{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.06815729925728017,
  "eval_steps": 500,
  "global_step": 35000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.9210271110262932e-05,
      "grad_norm": 5.397985935211182,
      "learning_rate": 2.9999999936841446e-05,
      "loss": 2.1156,
      "step": 15
    },
    {
      "epoch": 5.8420542220525864e-05,
      "grad_norm": 3.999190092086792,
      "learning_rate": 2.999999974736578e-05,
      "loss": 2.5268,
      "step": 30
    },
    {
      "epoch": 8.76308133307888e-05,
      "grad_norm": 5.580333709716797,
      "learning_rate": 2.999999943157301e-05,
      "loss": 2.3955,
      "step": 45
    },
    {
      "epoch": 0.00011684108444105173,
      "grad_norm": 3.6752560138702393,
      "learning_rate": 2.9999998989463132e-05,
      "loss": 2.171,
      "step": 60
    },
    {
      "epoch": 0.00014605135555131467,
      "grad_norm": 2.655461072921753,
      "learning_rate": 2.9999998421036153e-05,
      "loss": 2.188,
      "step": 75
    },
    {
      "epoch": 0.0001752616266615776,
      "grad_norm": 2.572371006011963,
      "learning_rate": 2.9999997726292083e-05,
      "loss": 2.1977,
      "step": 90
    },
    {
      "epoch": 0.00020447189777184053,
      "grad_norm": 8.544811248779297,
      "learning_rate": 2.999999690523092e-05,
      "loss": 2.1871,
      "step": 105
    },
    {
      "epoch": 0.00023368216888210345,
      "grad_norm": 3.530529737472534,
      "learning_rate": 2.999999595785267e-05,
      "loss": 2.2238,
      "step": 120
    },
    {
      "epoch": 0.00026289243999236636,
      "grad_norm": 3.671182155609131,
      "learning_rate": 2.9999994884157345e-05,
      "loss": 2.1338,
      "step": 135
    },
    {
      "epoch": 0.00029210271110262934,
      "grad_norm": 3.9643137454986572,
      "learning_rate": 2.9999993684144956e-05,
      "loss": 2.0788,
      "step": 150
    },
    {
      "epoch": 0.00032131298221289227,
      "grad_norm": 2.62164044380188,
      "learning_rate": 2.9999992357815508e-05,
      "loss": 2.117,
      "step": 165
    },
    {
      "epoch": 0.0003505232533231552,
      "grad_norm": 4.189517021179199,
      "learning_rate": 2.999999090516902e-05,
      "loss": 2.262,
      "step": 180
    },
    {
      "epoch": 0.0003797335244334181,
      "grad_norm": 2.880622625350952,
      "learning_rate": 2.9999989326205494e-05,
      "loss": 2.1102,
      "step": 195
    },
    {
      "epoch": 0.00040894379554368105,
      "grad_norm": 5.852839469909668,
      "learning_rate": 2.9999987620924948e-05,
      "loss": 2.0684,
      "step": 210
    },
    {
      "epoch": 0.000438154066653944,
      "grad_norm": 3.6339924335479736,
      "learning_rate": 2.9999985789327394e-05,
      "loss": 2.2518,
      "step": 225
    },
    {
      "epoch": 0.0004673643377642069,
      "grad_norm": 2.743267297744751,
      "learning_rate": 2.9999983831412858e-05,
      "loss": 2.1236,
      "step": 240
    },
    {
      "epoch": 0.0004965746088744698,
      "grad_norm": 3.9490363597869873,
      "learning_rate": 2.9999981747181345e-05,
      "loss": 2.0878,
      "step": 255
    },
    {
      "epoch": 0.0005257848799847327,
      "grad_norm": 3.0342047214508057,
      "learning_rate": 2.9999979536632872e-05,
      "loss": 2.1415,
      "step": 270
    },
    {
      "epoch": 0.0005549951510949957,
      "grad_norm": 4.045855522155762,
      "learning_rate": 2.9999977199767467e-05,
      "loss": 2.0558,
      "step": 285
    },
    {
      "epoch": 0.0005842054222052587,
