoch": 0.9393352694417554,
      "grad_norm": 0.8614898920059204,
      "learning_rate": 0.00010578489040038503,
      "loss": 1.9220222234725952,
      "step": 5822
    },
    {
      "epoch": 0.9394966118102613,
      "grad_norm": 1.3714327812194824,
      "learning_rate": 0.00010526756054207736,
      "loss": 2.340627670288086,
      "step": 5823
    },
    {
      "epoch": 0.9396579541787673,
      "grad_norm": 0.9928945302963257,
      "learning_rate": 0.00010475116119749223,
      "loss": 2.4968509674072266,
      "step": 5824
    },
    {
      "epoch": 0.9398192965472734,
      "grad_norm": 0.9119070172309875,
      "learning_rate": 0.00010423569568666396,
      "loss": 2.301041841506958,
      "step": 5825
    },
    {
      "epoch": 0.9399806389157793,
      "grad_norm": 1.0850622653961182,
      "learning_rate": 0.00010372116732362305,
      "loss": 2.2670631408691406,
      "step": 5826
    },
    {
      "epoch": 0.9401419812842853,
      "grad_norm": 1.1449264287948608,
      "learning_rate": 0.00010320757941637504,
      "loss": 2.4115495681762695,
      "step": 5827
    },
    {
      "epoch": 0.9403033236527912,
      "grad_norm": 0.9062164425849915,
      "learning_rate": 0.00010269493526687915,
      "loss": 2.837961435317993,
      "step": 5828
    },
    {
      "epoch": 0.9404646660212972,
      "grad_norm": 0.9329729676246643,
      "learning_rate": 0.00010218323817102668,
      "loss": 2.039102554321289,
      "step": 5829
    },
    {
      "epoch": 0.9406260083898031,
      "grad_norm": 1.2571756839752197,
      "learning_rate": 0.0001016724914186204,
      "loss": 2.1201953887939453,
      "step": 5830
    },
    {
      "epoch": 0.9407873507583091,
      "grad_norm": 1.933348298072815,
      "learning_rate": 0.00010116269829335315,
      "loss": 2.3152618408203125,
      "step": 5831
    },
    {
      "epoch": 0.9409486931268151,
      "grad_norm": 2.459836006164551,
      "learning_rate": 0.00010065386207278643,
      "loss": 2.174743413925171,
      "step": 5832
    },
    {
      "epoch": 0.941110035495321,
      "grad_norm": 1.153987169265747,
      "learning_rate": 0.00010014598602832995,
      "loss": 2.1230154037475586,
      "step": 5833
    },
    {
      "epoch": 0.9412713778638271,
      "grad_norm": 0.8744239211082458,
      "learning_rate": 9.96390734252201e-05,
      "loss": 2.2437918186187744,
      "step": 5834
    },
    {
      "epoch": 0.941432720232333,
      "grad_norm": 1.3209692239761353,
      "learning_rate": 9.913312752249903e-05,
      "loss": 2.169578790664673,
      "step": 5835
    },
    {
      "epoch": 0.941594062600839,
      "grad_norm": 1.1939496994018555,
      "learning_rate": 9.862815157299391e-05,
      "loss": 2.2229056358337402,
      "step": 5836
    },
    {
      "epoch": 0.941755404969345,
      "grad_norm": 1.092742681503296,
      "learning_rate": 9.81241488232959e-05,
      "loss": 2.281332015991211,
      "step": 5837
    },
    {
      "epoch": 0.9419167473378509,
      "grad_norm": 0.8754634857177734,
      "learning_rate": 9.762112251373919e-05,
      "loss": 2.4278247356414795,
      "step": 5838
    },
    {
      "epoch": 0.9420780897063569,
      "grad_norm": 1.0347248315811157,
      "learning_rate": 9.711907587838012e-05,
      "loss": 2.2677736282348633,
      "step": 5839
    },
    {
      "epoch": 0.9422394320748628,
      "grad_norm": 1.0634666681289673,
      "learning_rate": 9.661801214497679e-05,
      "loss": 2.353100299835205,
      "step": 5840
    },
    {
      "epoch": 0.9424007744433688,
      "grad_norm": 1.1821062564849854,
      "learning_rate": 9.611793453496793e-05,
      "loss": 2.3408327102661133,
      "step": 5841
    },
    {
      "epoch": 0.9425621168118749,
      "grad_norm": 1.106264591217041,
      "learning_rate": 9.561884626345205e-05,
      "loss": 2.6972453594207764,
      "step": 5842
    },
    {
      "epoch": 0.9427234591803808,
      "grad_norm": 1.3261256217956543,
      "learning_rate": 9.512075053916735e-05,
      "loss": 2.655799150466919,
      "step": 5843
    },
    {
      "epoch": 0.9428848015488868,
      "grad_norm": 1.4740550518035889,
      "learning_rate": 9.462365056447064e-05,
      "loss": 2.2422738075256348,
      "step": 5844
    },
    {
      "epoch": 0.9430461439173927,
      "grad_norm": 0.9823197722434998,
      "learning_rate": 9.412754953531663e-05,
      "loss": 2.0172715187072754,
      "step": 5845
    },
    {
      "epoch": 0.9432074862858987,
      "grad_norm": 1.354600191116333,
      "learning_rate": 9.363245064123791e-05,
      "loss": 2.5491676330566406,
      "step": 5846
    },
    {
      "epoch": 0.9433688286544046,
      "grad_norm": 0.9982588887214661,
      "learning_rate": 9.313835706532411e-05,
      "loss": 2.3420650959014893,
      "step": 5847
    },
    {
      "epoch": 0.9435301710229106,
      "grad_norm": 0.9443960785865784,
      "learning_rate": 9.264527198420117e-05,
      "loss": 2.3396010398864746,
      "step": 5848
    },
    {
      "epoch": 0.9436915133914165,
      "grad_norm": 0.9450830221176147,
      "learning_rate": 9.215319856801157e-05,
      "loss": 2.2621781826019287,
      "step": 5849
    },
    {
      "epoch": 0.9438528557599225,
      "grad_norm": 0.9682110548019409,
      "learning_rate": 9.166213998039352e-05,
      "loss": 2.4163897037506104,
      "step": 5850
    },
    {
      "epoch": 0.9440141981284286,
      "grad_norm": 0.9580631256103516,
      "learning_rate": 9.117209937846053e-05,
      "loss": 2.3014895915985107,
      "step": 5851
    },
    {
      "epoch": 0.9441755404969345,
      "grad_norm": 0.8732711672782898,
      "learning_rate": 9.068307991278127e-05,
      "loss": 2.3414931297302246,
      "step": 5852
    },
    {
      "epoch": 0.9443368828654405,
      "grad_norm": 0.8579061031341553,
      "learning_rate": 9.019508472735957e-05,
      "loss": 2.208566665649414,
      "step": 5853
    },
    {
      "epoch": 0.9444982252339464,
      "grad_norm": 0.8005319237709045,
      "learning_rate": 8.970811695961387e-05,
      "loss": 2.534245491027832,
      "step": 5854
    },
    {
      "epoch": 0.9446595676024524,
      "grad_norm": 1.292481541633606,
      "learning_rate": 8.922217974035692e-05,
      "loss": 2.070258140563965,
      "step": 5855
    },
    {
      "epoch": 0.9448209099709584,
      "grad_norm": 1.4740824699401855,
      "learning_rate": 8.873727619377612e-05,
      "loss": 2.387894630432129,
      "step": 5856
    },
    {
      "epoch": 0.9449822523394643,
      "grad_norm": 0.9970455765724182,
      "learning_rate": 8.825340943741314e-05,
      "loss": 2.3391332626342773,
      "step": 5857
    },
    {
      "epoch": 0.9451435947079703,
      "grad_norm": 1.0826884508132935,
      "learning_rate": 8.777058258214376e-05,
      "loss": 2.5204410552978516,
      "step": 5858
    },
    {
      "epoch": 0.9453049370764763,
      "grad_norm": 0.7675654888153076,
      "learning_rate": 8.728879873215817e-05,
      "loss": 2.5895681381225586,
      "step": 5859
    },
    {
      "epoch": 0.9454662794449823,
      "grad_norm": 1.0693928003311157,
      "learning_rate": 8.680806098494088e-05,
      "loss": 2.5546984672546387,
      "step": 5860
    },
    {
      "epoch": 0.9456276218134883,
      "grad_norm": 1.1542857885360718,
      "learning_rate": 8.632837243125083e-05,
      "loss": 2.2957100868225098,
      "step": 5861
    },
    {
      "epoch": 0.9457889641819942,
      "grad_norm": 1.6219310760498047,
      "learning_rate": 8.584973615510124e-05,
      "loss": 2.167665481567383,
      "step": 5862
    },
    {
      "epoch": 0.9459503065505002,
      "grad_norm": 1.4134613275527954,
      "learning_rate": 8.537215523374037e-05,
      "loss": 2.8882269859313965,
      "step": 5863
    },
    {
      "epoch": 0.9461116489190061,
      "grad_norm": 0.8664407730102539,
      "learning_rate": 8.489563273763115e-05,
      "loss": 2.230116844177246,
      "step": 5864
    },
    {
      "epoch": 0.9462729912875121,
      "grad_norm": 1.0880736112594604,
      "learning_rate": 8.442017173043165e-05,
      "loss": 2.287720203399658,
      "step": 5865
    },
    {
      "epoch": 0.946434333656018,
      "grad_norm": 0.7907834053039551,
      "learning_rate": 8.394577526897565e-05,
      "loss": 2.26448392868042,
      "step": 5866
    },
    {
      "epoch": 0.946595676024524,
      "grad_norm": 1.257526159286499,
      "learning_rate": 8.347244640325258e-05,
      "loss": 2.1731252670288086,
      "step": 5867
    },
    {
      "epoch": 0.9467570183930301,
      "grad_norm": 1.245977520942688,
      "learning_rate": 8.300018817638825e-05,
      "loss": 2.511991024017334,
      "step": 5868
    },
    {
      "epoch": 0.946918360761536,
      "grad_norm": 0.9148598313331604,