      "grad_norm": 4.325068950653076,
      "learning_rate": 2.999997473658514e-05,
      "loss": 2.0617,
      "step": 300
    },
    {
      "epoch": 0.0006134156933155216,
      "grad_norm": 4.770871162414551,
      "learning_rate": 2.999997214708592e-05,
      "loss": 2.0922,
      "step": 315
    },
    {
      "epoch": 0.0006426259644257845,
      "grad_norm": 4.1802754402160645,
      "learning_rate": 2.999996943126982e-05,
      "loss": 2.217,
      "step": 330
    },
    {
      "epoch": 0.0006718362355360475,
      "grad_norm": 3.044260263442993,
      "learning_rate": 2.999996658913687e-05,
      "loss": 2.0508,
      "step": 345
    },
    {
      "epoch": 0.0007010465066463104,
      "grad_norm": 4.1525187492370605,
      "learning_rate": 2.9999963620687095e-05,
      "loss": 2.1678,
      "step": 360
    },
    {
      "epoch": 0.0007302567777565733,
      "grad_norm": 2.5676231384277344,
      "learning_rate": 2.9999960525920515e-05,
      "loss": 2.1693,
      "step": 375
    },
    {
      "epoch": 0.0007594670488668362,
      "grad_norm": 3.880307674407959,
      "learning_rate": 2.9999957304837156e-05,
      "loss": 1.9008,
      "step": 390
    },
    {
      "epoch": 0.0007886773199770992,
      "grad_norm": 3.3433754444122314,
      "learning_rate": 2.999995395743705e-05,
      "loss": 2.0734,
      "step": 405
    },
    {
      "epoch": 0.0008178875910873621,
      "grad_norm": 3.451378583908081,
      "learning_rate": 2.999995048372022e-05,
      "loss": 1.9901,
      "step": 420
    },
    {
      "epoch": 0.000847097862197625,
      "grad_norm": 3.4411580562591553,
      "learning_rate": 2.9999946883686695e-05,
      "loss": 1.9734,
      "step": 435
    },
    {
      "epoch": 0.000876308133307888,
      "grad_norm": 3.514651298522949,
      "learning_rate": 2.999994315733651e-05,
      "loss": 2.1632,
      "step": 450
    },
    {
      "epoch": 0.0009055184044181509,
      "grad_norm": 5.380824565887451,
      "learning_rate": 2.999993930466969e-05,
      "loss": 2.1534,
      "step": 465
    },
    {
      "epoch": 0.0009347286755284138,
      "grad_norm": 3.768618106842041,
      "learning_rate": 2.999993532568628e-05,
      "loss": 2.2182,
      "step": 480
    },
    {
      "epoch": 0.0009639389466386767,
      "grad_norm": 2.9490509033203125,
      "learning_rate": 2.99999312203863e-05,
      "loss": 1.9677,
      "step": 495
    },
    {
      "epoch": 0.0009931492177489396,
      "grad_norm": 3.4881184101104736,
      "learning_rate": 2.9999926988769788e-05,
      "loss": 2.0786,
      "step": 510
    },
    {
      "epoch": 0.0010223594888592025,
      "grad_norm": 3.788675546646118,
      "learning_rate": 2.9999922630836784e-05,
      "loss": 1.9696,
      "step": 525
    },
    {
      "epoch": 0.0010515697599694654,
      "grad_norm": 3.070878744125366,
      "learning_rate": 2.999991814658732e-05,
      "loss": 1.9053,
      "step": 540
    },
    {
      "epoch": 0.0010807800310797286,
      "grad_norm": 2.456150531768799,
      "learning_rate": 2.9999913536021436e-05,
      "loss": 1.9813,
      "step": 555
    },
    {
      "epoch": 0.0011099903021899915,
      "grad_norm": 4.81670618057251,
      "learning_rate": 2.999990879913917e-05,
      "loss": 1.9317,
      "step": 570
    },
    {
      "epoch": 0.0011392005733002544,
      "grad_norm": 4.014492988586426,
      "learning_rate": 2.9999903935940567e-05,
      "loss": 1.9709,
      "step": 585
    },
    {
      "epoch": 0.0011684108444105174,
      "grad_norm": 4.765859603881836,