      "learning_rate": 8.252900362462479e-05,
      "loss": 1.9371206760406494,
      "step": 5869
    },
    {
      "epoch": 0.947079703130042,
      "grad_norm": 0.968131422996521,
      "learning_rate": 8.20588957773018e-05,
      "loss": 2.078866720199585,
      "step": 5870
    },
    {
      "epoch": 0.9472410454985479,
      "grad_norm": 0.9166329503059387,
      "learning_rate": 8.158986765683646e-05,
      "loss": 2.9426612854003906,
      "step": 5871
    },
    {
      "epoch": 0.9474023878670539,
      "grad_norm": 1.004321813583374,
      "learning_rate": 8.112192227870394e-05,
      "loss": 1.8345093727111816,
      "step": 5872
    },
    {
      "epoch": 0.9475637302355598,
      "grad_norm": 1.0387128591537476,
      "learning_rate": 8.065506265141851e-05,
      "loss": 2.435163736343384,
      "step": 5873
    },
    {
      "epoch": 0.9477250726040658,
      "grad_norm": 1.5018932819366455,
      "learning_rate": 8.018929177651393e-05,
      "loss": 2.2763595581054688,
      "step": 5874
    },
    {
      "epoch": 0.9478864149725718,
      "grad_norm": 1.0632628202438354,
      "learning_rate": 7.972461264852387e-05,
      "loss": 2.009861469268799,
      "step": 5875
    },
    {
      "epoch": 0.9480477573410777,
      "grad_norm": 0.8474001884460449,
      "learning_rate": 7.926102825496323e-05,
      "loss": 2.2501087188720703,
      "step": 5876
    },
    {
      "epoch": 0.9482090997095838,
      "grad_norm": 1.0489658117294312,
      "learning_rate": 7.87985415763086e-05,
      "loss": 2.0585484504699707,
      "step": 5877
    },
    {
      "epoch": 0.9483704420780897,
      "grad_norm": 1.362769603729248,
      "learning_rate": 7.833715558597907e-05,
      "loss": 2.4235386848449707,
      "step": 5878
    },
    {
      "epoch": 0.9485317844465957,
      "grad_norm": 1.0463582277297974,
      "learning_rate": 7.787687325031703e-05,
      "loss": 2.262101411819458,
      "step": 5879
    },
    {
      "epoch": 0.9486931268151017,
      "grad_norm": 0.804507851600647,
      "learning_rate": 7.741769752856958e-05,
      "loss": 2.028214931488037,
      "step": 5880
    },
    {
      "epoch": 0.9488544691836076,
      "grad_norm": 0.9125126600265503,
      "learning_rate": 7.69596313728691e-05,
      "loss": 2.7057113647460938,
      "step": 5881
    },
    {
      "epoch": 0.9490158115521136,
      "grad_norm": 1.3032963275909424,
      "learning_rate": 7.650267772821409e-05,
      "loss": 1.9977740049362183,
      "step": 5882
    },
    {
      "epoch": 0.9491771539206195,
      "grad_norm": 1.1532365083694458,
      "learning_rate": 7.604683953245076e-05,
      "loss": 2.1786892414093018,
      "step": 5883
    },
    {
      "epoch": 0.9493384962891255,
      "grad_norm": 1.6059006452560425,
      "learning_rate": 7.559211971625385e-05,
      "loss": 2.2992544174194336,
      "step": 5884
    },
    {
      "epoch": 0.9494998386576315,
      "grad_norm": 1.0508393049240112,
      "learning_rate": 7.513852120310766e-05,
      "loss": 2.4279632568359375,
      "step": 5885
    },
    {
      "epoch": 0.9496611810261375,
      "grad_norm": 1.2950432300567627,
      "learning_rate": 7.468604690928754e-05,
      "loss": 1.9945350885391235,
      "step": 5886
    },
    {
      "epoch": 0.9498225233946435,
      "grad_norm": 0.7711794376373291,
      "learning_rate": 7.423469974384108e-05,
      "loss": 2.098043203353882,
      "step": 5887
    },
    {
      "epoch": 0.9499838657631494,
      "grad_norm": 0.9570376873016357,
      "learning_rate": 7.37844826085691e-05,
      "loss": 2.3795061111450195,
      "step": 5888
    },
    {
      "epoch": 0.9501452081316554,
      "grad_norm": 1.2817223072052002,
      "learning_rate": 7.333539839800738e-05,
      "loss": 2.0105366706848145,
      "step": 5889
    },
    {
      "epoch": 0.9503065505001613,
      "grad_norm": 1.0626938343048096,
      "learning_rate": 7.2887449999408e-05,
      "loss": 2.347933292388916,
      "step": 5890
    },
    {
      "epoch": 0.9504678928686673,
      "grad_norm": 1.0555347204208374,
      "learning_rate": 7.244064029272049e-05,
      "loss": 2.1869304180145264,
      "step": 5891
    },
    {
      "epoch": 0.9506292352371732,
      "grad_norm": 0.9768400192260742,
      "learning_rate": 7.199497215057352e-05,
      "loss": 2.312633514404297,
      "step": 5892
    },
    {
      "epoch": 0.9507905776056792,
      "grad_norm": 0.8752104043960571,
      "learning_rate": 7.15504484382565e-05,
      "loss": 2.195976734161377,
      "step": 5893
    },
    {
      "epoch": 0.9509519199741853,
      "grad_norm": 0.9778468608856201,
      "learning_rate": 7.110707201370115e-05,
      "loss": 2.327019691467285,
      "step": 5894
    },
    {
      "epoch": 0.9511132623426912,
      "grad_norm": 1.0217647552490234,
      "learning_rate": 7.066484572746279e-05,
      "loss": 2.6351184844970703,
      "step": 5895
    },
    {
      "epoch": 0.9512746047111972,
      "grad_norm": 1.126513957977295,
      "learning_rate": 7.022377242270251e-05,
      "loss": 2.718902826309204,
      "step": 5896
    },
    {
      "epoch": 0.9514359470797031,
      "grad_norm": 0.8994089365005493,
      "learning_rate": 6.978385493516859e-05,
      "loss": 2.4712977409362793,
      "step": 5897
    },
    {
      "epoch": 0.9515972894482091,
      "grad_norm": 1.0373797416687012,
      "learning_rate": 6.934509609317821e-05,
      "loss": 2.5214877128601074,
      "step": 5898
    },
    {
      "epoch": 0.951758631816715,
      "grad_norm": 1.1131243705749512,
      "learning_rate": 6.890749871759955e-05,
      "loss": 2.1832966804504395,
      "step": 5899
    },
    {
      "epoch": 0.951919974185221,
      "grad_norm": 1.0710166692733765,
      "learning_rate": 6.847106562183348e-05,
      "loss": 2.324366569519043,
      "step": 5900
    },
    {
      "epoch": 0.952081316553727,
      "grad_norm": 1.0932297706604004,
      "learning_rate": 6.80357996117953e-05,
      "loss": 2.3756935596466064,
      "step": 5901
    },
    {
      "epoch": 0.952242658922233,
      "grad_norm": 0.8282956480979919,
      "learning_rate": 6.760170348589704e-05,
      "loss": 2.515221118927002,
      "step": 5902
    },
    {
      "epoch": 0.952404001290739,
      "grad_norm": 0.7843517661094666,
      "learning_rate": 6.716878003502938e-05,
      "loss": 2.0666756629943848,
      "step": 5903
    },
    {
      "epoch": 0.952565343659245,
      "grad_norm": 1.1641261577606201,
      "learning_rate": 6.673703204254347e-05,
      "loss": 3.0505738258361816,
      "step": 5904
    },
    {
      "epoch": 0.9527266860277509,
      "grad_norm": 0.783150315284729,
      "learning_rate": 6.630646228423323e-05,
      "loss": 2.2554171085357666,
      "step": 5905
    },
    {
      "epoch": 0.9528880283962569,
      "grad_norm": 0.9690251350402832,
      "learning_rate": 6.587707352831754e-05,
      "loss": 2.51387095451355,
      "step": 5906
    },
    {
      "epoch": 0.9530493707647628,
      "grad_norm": 1.2202883958816528,
      "learning_rate": 6.544886853542246e-05,
      "loss": 2.1865713596343994,
      "step": 5907
    },
    {
      "epoch": 0.9532107131332688,
      "grad_norm": 1.155492901802063,
      "learning_rate": 6.502185005856312e-05,
      "loss": 2.175304651260376,
      "step": 5908
    },
    {
      "epoch": 0.9533720555017747,
      "grad_norm": 0.9066641926765442,
      "learning_rate": 6.45960208431266e-05,
      "loss": 2.215163230895996,
      "step": 5909
    },
    {
      "epoch": 0.9535333978702807,
      "grad_norm": 1.1214165687561035,
      "learning_rate": 6.417138362685394e-05,
      "loss": 2.075376033782959,
      "step": 5910
    },
    {
      "epoch": 0.9536947402387868,
      "grad_norm": 0.8012157678604126,
      "learning_rate": 6.374794113982232e-05,
      "loss": 2.0864927768707275,
      "step": 5911
    },
    {
      "epoch": 0.9538560826072927,
      "grad_norm": 1.1739611625671387,
      "learning_rate": 6.332569610442807e-05,
      "loss": 1.888679027557373,
      "step": 5912
    },
    {
      "epoch": 0.9540174249757987,
      "grad_norm": 1.1869603395462036,
      "learning_rate": 6.290465123536877e-05,
      "loss": 2.3067784309387207,
      "step": 5913
    },
    {
      "epoch": 0.9541787673443046,
      "grad_norm": 0.7357597947120667,
      "learning_rate": 6.248480923962577e-05,
      "loss": 2.0443315505981445,
      "step": 5914
    },
    {
      "epoch": 0.9543401097128106,
      "grad_norm": 1.0921027660369873,
      "learning_rate": 6.206617281644702e-05,
      "loss": 2.0958189964294434,
      "step": 5915
    },
    {
      "epoch": 0.9545014520813165,
      "grad_norm": 0.9545575380325317,
      "learning_rate": 6.164874465732973e-05,
      "loss": 2.631807327270508,
      "step": 5916
    },
    {
      "epoch": 0.9546627944498225,
      "grad_norm": 0.764599621295929,
      "learning_rate": 6.123252744600266e-05,
      "loss": 2.0249600410461426,
      "step": 5917
    },
    {
      "epoch": 0.9548241368183285,
      "grad_norm": 1.3433769941329956,
      "learning_rate": 6.0817523858409245e-05,
      "loss": 2.3848884105682373,
      "step": 5918
    },
    {
      "epoch": 0.9549854791868344,
      "grad_norm": 1.3006800413131714,
      "learning_rate": 6.040373656269041e-05,
      "loss": 1.856907844543457,
      "step": 5919
    },
    {
      "epoch": 0.9551468215553405,
      "grad_norm": 0.8924742937088013,
      "learning_rate": 5.999116821916728e-05,
      "loss": 2.222712993621826,
      "step": 5920
    },
    {
      "epoch": 0.9553081639238464,
      "grad_norm": 0.925900399684906,
      "learning_rate": 5.957982148032387e-05,
      "loss": 1.7850298881530762,
      "step": 5921
    },
    {
      "epoch": 0.9554695062923524,
      "grad_norm": 0.8628061413764954,
      "learning_rate": 5.916969899079047e-05,
      "loss": 2.6069183349609375,
      "step": 5922
    },
    {
      "epoch": 0.9556308486608583,
      "grad_norm": 1.2662640810012817,
      "learning_rate": 5.8760803387326425e-05,
      "loss": 1.8482184410095215,
      "step": 5923
    },
    {
      "epoch": 0.9557921910293643,
      "grad_norm": 1.1593716144561768,
      "learning_rate": 5.8353137298803174e-05,
      "loss": 2.5276808738708496,
      "step": 5924
    },
    {
      "epoch": 0.9559535333978703,
      "grad_norm": 0.8375536799430847,
      "learning_rate": 5.794670334618715e-05,
      "loss": 2.053475856781006,
      "step": 5925
    },
    {
      "epoch": 0.9561148757663762,
      "grad_norm": 0.920303463935852,
      "learning_rate": 5.7541504142523406e-05,
      "loss": 2.2046773433685303,
      "step": 5926
    },
    {
      "epoch": 0.9562762181348822,
      "grad_norm": 1.1993811130523682,
      "learning_rate": 5.713754229291851e-05,
      "loss": 1.7521297931671143,
      "step": 5927
    },
    {
      "epoch": 0.9564375605033882,
      "grad_norm": 1.080795168876648,
      "learning_rate": 5.6734820394523626e-05,
      "loss": 2.195782423019409,
      "step": 5928
    },
    {
      "epoch": 0.9565989028718942,
      "grad_norm": 0.968019962310791,
      "learning_rate": 5.63333410365183e-05,
      "loss": 1.9476864337921143,
      "step": 5929
    },
    {
      "epoch": 0.9567602452404002,
      "grad_norm": 1.916200876235962,
      "learning_rate": 5.593310680009328e-05,
      "loss": 2.169649124145508,
      "step": 5930
    },
    {
      "epoch": 0.9569215876089061,
      "grad_norm": 1.1338576078414917,
      "learning_rate": 5.553412025843444e-05,
      "loss": 2.0924160480499268,
      "step": 5931
    },
    {
      "epoch": 0.9570829299774121,
      "grad_norm": 1.1434519290924072,
      "learning_rate": 5.513638397670567e-05,
      "loss": 2.3193655014038086,
      "step": 5932
    },
    {
      "epoch": 0.957244272345918,
      "grad_norm": 0.9276216626167297,
      "learning_rate": 5.473990051203298e-05,
      "loss": 2.3918650150299072,
      "step": 5933
    },
    {
      "epoch": 0.957405614714424,
      "grad_norm": 0.9209693074226379,
      "learning_rate": 5.434467241348773e-05,
      "loss": 2.18891978263855,
      "step": 5934
    },
    {
      "epoch": 0.9575669570829299,
      "grad_norm": 1.0139374732971191,
      "learning_rate": 5.395070222206999e-05,
      "loss": 1.9244165420532227,
      "step": 5935
    },
    {
      "epoch": 0.9577282994514359,
      "grad_norm": 1.3191357851028442,
      "learning_rate": 5.3557992470692817e-05,
      "loss": 1.8978378772735596,
      "step": 5936
    },
    {
      "epoch": 0.957889641819942,
      "grad_norm": 0.9570164680480957,
      "learning_rate": 5.31665456841656e-05,
      "loss": 2.667968273162842,
      "step": 5937
    },
    {
      "epoch": 0.9580509841884479,
      "grad_norm": 1.0322397947311401,
      "learning_rate": 5.277636437917768e-05,
      "loss": 2.246629238128662,
      "step": 5938
    },
    {
      "epoch": 0.9582123265569539,
      "grad_norm": 1.7633867263793945,
      "learning_rate": 5.2387451064282556e-05,
      "loss": 2.1514196395874023,
      "step": 5939
    },
    {
      "epoch": 0.9583736689254598,
      "grad_norm": 0.8518823981285095,
      "learning_rate": 5.1999808239881565e-05,
      "loss": 2.3791494369506836,
      "step": 5940
    },
    {
      "epoch": 0.9585350112939658,
      "grad_norm": 0.8306836485862732,
      "learning_rate": 5.161343839820762e-05,
      "loss": 2.1602354049682617,
      "step": 5941
    },
    {
      "epoch": 0.9586963536624717,
      "grad_norm": 1.104798436164856,
      "learning_rate": 5.122834402330964e-05,
      "loss": 2.188157796859741,
      "step": 5942
    },
    {
      "epoch": 0.9588576960309777,
      "grad_norm": 1.0390794277191162,
      "learning_rate": 5.084452759103603e-05,
      "loss": 2.0261383056640625,
      "step": 5943
    },
    {
      "epoch": 0.9590190383994837,
      "grad_norm": 0.9608902931213379,
      "learning_rate": 5.046199156901931e-05,
      "loss": 2.3556692600250244,
      "step": 5944
    },
    {
      "epoch": 0.9591803807679897,
      "grad_norm": 0.8813768625259399,
      "learning_rate": 5.0080738416659776e-05,
      "loss": 2.3392271995544434,
      "step": 5945
    },
    {
      "epoch": 0.9593417231364957,
      "grad_norm": 0.834064245223999,
      "learning_rate": 4.970077058511002e-05,
      "loss": 2.327162742614746,
      "step": 5946
    },
    {
      "epoch": 0.9595030655050016,
      "grad_norm": 0.8225624561309814,
      "learning_rate": 4.932209051725914e-05,
      "loss": 2.146210193634033,
      "step": 5947
    },
    {
      "epoch": 0.9596644078735076,
      "grad_norm": 0.7610519528388977,
      "learning_rate": 4.8944700647716614e-05,
      "loss": 2.173802137374878,
      "step": 5948
    },
    {
      "epoch": 0.9598257502420136,
      "grad_norm": 1.061968207359314,
      "learning_rate": 4.856860340279731e-05,
      "loss": 2.047680377960205,
      "step": 5949
    },
    {
      "epoch": 0.9599870926105195,
      "grad_norm": 1.2544047832489014,
      "learning_rate": 4.819380120050551e-05,
      "loss": 2.419529438018799,
      "step": 5950
    },
    {
      "epoch": 0.9601484349790255,
      "grad_norm": 1.004615306854248,
      "learning_rate": 4.782029645051916e-05,
      "loss": 2.076169013977051,
      "step": 5951
    },
    {
      "epoch": 0.9603097773475314,
      "grad_norm": 1.2832516431808472,
      "learning_rate": 4.744809155417487e-05,
      "loss": 2.135134696960449,
      "step": 5952
    },
    {
      "epoch": 0.9604711197160374,
      "grad_norm": 0.9916216731071472,
      "learning_rate": 4.707718890445226e-05,
      "loss": 2.4105615615844727,
      "step": 5953
    },
    {
      "epoch": 0.9606324620845434,
      "grad_norm": 0.7511906623840332,
      "learning_rate": 4.67075908859583e-05,
      "loss": 2.500070571899414,
      "step": 5954
    },
    {
      "epoch": 0.9607938044530494,
      "grad_norm": 0.7264192700386047,
      "learning_rate": 4.633929987491245e-05,
      "loss": 2.049577236175537,
      "step": 5955
    },
    {
      "epoch": 0.9609551468215554,
      "grad_norm": 1.1141055822372437,
      "learning_rate": 4.597231823913112e-05,
      "loss": 2.126883029937744,
      "step": 5956
    },
    {
      "epoch": 0.9611164891900613,
      "grad_norm": 0.8749083876609802,
      "learning_rate": 4.560664833801237e-05,
      "loss": 2.195160388946533,
      "step": 5957
    },
    {
      "epoch": 0.9612778315585673,
      "grad_norm": 0.903946042060852,
      "learning_rate": 4.524229252252091e-05,
      "loss": 2.4152777194976807,
      "step": 5958
    },
    {
      "epoch": 0.9614391739270732,
      "grad_norm": 0.9539807438850403,
      "learning_rate": 4.487925313517302e-05,
      "loss": 2.3006863594055176,
      "step": 5959
    },
    {
      "epoch": 0.9616005162955792,
      "grad_norm": 0.8687537312507629,
      "learning_rate": 4.4517532510021356e-05,
      "loss": 1.9652690887451172,
      "step": 5960
    },
    {
      "epoch": 0.9617618586640851,
      "grad_norm": 1.3254128694534302,
      "learning_rate": 4.4157132972639875e-05,
      "loss": 2.6896767616271973,
      "step": 5961
    },
    {
      "epoch": 0.9619232010325912,
      "grad_norm": 1.086984634399414,
      "learning_rate": 4.3798056840109154e-05,
      "loss": 2.687197685241699,
      "step": 5962
    },
    {
      "epoch": 0.9620845434010972,
      "grad_norm": 1.127128005027771,
      "learning_rate": 4.344030642100133e-05,
      "loss": 2.1340253353118896,
      "step": 5963
    },
    {