      "learning_rate": 2.999989894642566e-05,
      "loss": 1.9252,
      "step": 600
    },
    {
      "epoch": 0.0011976211155207803,
      "grad_norm": 3.5599727630615234,
      "learning_rate": 2.9999893830594492e-05,
      "loss": 2.0175,
      "step": 615
    },
    {
      "epoch": 0.0012268313866310432,
      "grad_norm": 4.441170692443848,
      "learning_rate": 2.999988858844711e-05,
      "loss": 2.1166,
      "step": 630
    },
    {
      "epoch": 0.0012560416577413061,
      "grad_norm": 2.691765308380127,
      "learning_rate": 2.9999883219983558e-05,
      "loss": 1.991,
      "step": 645
    },
    {
      "epoch": 0.001285251928851569,
      "grad_norm": 4.039743423461914,
      "learning_rate": 2.9999877725203878e-05,
      "loss": 1.9917,
      "step": 660
    },
    {
      "epoch": 0.001314462199961832,
      "grad_norm": 3.0612881183624268,
      "learning_rate": 2.9999872104108115e-05,
      "loss": 2.0436,
      "step": 675
    },
    {
      "epoch": 0.001343672471072095,
      "grad_norm": 3.3579518795013428,
      "learning_rate": 2.9999866356696326e-05,
      "loss": 1.9659,
      "step": 690
    },
    {
      "epoch": 0.0013728827421823579,
      "grad_norm": 4.145946502685547,
      "learning_rate": 2.9999860482968552e-05,
      "loss": 2.0267,
      "step": 705
    },
    {
      "epoch": 0.0014020930132926208,
      "grad_norm": 5.213028430938721,
      "learning_rate": 2.999985448292484e-05,
      "loss": 2.0953,
      "step": 720
    },
    {
      "epoch": 0.0014313032844028837,
      "grad_norm": 2.440793752670288,
      "learning_rate": 2.9999848356565246e-05,
      "loss": 2.0218,
      "step": 735
    },
    {
      "epoch": 0.0014605135555131466,
      "grad_norm": 2.6167280673980713,
      "learning_rate": 2.9999842103889813e-05,
      "loss": 2.0638,
      "step": 750
    },
    {
      "epoch": 0.0014897238266234096,
      "grad_norm": 2.2680811882019043,
      "learning_rate": 2.999983572489861e-05,
      "loss": 1.973,
      "step": 765
    },
    {
      "epoch": 0.0015189340977336725,
      "grad_norm": 2.905947208404541,
      "learning_rate": 2.999982921959167e-05,
      "loss": 1.9346,
      "step": 780
    },
    {
      "epoch": 0.0015481443688439354,
      "grad_norm": 4.673079967498779,
      "learning_rate": 2.999982258796907e-05,
      "loss": 2.1666,
      "step": 795
    },
    {
      "epoch": 0.0015773546399541984,
      "grad_norm": 2.5963408946990967,
      "learning_rate": 2.9999815830030846e-05,
      "loss": 2.1063,
      "step": 810
    },
    {
      "epoch": 0.0016065649110644613,
      "grad_norm": 3.656632661819458,
      "learning_rate": 2.9999808945777066e-05,
      "loss": 2.1328,
      "step": 825
    },
    {
      "epoch": 0.0016357751821747242,
      "grad_norm": 4.4109368324279785,
      "learning_rate": 2.9999801935207786e-05,
      "loss": 1.9602,
      "step": 840
    },
    {
      "epoch": 0.0016649854532849871,
      "grad_norm": 2.10768985748291,
      "learning_rate": 2.9999794798323065e-05,
      "loss": 1.9766,
      "step": 855
    },
    {
      "epoch": 0.00169419572439525,
      "grad_norm": 4.301672458648682,
      "learning_rate": 2.999978753512296e-05,
      "loss": 1.9386,
      "step": 870
    },
    {
      "epoch": 0.001723405995505513,
      "grad_norm": 8.171302795410156,
      "learning_rate": 2.9999780145607538e-05,
      "loss": 2.1288,
      "step": 885
    },
    {
      "epoch": 0.001752616266615776,
      "grad_norm": 3.4784069061279297,
      "learning_rate": 2.999977262977685e-05,