      "epoch": 0.9622458857696031,
      "grad_norm": 0.8875247240066528,
      "learning_rate": 4.308388401536506e-05,
      "loss": 2.409306049346924,
      "step": 5964
    },
    {
      "epoch": 0.9624072281381091,
      "grad_norm": 0.8242495059967041,
      "learning_rate": 4.2728791914711156e-05,
      "loss": 1.9652667045593262,
      "step": 5965
    },
    {
      "epoch": 0.962568570506615,
      "grad_norm": 1.345842957496643,
      "learning_rate": 4.2375032401997636e-05,
      "loss": 2.3139915466308594,
      "step": 5966
    },
    {
      "epoch": 0.962729912875121,
      "grad_norm": 0.709181547164917,
      "learning_rate": 4.2022607751614874e-05,
      "loss": 1.9282453060150146,
      "step": 5967
    },
    {
      "epoch": 0.962891255243627,
      "grad_norm": 1.0521901845932007,
      "learning_rate": 4.1671520229371234e-05,
      "loss": 2.388934373855591,
      "step": 5968
    },
    {
      "epoch": 0.9630525976121329,
      "grad_norm": 0.9293901920318604,
      "learning_rate": 4.1321772092478533e-05,
      "loss": 2.3292953968048096,
      "step": 5969
    },
    {
      "epoch": 0.9632139399806389,
      "grad_norm": 1.1587107181549072,
      "learning_rate": 4.097336558953726e-05,
      "loss": 2.363431930541992,
      "step": 5970
    },
    {
      "epoch": 0.9633752823491449,
      "grad_norm": 1.157167911529541,
      "learning_rate": 4.062630296052222e-05,
      "loss": 2.141214370727539,
      "step": 5971
    },
    {
      "epoch": 0.9635366247176509,
      "grad_norm": 0.979958713054657,
      "learning_rate": 4.028058643676835e-05,
      "loss": 2.1154959201812744,
      "step": 5972
    },
    {
      "epoch": 0.9636979670861568,
      "grad_norm": 1.2411192655563354,
      "learning_rate": 3.993621824095622e-05,
      "loss": 2.6492061614990234,
      "step": 5973
    },
    {
      "epoch": 0.9638593094546628,
      "grad_norm": 0.8723487854003906,
      "learning_rate": 3.9593200587097535e-05,
      "loss": 2.3052077293395996,
      "step": 5974
    },
    {
      "epoch": 0.9640206518231688,
      "grad_norm": 1.2757697105407715,
      "learning_rate": 3.925153568052123e-05,
      "loss": 2.2843685150146484,
      "step": 5975
    },
    {
      "epoch": 0.9641819941916747,
      "grad_norm": 0.9626977443695068,
      "learning_rate": 3.891122571785924e-05,
      "loss": 2.07535719871521,
      "step": 5976
    },
    {
      "epoch": 0.9643433365601807,
      "grad_norm": 0.9485663771629333,
      "learning_rate": 3.857227288703205e-05,
      "loss": 2.238638401031494,
      "step": 5977
    },
    {
      "epoch": 0.9645046789286866,
      "grad_norm": 1.1879147291183472,
      "learning_rate": 3.8234679367235096e-05,
      "loss": 2.268632650375366,
      "step": 5978
    },
    {
      "epoch": 0.9646660212971926,
      "grad_norm": 1.6962645053863525,
      "learning_rate": 3.789844732892445e-05,
      "loss": 2.114011287689209,
      "step": 5979
    },
    {
      "epoch": 0.9648273636656987,
      "grad_norm": 0.9601020812988281,
      "learning_rate": 3.756357893380302e-05,
      "loss": 2.3047285079956055,
      "step": 5980
    },
    {
      "epoch": 0.9649887060342046,
      "grad_norm": 0.8221027851104736,
      "learning_rate": 3.7230076334806376e-05,
      "loss": 2.028743267059326,
      "step": 5981
    },
    {
      "epoch": 0.9651500484027106,
      "grad_norm": 0.8103053569793701,
      "learning_rate": 3.689794167608937e-05,
      "loss": 2.222897529602051,
      "step": 5982
    },
    {
      "epoch": 0.9653113907712165,
      "grad_norm": 0.8287355899810791,
      "learning_rate": 3.656717709301194e-05,
      "loss": 2.2560906410217285,
      "step": 5983
    },
    {
      "epoch": 0.9654727331397225,
      "grad_norm": 0.7861142158508301,
      "learning_rate": 3.6237784712125464e-05,
      "loss": 2.125624179840088,
      "step": 5984
    },
    {
      "epoch": 0.9656340755082284,
      "grad_norm": 1.1747512817382812,
      "learning_rate": 3.5909766651159384e-05,
      "loss": 2.249483585357666,
      "step": 5985
    },
    {
      "epoch": 0.9657954178767344,
      "grad_norm": 0.8091188669204712,
      "learning_rate": 3.558312501900718e-05,
      "loss": 1.9903709888458252,
      "step": 5986
    },
    {
      "epoch": 0.9659567602452404,
      "grad_norm": 0.9120203256607056,
      "learning_rate": 3.525786191571314e-05,
      "loss": 1.954100489616394,
      "step": 5987
    },
    {
      "epoch": 0.9661181026137464,
      "grad_norm": 0.8058664798736572,
      "learning_rate": 3.4933979432458516e-05,
      "loss": 2.5025885105133057,
      "step": 5988
    },
    {
      "epoch": 0.9662794449822524,
      "grad_norm": 0.8272390365600586,
      "learning_rate": 3.461147965154846e-05,
      "loss": 2.0019466876983643,
      "step": 5989
    },
    {
      "epoch": 0.9664407873507583,
      "grad_norm": 0.9787078499794006,
      "learning_rate": 3.4290364646398456e-05,
      "loss": 2.000649929046631,
      "step": 5990
    },
    {
      "epoch": 0.9666021297192643,
      "grad_norm": 1.2863887548446655,
      "learning_rate": 3.397063648152077e-05,
      "loss": 2.2835521697998047,
      "step": 5991
    },
    {
      "epoch": 0.9667634720877702,
      "grad_norm": 0.9254443049430847,
      "learning_rate": 3.3652297212511644e-05,
      "loss": 2.5589170455932617,
      "step": 5992
    },
    {
      "epoch": 0.9669248144562762,
      "grad_norm": 1.0107896327972412,
      "learning_rate": 3.3335348886037815e-05,
      "loss": 2.2842140197753906,
      "step": 5993
    },
    {
      "epoch": 0.9670861568247822,
      "grad_norm": 0.9892925024032593,
      "learning_rate": 3.301979353982312e-05,
      "loss": 2.9305033683776855,
      "step": 5994
    },
    {
      "epoch": 0.9672474991932881,
      "grad_norm": 1.2683720588684082,
      "learning_rate": 3.2705633202635974e-05,
      "loss": 2.1531901359558105,
      "step": 5995
    },
    {
      "epoch": 0.9674088415617941,
      "grad_norm": 0.8013840913772583,
      "learning_rate": 3.239286989427573e-05,
      "loss": 1.73250150680542,
      "step": 5996
    },
    {
      "epoch": 0.9675701839303001,
      "grad_norm": 0.9264519214630127,
      "learning_rate": 3.208150562556017e-05,
      "loss": 2.0726735591888428,
      "step": 5997
    },
    {
      "epoch": 0.9677315262988061,
      "grad_norm": 1.0437986850738525,
      "learning_rate": 3.177154239831223e-05,
      "loss": 1.7007440328598022,
      "step": 5998
    },
    {
      "epoch": 0.9678928686673121,
      "grad_norm": 1.1392425298690796,
      "learning_rate": 3.1462982205347327e-05,
      "loss": 2.4075722694396973,
      "step": 5999
    },
    {
      "epoch": 0.968054211035818,
      "grad_norm": 1.440717339515686,
      "learning_rate": 3.115582703046055e-05,
      "loss": 2.3516759872436523,
      "step": 6000
    },
    {
      "epoch": 0.968215553404324,
      "grad_norm": 0.9685876965522766,
      "learning_rate": 3.0850078848413704e-05,
      "loss": 2.031785011291504,
      "step": 6001
    },
    {
      "epoch": 0.9683768957728299,
      "grad_norm": 1.1450544595718384,
      "learning_rate": 3.054573962492288e-05,
      "loss": 1.8966705799102783,
      "step": 6002
    },
    {
      "epoch": 0.9685382381413359,
      "grad_norm": 0.8960651755332947,
      "learning_rate": 3.024281131664569e-05,
      "loss": 1.822111964225769,
      "step": 6003
    },
    {
      "epoch": 0.9686995805098418,
      "grad_norm": 0.9561957716941833,
      "learning_rate": 2.9941295871168562e-05,
      "loss": 2.1708078384399414,
      "step": 6004
    },
    {
      "epoch": 0.9688609228783479,
      "grad_norm": 1.2167119979858398,
      "learning_rate": 2.964119522699449e-05,
      "loss": 2.147610902786255,
      "step": 6005
    },
    {
      "epoch": 0.9690222652468539,
      "grad_norm": 0.9208367466926575,
      "learning_rate": 2.9342511313530413e-05,
      "loss": 2.072706699371338,
      "step": 6006
    },
    {
      "epoch": 0.9691836076153598,
      "grad_norm": 1.2642698287963867,
      "learning_rate": 2.9045246051074698e-05,
      "loss": 2.488372325897217,
      "step": 6007
    },
    {
      "epoch": 0.9693449499838658,
      "grad_norm": 0.970229983329773,
      "learning_rate": 2.8749401350805115e-05,
      "loss": 2.049020767211914,
      "step": 6008
    },
    {
      "epoch": 0.9695062923523717,
      "grad_norm": 1.034257173538208,
      "learning_rate": 2.8454979114766134e-05,
      "loss": 2.509824275970459,
      "step": 6009
    },
    {
      "epoch": 0.9696676347208777,
      "grad_norm": 0.8426554203033447,
      "learning_rate": 2.816198123585714e-05,
      "loss": 1.8803200721740723,
      "step": 6010
    },
    {
      "epoch": 0.9698289770893836,
      "grad_norm": 1.0767226219177246,
      "learning_rate": 2.7870409597819864e-05,
      "loss": 2.73358154296875,
      "step": 6011
    },
    {
      "epoch": 0.9699903194578896,
      "grad_norm": 1.7364604473114014,
      "learning_rate": 2.758026607522657e-05,
      "loss": 2.0516223907470703,
      "step": 6012
    },
    {
      "epoch": 0.9701516618263956,
      "grad_norm": 0.9173264503479004,
      "learning_rate": 2.7291552533467852e-05,
      "loss": 2.251721143722534,
      "step": 6013
    },
    {
      "epoch": 0.9703130041949016,
      "grad_norm": 1.073049783706665,
      "learning_rate": 2.700427082874063e-05,
      "loss": 2.4627726078033447,
      "step": 6014
    },
    {
      "epoch": 0.9704743465634076,
      "grad_norm": 1.3446921110153198,
      "learning_rate": 2.671842280803624e-05,
      "loss": 2.145637035369873,
      "step": 6015
    },
    {
      "epoch": 0.9706356889319135,
      "grad_norm": 1.0672063827514648,
      "learning_rate": 2.643401030912876e-05,
      "loss": 2.375817060470581,
      "step": 6016
    },
    {
      "epoch": 0.9707970313004195,
      "grad_norm": 1.5550585985183716,
      "learning_rate": 2.615103516056275e-05,
      "loss": 2.2677431106567383,
      "step": 6017
    },
    {
      "epoch": 0.9709583736689255,
      "grad_norm": 0.9091049432754517,
      "learning_rate": 2.5869499181641913e-05,
      "loss": 2.4013800621032715,
      "step": 6018
    },
    {
      "epoch": 0.9711197160374314,
      "grad_norm": 1.014593482017517,
      "learning_rate": 2.558940418241726e-05,
      "loss": 2.491049289703369,
      "step": 6019
    },
    {
      "epoch": 0.9712810584059374,
      "grad_norm": 0.7554174661636353,
      "learning_rate": 2.5310751963675355e-05,
      "loss": 2.331300735473633,
      "step": 6020
    },
    {
      "epoch": 0.9714424007744433,
      "grad_norm": 0.9715957045555115,
      "learning_rate": 2.503354431692689e-05,
      "loss": 2.451331615447998,
      "step": 6021
    },
    {
      "epoch": 0.9716037431429493,
      "grad_norm": 0.9662553071975708,
      "learning_rate": 2.4757783024395242e-05,
      "loss": 2.1850008964538574,
      "step": 6022
    },
    {
      "epoch": 0.9717650855114554,
      "grad_norm": 0.7867013812065125,
      "learning_rate": 2.4483469859004625e-05,
      "loss": 1.9174530506134033,
      "step": 6023
    },
    {
      "epoch": 0.9719264278799613,
      "grad_norm": 0.9936076402664185,
      "learning_rate": 2.4210606584369106e-05,
      "loss": 2.002171516418457,
      "step": 6024
    },
    {
      "epoch": 0.9720877702484673,
      "grad_norm": 0.9484394788742065,
      "learning_rate": 2.3939194954781142e-05,
      "loss": 2.457183837890625,
      "step": 6025
    },
    {
      "epoch": 0.9722491126169732,
      "grad_norm": 1.4192125797271729,
      "learning_rate": 2.3669236715200298e-05,
      "loss": 2.068779945373535,
      "step": 6026
    },
    {
      "epoch": 0.9724104549854792,
      "grad_norm": 1.167372465133667,
      "learning_rate": 2.3400733601241812e-05,
      "loss": 2.2650527954101562,
      "step": 6027
    },
    {
      "epoch": 0.9725717973539851,
      "grad_norm": 0.7478707432746887,
      "learning_rate": 2.313368733916585e-05,
      "loss": 1.908159613609314,
      "step": 6028
    },
    {
      "epoch": 0.9727331397224911,
      "grad_norm": 0.7986984252929688,
      "learning_rate": 2.2868099645866088e-05,
      "loss": 2.2257888317108154,
      "step": 6029
    },
    {
      "epoch": 0.972894482090997,
      "grad_norm": 0.8828611373901367,
      "learning_rate": 2.2603972228858737e-05,
      "loss": 2.219618320465088,
      "step": 6030
    },
    {
      "epoch": 0.9730558244595031,
      "grad_norm": 0.8775841593742371,
      "learning_rate": 2.234130678627169e-05,
      "loss": 2.5289759635925293,
      "step": 6031
    },
    {
      "epoch": 0.9732171668280091,
      "grad_norm": 1.1820279359817505,
      "learning_rate": 2.2080105006833495e-05,
      "loss": 2.035522222518921,
      "step": 6032
    },
    {
      "epoch": 0.973378509196515,
      "grad_norm": 1.2762315273284912,
      "learning_rate": 2.1820368569862443e-05,
      "loss": 1.8594800233840942,
      "step": 6033
    },
    {
      "epoch": 0.973539851565021,
      "grad_norm": 1.236182689666748,
      "learning_rate": 2.1562099145255866e-05,
      "loss": 2.3505783081054688,
      "step": 6034
    },
    {
      "epoch": 0.9737011939335269,
      "grad_norm": 0.999071478843689,
      "learning_rate": 2.130529839347953e-05,
      "loss": 2.3602142333984375,
      "step": 6035
    },
    {
      "epoch": 0.9738625363020329,
      "grad_norm": 0.7087242007255554,
      "learning_rate": 2.1049967965556583e-05,
      "loss": 2.2429022789001465,
      "step": 6036
    },
    {
      "epoch": 0.9740238786705389,
      "grad_norm": 1.0329786539077759,
      "learning_rate": 2.07961095030573e-05,
      "loss": 2.6023545265197754,
      "step": 6037
    },
    {
      "epoch": 0.9741852210390448,
      "grad_norm": 1.0305209159851074,
      "learning_rate": 2.0543724638088347e-05,
      "loss": 1.8393745422363281,
      "step": 6038
    },
    {
      "epoch": 0.9743465634075508,
      "grad_norm": 1.6528981924057007,
      "learning_rate": 2.0292814993282477e-05,
      "loss": 2.7104058265686035,
      "step": 6039
    },
    {
      "epoch": 0.9745079057760568,
      "grad_norm": 0.8192885518074036,
      "learning_rate": 2.0043382181787696e-05,
      "loss": 1.861919641494751,
      "step": 6040
    },
    {
      "epoch": 0.9746692481445628,
      "grad_norm": 0.9128857851028442,
      "learning_rate": 1.9795427807257375e-05,
      "loss": 2.27054500579834,
      "step": 6041
    },
    {
      "epoch": 0.9748305905130688,
      "grad_norm": 0.9423172473907471,
      "learning_rate": 1.954895346383967e-05,
      "loss": 2.255413293838501,
      "step": 6042
    },
    {
      "epoch": 0.9749919328815747,
      "grad_norm": 1.1158852577209473,
      "learning_rate": 1.930396073616725e-05,
      "loss": 2.5537590980529785,
      "step": 6043
    },
    {
      "epoch": 0.9751532752500807,
      "grad_norm": 0.9724969863891602,
      "learning_rate": 1.906045119934724e-05,
      "loss": 2.15265154838562,
      "step": 6044
    },
    {
      "epoch": 0.9753146176185866,
      "grad_norm": 0.9009048342704773,
      "learning_rate": 1.881842641895104e-05,
      "loss": 2.506720542907715,
      "step": 6045
    },
    {
      "epoch": 0.9754759599870926,
      "grad_norm": 0.7115519642829895,
      "learning_rate": 1.8577887951004264e-05,
      "loss": 1.8713715076446533,
      "step": 6046
    },
    {
      "epoch": 0.9756373023555985,
      "grad_norm": 1.1183743476867676,
      "learning_rate": 1.833883734197664e-05,
      "loss": 2.6109619140625,
      "step": 6047
    },
    {
      "epoch": 0.9757986447241046,
      "grad_norm": 1.0063577890396118,
      "learning_rate": 1.810127612877227e-05,
      "loss": 2.10982084274292,
      "step": 6048
    },
    {
      "epoch": 0.9759599870926106,
      "grad_norm": 0.8615561723709106,
      "learning_rate": 1.7865205838719563e-05,
      "loss": 2.402695655822754,
      "step": 6049
    },
    {
      "epoch": 0.9761213294611165,
      "grad_norm": 1.0363333225250244,
      "learning_rate": 1.763062798956147e-05,
      "loss": 2.2311904430389404,
      "step": 6050
    },
    {
      "epoch": 0.9762826718296225,
      "grad_norm": 1.4664112329483032,
      "learning_rate": 1.7397544089445772e-05,
      "loss": 2.1398682594299316,
      "step": 6051
    },
    {
      "epoch": 0.9764440141981284,
      "grad_norm": 0.8586583137512207,
      "learning_rate": 1.716595563691539e-05,
      "loss": 2.555011749267578,
      "step": 6052
    },
    {
      "epoch": 0.9766053565666344,
      "grad_norm": 0.9759625196456909,
      "learning_rate": 1.6935864120898703e-05,
      "loss": 2.2735018730163574,
      "step": 6053
    },
    {
      "epoch": 0.9767666989351403,
      "grad_norm": 0.7871874570846558,
      "learning_rate": 1.6707271020699882e-05,
      "loss": 1.8991920948028564,
      "step": 6054
    },
    {
      "epoch": 0.9769280413036463,
      "grad_norm": 1.0273021459579468,
      "learning_rate": 1.6480177805989623e-05,
      "loss": 2.4131951332092285,
      "step": 6055
    },
    {
      "epoch": 0.9770893836721523,
      "grad_norm": 1.3816227912902832,
      "learning_rate": 1.6254585936795547e-05,
      "loss": 2.575745105743408,
      "step": 6056
    },
    {
      "epoch": 0.9772507260406583,
      "grad_norm": 0.9226644039154053,
      "learning_rate": 1.6030496863492645e-05,
      "loss": 2.3723621368408203,
      "step": 6057
    },
    {
      "epoch": 0.9774120684091643,
      "grad_norm": 0.8330327868461609,
      "learning_rate": 1.580791202679438e-05,
      "loss": 2.192279815673828,
      "step": 6058
    },
    {
      "epoch": 0.9775734107776702,
      "grad_norm": 1.0661176443099976,
      "learning_rate": 1.5586832857743037e-05,
      "loss": 2.071718215942383,
      "step": 6059
    },
    {
      "epoch": 0.9777347531461762,
      "grad_norm": 0.7701851725578308,
      "learning_rate": 1.5367260777700676e-05,
      "loss": 2.3488047122955322,
      "step": 6060
    },
    {
      "epoch": 0.9778960955146822,
      "grad_norm": 0.7680377960205078,
      "learning_rate": 1.5149197198340014e-05,
      "loss": 2.4198291301727295,
      "step": 6061
    },
    {
      "epoch": 0.9780574378831881,
      "grad_norm": 1.1656229496002197,
      "learning_rate": 1.4932643521635258e-05,
      "loss": 2.046779155731201,
      "step": 6062
    },
    {
      "epoch": 0.9782187802516941,
      "grad_norm": 0.7945085167884827,
      "learning_rate": 1.4717601139853265e-05,
      "loss": 1.8886640071868896,
      "step": 6063
    },
    {
      "epoch": 0.9783801226202,
      "grad_norm": 1.106328010559082,
      "learning_rate": 1.4504071435544331e-05,
      "loss": 2.5415256023406982,
      "step": 6064
    },
    {
      "epoch": 0.9785414649887061,
      "grad_norm": 0.5602567195892334,
      "learning_rate": 1.4292055781533531e-05,
      "loss": 1.6844977140426636,
      "step": 6065
    },
    {
      "epoch": 0.978702807357212,
      "grad_norm": 1.101211428642273,
      "learning_rate": 1.408155554091184e-05,
      "loss": 2.098046064376831,
      "step": 6066
    },
    {
      "epoch": 0.978864149725718,
      "grad_norm": 0.9286410212516785,
      "learning_rate": 1.3872572067027216e-05,
      "loss": 2.4670538902282715,
      "step": 6067
    },
    {
      "epoch": 0.979025492094224,
      "grad_norm": 0.8980381488800049,
      "learning_rate": 1.3665106703476177e-05,
      "loss": 2.2299511432647705,
      "step": 6068
    },
    {
      "epoch": 0.9791868344627299,
      "grad_norm": 1.1344776153564453,
      "learning_rate": 1.3459160784094898e-05,
      "loss": 2.1600067615509033,
      "step": 6069
    },
    {
      "epoch": 0.9793481768312359,
      "grad_norm": 0.8696734309196472,
      "learning_rate": 1.3254735632950738e-05,
      "loss": 2.099778652191162,
      "step": 6070
    },
    {
      "epoch": 0.9795095191997418,
      "grad_norm": 0.9491071701049805,
      "learning_rate": 1.3051832564333816e-05,
      "loss": 2.520205497741699,
      "step": 6071
    },
    {
      "epoch": 0.9796708615682478,
      "grad_norm": 0.9207913875579834,
      "learning_rate": 1.2850452882748415e-05,
      "loss": 1.7973731756210327,
      "step": 6072
    },
    {
      "epoch": 0.9798322039367537,
      "grad_norm": 1.1050584316253662,
      "learning_rate": 1.265059788290468e-05,
      "loss": 2.6098790168762207,
      "step": 6073
    },
    {
      "epoch": 0.9799935463052598,
      "grad_norm": 0.9603385925292969,
      "learning_rate": 1.2452268849710292e-05,
      "loss": 2.408844470977783,
      "step": 6074
    },
    {
      "epoch": 0.9801548886737658,
      "grad_norm": 1.163619041442871,
      "learning_rate": 1.2255467058262082e-05,
      "loss": 2.29349684715271,
      "step": 6075
    },
    {
      "epoch": 0.9803162310422717,
      "grad_norm": 1.1686879396438599,
      "learning_rate": 1.206019377383813e-05,
      "loss": 2.3506810665130615,
      "step": 6076
    },
    {
      "epoch": 0.9804775734107777,
      "grad_norm": 0.9817716479301453,
      "learning_rate": 1.186645025188926e-05,
      "loss": 2.410287618637085,
      "step": 6077
    },
    {
      "epoch": 0.9806389157792836,
      "grad_norm": 1.2309532165527344,
      "learning_rate": 1.1674237738031224e-05,
      "loss": 2.3071401119232178,
      "step": 6078
    },
    {
      "epoch": 0.9808002581477896,
      "grad_norm": 1.1476224660873413,
      "learning_rate": 1.1483557468036671e-05,
      "loss": 2.052460193634033,
      "step": 6079
    },
    {
      "epoch": 0.9809616005162956,
      "grad_norm": 0.9722781181335449,
      "learning_rate": 1.129441066782702e-05,
      "loss": 2.6437716484069824,
      "step": 6080
    },
    {
      "epoch": 0.9811229428848015,
      "grad_norm": 1.0411872863769531,
      "learning_rate": 1.1106798553464804e-05,
      "loss": 2.167553424835205,
      "step": 6081
    },
    {
      "epoch": 0.9812842852533075,
      "grad_norm": 0.886066734790802,
      "learning_rate": 1.0920722331145771e-05,
      "loss": 2.2379565238952637,
      "step": 6082
    },
    {
      "epoch": 0.9814456276218135,
      "grad_norm": 1.4343388080596924,
      "learning_rate": 1.0736183197191024e-05,
      "loss": 3.011934995651245,
      "step": 6083
    },
    {
      "epoch": 0.9816069699903195,
      "grad_norm": 0.8794329762458801,
      "learning_rate": 1.0553182338039452e-05,
      "loss": 2.207777976989746,
      "step": 6084
    },
    {
      "epoch": 0.9817683123588254,
      "grad_norm": 1.5900734663009644,
      "learning_rate": 1.0371720930240108e-05,
      "loss": 2.3223843574523926,
      "step": 6085
    },
    {
      "epoch": 0.9819296547273314,
      "grad_norm": 0.8503746390342712,
      "learning_rate": 1.0191800140444574e-05,
      "loss": 2.054861307144165,
      "step": 6086
    },
    {
      "epoch": 0.9820909970958374,
      "grad_norm": 0.8969930410385132,
      "learning_rate": 1.0013421125399518e-05,
      "loss": 2.2213239669799805,
      "step": 6087
    },
    {
      "epoch": 0.9822523394643433,
      "grad_norm": 1.075161337852478,
      "learning_rate": 9.836585031939155e-06,
      "loss": 2.1551811695098877,
      "step": 6088
    },
    {
      "epoch": 0.9824136818328493,
      "grad_norm": 1.1293619871139526,
      "learning_rate": 9.661292996978072e-06,
      "loss": 1.9467893838882446,
      "step": 6089
    },
    {
      "epoch": 0.9825750242013552,
      "grad_norm": 0.8872674107551575,
      "learning_rate": 9.487546147503635e-06,
      "loss": 2.4157512187957764,
      "step": 6090
    },
    {
      "epoch": 0.9827363665698613,
      "grad_norm": 1.151003360748291,
      "learning_rate": 9.315345600569069e-06,
      "loss": 2.585761547088623,
      "step": 6091
    },
    {
      "epoch": 0.9828977089383673,
      "grad_norm": 1.2867331504821777,
      "learning_rate": 9.144692463286031e-06,
      "loss": 2.4912075996398926,
      "step": 6092
    },
    {
      "epoch": 0.9830590513068732,
      "grad_norm": 0.9241847991943359,
      "learning_rate": 8.975587832817544e-06,
      "loss": 2.3276138305664062,
      "step": 6093
    },
    {
      "epoch": 0.9832203936753792,
      "grad_norm": 0.7065495848655701,
      "learning_rate": 8.808032796371018e-06,
      "loss": 1.9765678644180298,
      "step": 6094
    },
    {
      "epoch": 0.9833817360438851,
      "grad_norm": 1.0029101371765137,
      "learning_rate": 8.642028431191246e-06,
      "loss": 2.2302310466766357,
      "step": 6095
    },
    {
      "epoch": 0.9835430784123911,
      "grad_norm": 0.7496291399002075,
      "learning_rate": 8.477575804553355e-06,
      "loss": 1.8351929187774658,
      "step": 6096
    },
    {
      "epoch": 0.983704420780897,
      "grad_norm": 0.9330483675003052,
      "learning_rate": 8.314675973756097e-06,
      "loss": 1.9436241388320923,
      "step": 6097
    },
    {
      "epoch": 0.983865763149403,
      "grad_norm": 0.9277779459953308,
      "learning_rate": 8.15332998611501e-06,
      "loss": 2.06622314453125,
      "step": 6098
    },
    {
      "epoch": 0.984027105517909,
      "grad_norm": 0.9686710238456726,
      "learning_rate": 7.99353887895557e-06,
      "loss": 2.290393352508545,
      "step": 6099
    },
    {
      "epoch": 0.984188447886415,
      "grad_norm": 0.9608275890350342,
      "learning_rate": 7.83530367960672e-06,
      "loss": 2.0917410850524902,
      "step": 6100
    },
    {
      "epoch": 0.984349790254921,
      "grad_norm": 1.6383705139160156,
      "learning_rate": 7.678625405394158e-06,
      "loss": 2.526249408721924,
      "step": 6101
    },
    {
      "epoch": 0.9845111326234269,
      "grad_norm": 1.014400601387024,
      "learning_rate": 7.523505063633695e-06,
      "loss": 2.2276082038879395,
      "step": 6102
    },
    {
      "epoch": 0.9846724749919329,
      "grad_norm": 1.5066064596176147,
      "learning_rate": 7.369943651624939e-06,
      "loss": 2.241896629333496,
      "step": 6103
    },
    {
      "epoch": 0.9848338173604388,
      "grad_norm": 0.7488582134246826,
      "learning_rate": 7.217942156644869e-06,
      "loss": 2.1147379875183105,
      "step": 6104
    },
    {
      "epoch": 0.9849951597289448,
      "grad_norm": 0.9124037027359009,
      "learning_rate": 7.067501555941464e-06,
      "loss": 2.0577731132507324,
      "step": 6105
    },
    {
      "epoch": 0.9851565020974508,
      "grad_norm": 1.1113595962524414,
      "learning_rate": 6.918622816727255e-06,
      "loss": 2.6082875728607178,
      "step": 6106
    },
    {
      "epoch": 0.9853178444659567,
      "grad_norm": 1.0441111326217651,
      "learning_rate": 6.7713068961733905e-06,
      "loss": 2.267536163330078,
      "step": 6107
    },
    {
      "epoch": 0.9854791868344628,
      "grad_norm": 0.9702252745628357,
      "learning_rate": 6.625554741403333e-06,
      "loss": 2.5811846256256104,
      "step": 6108
    },
    {
      "epoch": 0.9856405292029687,
      "grad_norm": 0.9865253567695618,
      "learning_rate": 6.481367289486784e-06,
      "loss": 1.8807036876678467,
      "step": 6109
    },
    {
      "epoch": 0.9858018715714747,
      "grad_norm": 1.107285499572754,
      "learning_rate": 6.3387454674335435e-06,
      "loss": 2.311748504638672,
      "step": 6110
    },
    {
      "epoch": 0.9859632139399807,
      "grad_norm": 1.13225257396698,
      "learning_rate": 6.197690192187827e-06,
      "loss": 2.4876022338867188,
      "step": 6111
    },
    {
      "epoch": 0.9861245563084866,
      "grad_norm": 1.0713584423065186,
      "learning_rate": 6.0582023706221014e-06,
      "loss": 2.4596123695373535,
      "step": 6112
    },
    {
      "epoch": 0.9862858986769926,
      "grad_norm": 0.9177340865135193,
      "learning_rate": 5.920282899531421e-06,
      "loss": 2.0642268657684326,
      "step": 6113
    },
    {
      "epoch": 0.9864472410454985,
      "grad_norm": 0.8807711601257324,
      "learning_rate": 5.783932665627545e-06,
      "loss": 1.9156864881515503,
      "step": 6114
    },
    {
      "epoch": 0.9866085834140045,
      "grad_norm": 0.8799114227294922,
      "learning_rate": 5.649152545533331e-06,
      "loss": 2.1189229488372803,
      "step": 6115
    },
    {
      "epoch": 0.9867699257825104,
      "grad_norm": 0.9380196928977966,
      "learning_rate": 5.515943405777102e-06,
      "loss": 2.0733180046081543,
      "step": 6116
    },
    {
      "epoch": 0.9869312681510165,
      "grad_norm": 1.1630345582962036,
      "learning_rate": 5.384306102786896e-06,
      "loss": 2.3741273880004883,
      "step": 6117
    },
    {
      "epoch": 0.9870926105195225,
      "grad_norm": 1.0493115186691284,
      "learning_rate": 5.254241482885253e-06,
      "loss": 1.9614596366882324,
      "step": 6118
    },
    {
      "epoch": 0.9872539528880284,
      "grad_norm": 0.7703579664230347,
      "learning_rate": 5.125750382283578e-06,
      "loss": 2.0368001461029053,
      "step": 6119
    },
    {
      "epoch": 0.9874152952565344,
      "grad_norm": 1.440340518951416,
      "learning_rate": 4.998833627076732e-06,
      "loss": 2.359316825866699,
      "step": 6120
    },
    {
      "epoch": 0.9875766376250403,
      "grad_norm": 1.0268266201019287,
      "learning_rate": 4.873492033237864e-06,
      "loss": 2.166057825088501,
      "step": 6121
    },
    {
      "epoch": 0.9877379799935463,
      "grad_norm": 1.1866154670715332,
      "learning_rate": 4.749726406613142e-06,
      "loss": 2.6153273582458496,
      "step": 6122
    },
    {
      "epoch": 0.9878993223620522,
      "grad_norm": 0.893658459186554,
      "learning_rate": 4.627537542916366e-06,
      "loss": 2.2236297130584717,
      "step": 6123
    },
    {
      "epoch": 0.9880606647305582,
      "grad_norm": 1.1610020399093628,
      "learning_rate": 4.506926227724168e-06,
      "loss": 2.2056944370269775,
      "step": 6124
    },
    {
      "epoch": 0.9882220070990642,
      "grad_norm": 0.9307556748390198,
      "learning_rate": 4.3878932364707635e-06,
      "loss": 2.561267375946045,
      "step": 6125
    },
    {
      "epoch": 0.9883833494675702,
      "grad_norm": 1.104583740234375,
      "learning_rate": 4.2704393344429885e-06,
      "loss": 2.3761324882507324,
      "step": 6126
    },
    {
      "epoch": 0.9885446918360762,
      "grad_norm": 0.8707414865493774,
      "learning_rate": 4.154565276775407e-06,
      "loss": 2.4512529373168945,
      "step": 6127
    },
    {
      "epoch": 0.9887060342045821,
      "grad_norm": 0.7759079337120056,
      "learning_rate": 4.040271808445406e-06,
      "loss": 2.0296707153320312,
      "step": 6128
    },
    {
      "epoch": 0.9888673765730881,
      "grad_norm": 1.2680777311325073,
      "learning_rate": 3.927559664268554e-06,
      "loss": 2.0396900177001953,
      "step": 6129
    },
    {
      "epoch": 0.989028718941594,
      "grad_norm": 0.90611732006073,
      "learning_rate": 3.816429568893665e-06,
      "loss": 2.066769599914551,
      "step": 6130
    },
    {
      "epoch": 0.9891900613101,
      "grad_norm": 1.0506030321121216,
      "learning_rate": 3.706882236798298e-06,
      "loss": 1.7000064849853516,
      "step": 6131
    },
    {
      "epoch": 0.989351403678606,
      "grad_norm": 0.7001917958259583,
      "learning_rate": 3.5989183722841267e-06,
      "loss": 1.7491871118545532,
      "step": 6132
    },
    {
      "epoch": 0.9895127460471119,
      "grad_norm": 0.8260694146156311,
      "learning_rate": 3.4925386694723284e-06,
      "loss": 2.000899314880371,
      "step": 6133
    },
    {
      "epoch": 0.989674088415618,
      "grad_norm": 0.9172551035881042,
      "learning_rate": 3.3877438122992545e-06,
      "loss": 2.407416820526123,
      "step": 6134
    },
    {
      "epoch": 0.989835430784124,
      "grad_norm": 1.1950404644012451,
      "learning_rate": 3.284534474511991e-06,
      "loss": 2.387174129486084,
      "step": 6135
    },
    {
      "epoch": 0.9899967731526299,
      "grad_norm": 0.8701843023300171,
      "learning_rate": 3.1829113196638614e-06,
      "loss": 2.4384493827819824,
      "step": 6136
    },
    {
      "epoch": 0.9901581155211359,
      "grad_norm": 1.040989637374878,
      "learning_rate": 3.0828750011104578e-06,
      "loss": 1.9802281856536865,
      "step": 6137
    },
    {
      "epoch": 0.9903194578896418,
      "grad_norm": 2.6996219158172607,
      "learning_rate": 2.984426162005227e-06,
      "loss": 2.2829768657684326,
      "step": 6138
    },
    {
      "epoch": 0.9904808002581478,
      "grad_norm": 1.1102136373519897,
      "learning_rate": 2.887565435295392e-06,
      "loss": 1.9903762340545654,
      "step": 6139
    },
    {
      "epoch": 0.9906421426266537,
      "grad_norm": 1.3116908073425293,
      "learning_rate": 2.7922934437178693e-06,
      "loss": 2.774747610092163,
      "step": 6140
    },
    {
      "epoch": 0.9908034849951597,
      "grad_norm": 1.2011369466781616,
      "learning_rate": 2.6986107997953036e-06,
      "loss": 2.131563663482666,
      "step": 6141
    },
    {
      "epoch": 0.9909648273636656,
      "grad_norm": 0.9280456304550171,
      "learning_rate": 2.6065181058320954e-06,
      "loss": 2.2014894485473633,
      "step": 6142
    },
    {
      "epoch": 0.9911261697321717,
      "grad_norm": 1.0028424263000488,
      "learning_rate": 2.516015953910544e-06,
      "loss": 2.3799920082092285,
      "step": 6143
    },
    {
      "epoch": 0.9912875121006777,
      "grad_norm": 0.7039561867713928,
      "learning_rate": 2.427104925887047e-06,
      "loss": 2.0499892234802246,
      "step": 6144
    },
    {
      "epoch": 0.9914488544691836,
      "grad_norm": 1.248971700668335,
      "learning_rate": 2.3397855933883216e-06,
      "loss": 2.61380672454834,
      "step": 6145
    },
    {
      "epoch": 0.9916101968376896,
      "grad_norm": 1.0693496465682983,
      "learning_rate": 2.2540585178078e-06,
      "loss": 2.4740633964538574,
      "step": 6146
    },
    {
      "epoch": 0.9917715392061955,
      "grad_norm": 0.8060062527656555,
      "learning_rate": 2.169924250301908e-06,
      "loss": 2.1322460174560547,
      "step": 6147
    },
    {
      "epoch": 0.9919328815747015,
      "grad_norm": 1.1439803838729858,
      "learning_rate": 2.0873833317866796e-06,
      "loss": 1.9969699382781982,
      "step": 6148
    },
    {
      "epoch": 0.9920942239432075,
      "grad_norm": 1.0229254961013794,
      "learning_rate": 2.006436292934094e-06,
      "loss": 2.0745716094970703,
      "step": 6149
    },
    {
      "epoch": 0.9922555663117134,
      "grad_norm": 1.1616977453231812,
      "learning_rate": 1.9270836541688543e-06,
      "loss": 2.4098873138427734,
      "step": 6150
    },
    {
      "epoch": 0.9924169086802195,
      "grad_norm": 0.9689063429832458,
      "learning_rate": 1.8493259256649187e-06,
      "loss": 2.362309694290161,
      "step": 6151
    },
    {
      "epoch": 0.9925782510487254,
      "grad_norm": 1.064650058746338,
      "learning_rate": 1.7731636073421709e-06,
      "loss": 2.002988815307617,
      "step": 6152
    },
    {
      "epoch": 0.9927395934172314,
      "grad_norm": 1.1330592632293701,
      "learning_rate": 1.6985971888633933e-06,
      "loss": 2.5287251472473145,
      "step": 6153
    },
    {
      "epoch": 0.9929009357857373,
      "grad_norm": 1.0234962701797485,
      "learning_rate": 1.6256271496309371e-06,
      "loss": 2.1432547569274902,
      "step": 6154
    },
    {
      "epoch": 0.9930622781542433,
      "grad_norm": 1.1904211044311523,
      "learning_rate": 1.5542539587837523e-06,
      "loss": 1.9867236614227295,
      "step": 6155
    },
    {
      "epoch": 0.9932236205227493,
      "grad_norm": 1.1386644840240479,
      "learning_rate": 1.4844780751943343e-06,
      "loss": 2.246623992919922,
      "step": 6156
    },
    {
      "epoch": 0.9933849628912552,
      "grad_norm": 0.686682939529419,
      "learning_rate": 1.4162999474657267e-06,
      "loss": 1.8128349781036377,
      "step": 6157
    },
    {
      "epoch": 0.9935463052597612,
      "grad_norm": 1.11471426486969,
      "learning_rate": 1.349720013928718e-06,
      "loss": 2.284450054168701,
      "step": 6158
    },
    {
      "epoch": 0.9937076476282671,
      "grad_norm": 1.119571328163147,
      "learning_rate": 1.28473870263901e-06,
      "loss": 2.5564918518066406,
      "step": 6159
    },
    {
      "epoch": 0.9938689899967732,
      "grad_norm": 1.212131142616272,
      "learning_rate": 1.2213564313743874e-06,
      "loss": 2.342081308364868,
      "step": 6160
    },
    {
      "epoch": 0.9940303323652792,
      "grad_norm": 0.7850072979927063,
      "learning_rate": 1.1595736076321362e-06,
      "loss": 2.510138988494873,
      "step": 6161
    },
    {
      "epoch": 0.9941916747337851,
      "grad_norm": 1.0850344896316528,
      "learning_rate": 1.0993906286263234e-06,
      "loss": 2.0253512859344482,
      "step": 6162
    },
    {
      "epoch": 0.9943530171022911,
      "grad_norm": 1.7543383836746216,
      "learning_rate": 1.0408078812853272e-06,
      "loss": 2.244746685028076,
      "step": 6163
    },
    {
      "epoch": 0.994514359470797,
      "grad_norm": 1.0425610542297363,
      "learning_rate": 9.83825742249367e-07,
      "loss": 2.2081875801086426,
      "step": 6164
    },
    {
      "epoch": 0.994675701839303,
      "grad_norm": 0.9716667532920837,
      "learning_rate": 9.284445778679207e-07,
      "loss": 2.021078586578369,
      "step": 6165
    },
    {
      "epoch": 0.9948370442078089,
      "grad_norm": 0.9853355288505554,
      "learning_rate": 8.746647441975619e-07,
      "loss": 2.3544650077819824,
      "step": 6166
    },
    {
      "epoch": 0.9949983865763149,
      "grad_norm": 1.073170781135559,
      "learning_rate": 8.224865869995712e-07,
      "loss": 2.5240707397460938,
      "step": 6167
    },
    {
      "epoch": 0.995159728944821,
      "grad_norm": 0.8451960682868958,
      "learning_rate": 7.719104417377442e-07,
      "loss": 2.2321105003356934,
      "step": 6168
    },
    {
      "epoch": 0.9953210713133269,
      "grad_norm": 1.1042381525039673,
      "learning_rate": 7.229366335761989e-07,
      "loss": 2.3929922580718994,
      "step": 6169
    },
    {
      "epoch": 0.9954824136818329,
      "grad_norm": 1.594201683998108,
      "learning_rate": 6.755654773772657e-07,
      "loss": 2.6285054683685303,
      "step": 6170
    },
    {
      "epoch": 0.9956437560503388,
      "grad_norm": 1.1400349140167236,
      "learning_rate": 6.297972776996286e-07,
      "loss": 2.671910285949707,
      "step": 6171
    },
    {
      "epoch": 0.9958050984188448,
      "grad_norm": 0.8163478374481201,
      "learning_rate": 5.856323287961318e-07,
      "loss": 2.219989776611328,
      "step": 6172
    },
    {
      "epoch": 0.9959664407873507,
      "grad_norm": 0.8686359524726868,
      "learning_rate": 5.430709146120593e-07,
      "loss": 1.7557708024978638,
      "step": 6173
    },
    {
      "epoch": 0.9961277831558567,
      "grad_norm": 0.7409380674362183,
      "learning_rate": 5.021133087832475e-07,
      "loss": 2.0358710289001465,
      "step": 6174
    },
    {
      "epoch": 0.9962891255243627,
      "grad_norm": 1.043488621711731,
      "learning_rate": 4.627597746342804e-07,
      "loss": 1.8450884819030762,
      "step": 6175
    },
    {
      "epoch": 0.9964504678928686,
      "grad_norm": 0.9233027696609497,
      "learning_rate": 4.2501056517688097e-07,
      "loss": 2.7609987258911133,
      "step": 6176
    },
    {
      "epoch": 0.9966118102613747,
      "grad_norm": 0.9349467754364014,
      "learning_rate": 3.8886592310824477e-07,
      "loss": 2.3394176959991455,
      "step": 6177
    },
    {
      "epoch": 0.9967731526298806,
      "grad_norm": 1.3075230121612549,
      "learning_rate": 3.543260808095139e-07,
      "loss": 2.192835807800293,
      "step": 6178
    },
    {
      "epoch": 0.9969344949983866,
      "grad_norm": 1.0985404253005981,
      "learning_rate": 3.2139126034419487e-07,
      "loss": 1.6667598485946655,
      "step": 6179
    },
    {
      "epoch": 0.9970958373668926,
      "grad_norm": 1.1137511730194092,
      "learning_rate": 2.9006167345685397e-07,
      "loss": 2.262258768081665,
      "step": 6180
    },
    {
      "epoch": 0.9972571797353985,
      "grad_norm": 1.3338944911956787,
      "learning_rate": 2.603375215716186e-07,
      "loss": 1.9340183734893799,
      "step": 6181
    },
    {
      "epoch": 0.9974185221039045,
      "grad_norm": 1.2517050504684448,
      "learning_rate": 2.3221899579098372e-07,
      "loss": 2.127470016479492,
      "step": 6182
    },
    {
      "epoch": 0.9975798644724104,
      "grad_norm": 0.9439082145690918,
      "learning_rate": 2.0570627689459054e-07,
      "loss": 2.1405835151672363,
      "step": 6183
    },
    {
      "epoch": 0.9977412068409164,
      "grad_norm": 0.9046352505683899,
      "learning_rate": 1.807995353380054e-07,
      "loss": 2.3711466789245605,
      "step": 6184
    },
    {
      "epoch": 0.9979025492094223,
      "grad_norm": 0.770023763179779,
      "learning_rate": 1.574989312516095e-07,
      "loss": 2.054539203643799,
      "step": 6185
    },
    {
      "epoch": 0.9980638915779284,
      "grad_norm": 1.0584615468978882,
      "learning_rate": 1.3580461443962743e-07,
      "loss": 2.330411672592163,
      "step": 6186
    },
    {
      "epoch": 0.9982252339464344,
      "grad_norm": 1.0023915767669678,
      "learning_rate": 1.157167243791557e-07,
      "loss": 2.2651820182800293,
      "step": 6187
    },
    {
      "epoch": 0.9983865763149403,
      "grad_norm": 1.0088281631469727,
      "learning_rate": 9.723539021927463e-08,
      "loss": 1.8838385343551636,
      "step": 6188
    },
    {
      "epoch": 0.9985479186834463,
      "grad_norm": 1.1819628477096558,
      "learning_rate": 8.036073078016015e-08,
      "loss": 1.9939017295837402,
      "step": 6189
    },
    {
      "epoch": 0.9987092610519522,
      "grad_norm": 0.8517775535583496,
      "learning_rate": 6.509285455233438e-08,
      "loss": 2.274993419647217,
      "step": 6190
    },
    {
      "epoch": 0.9988706034204582,
      "grad_norm": 0.8568407893180847,
      "learning_rate": 5.143185969602726e-08,
      "loss": 2.2528200149536133,
      "step": 6191
    },
    {
      "epoch": 0.9990319457889641,
      "grad_norm": 0.9140080213546753,
      "learning_rate": 3.9377834040538187e-08,
      "loss": 2.2739553451538086,
      "step": 6192
    },
    {
      "epoch": 0.9991932881574701,
      "grad_norm": 1.1259822845458984,
      "learning_rate": 2.8930855083542095e-08,
      "loss": 2.0543746948242188,
      "step": 6193
    },
    {
      "epoch": 0.9993546305259762,
      "grad_norm": 1.0607000589370728,
      "learning_rate": 2.009098999078418e-08,
      "loss": 2.1558632850646973,
      "step": 6194
    },
    {
      "epoch": 0.9995159728944821,
      "grad_norm": 0.9089713096618652,
      "learning_rate": 1.2858295595552516e-08,
      "loss": 2.109231472015381,
      "step": 6195
    },
    {
      "epoch": 0.9996773152629881,
      "grad_norm": 1.2143439054489136,
      "learning_rate": 7.23281839820622e-09,
      "loss": 2.397637367248535,
      "step": 6196
    },
    {
      "epoch": 0.999838657631494,
      "grad_norm": 1.3220008611679077,
      "learning_rate": 3.214594566119944e-09,
      "loss": 2.699350595474243,
      "step": 6197
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1640057563781738,
      "learning_rate": 8.036499332397807e-10,
      "loss": 1.7689704895019531,
      "step": 6198
    }
  ],
  "logging_steps": 1,
  "max_steps": 6198,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 620,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9.580327193691554e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}