      "loss": 2.1346,
      "step": 900
    },
    {
      "epoch": 0.0017818265377260389,
      "grad_norm": 2.9945054054260254,
      "learning_rate": 2.9999764987630976e-05,
      "loss": 2.0432,
      "step": 915
    },
    {
      "epoch": 0.0018110368088363018,
      "grad_norm": 3.632039785385132,
      "learning_rate": 2.9999757219169964e-05,
      "loss": 2.0642,
      "step": 930
    },
    {
      "epoch": 0.0018402470799465647,
      "grad_norm": 3.9847047328948975,
      "learning_rate": 2.999974932439389e-05,
      "loss": 2.016,
      "step": 945
    },
    {
      "epoch": 0.0018694573510568276,
      "grad_norm": 1.89451003074646,
      "learning_rate": 2.9999741303302816e-05,
      "loss": 2.0355,
      "step": 960
    },
    {
      "epoch": 0.0018986676221670906,
      "grad_norm": 6.945189952850342,
      "learning_rate": 2.9999733155896814e-05,
      "loss": 1.97,
      "step": 975
    },
    {
      "epoch": 0.0019278778932773535,
      "grad_norm": 3.347123384475708,
      "learning_rate": 2.9999724882175947e-05,
      "loss": 2.0421,
      "step": 990
    },
    {
      "epoch": 0.0019570881643876164,
      "grad_norm": 4.331577777862549,
      "learning_rate": 2.9999716482140292e-05,
      "loss": 2.133,
      "step": 1005
    },
    {
      "epoch": 0.001986298435497879,
      "grad_norm": 4.221049785614014,
      "learning_rate": 2.999970795578991e-05,
      "loss": 1.9272,
      "step": 1020
    },
    {
      "epoch": 0.0020155087066081423,
      "grad_norm": 2.820549488067627,
      "learning_rate": 2.9999699303124876e-05,
      "loss": 2.0116,
      "step": 1035
    },
    {
      "epoch": 0.002044718977718405,
      "grad_norm": 2.2391343116760254,
      "learning_rate": 2.9999690524145268e-05,
      "loss": 1.9337,
      "step": 1050
    },
    {
      "epoch": 0.002073929248828668,
      "grad_norm": 2.285027503967285,
      "learning_rate": 2.9999681618851156e-05,
      "loss": 1.9773,
      "step": 1065
    },
    {
      "epoch": 0.002103139519938931,
      "grad_norm": 3.266909599304199,
      "learning_rate": 2.9999672587242616e-05,
      "loss": 1.9471,
      "step": 1080
    },
    {
      "epoch": 0.002132349791049194,
      "grad_norm": 5.03222131729126,
      "learning_rate": 2.9999663429319722e-05,
      "loss": 2.1337,
      "step": 1095
    },
    {
      "epoch": 0.002161560062159457,
      "grad_norm": 2.044246196746826,
      "learning_rate": 2.9999654145082552e-05,
      "loss": 2.06,
      "step": 1110
    },
    {
      "epoch": 0.00219077033326972,
      "grad_norm": 3.1399905681610107,
      "learning_rate": 2.9999644734531183e-05,
      "loss": 1.9174,
      "step": 1125
    },
    {
      "epoch": 0.002219980604379983,
      "grad_norm": 3.2445454597473145,
      "learning_rate": 2.99996351976657e-05,
      "loss": 1.9728,
      "step": 1140
    },
    {
      "epoch": 0.0022491908754902457,
      "grad_norm": 2.3682138919830322,
      "learning_rate": 2.999962553448618e-05,
      "loss": 1.9779,
      "step": 1155
    },
    {
      "epoch": 0.002278401146600509,
      "grad_norm": 2.283808708190918,
      "learning_rate": 2.99996157449927e-05,
      "loss": 1.964,
      "step": 1170
    },
    {
      "epoch": 0.0023076114177107716,
      "grad_norm": 3.048382043838501,
      "learning_rate": 2.9999605829185353e-05,
      "loss": 2.0908,
      "step": 1185
    },
    {
      "epoch": 0.0023368216888210347,
      "grad_norm": 6.139120101928711,
      "learning_rate": 2.999959578706421e-05,
      "loss": 1.9535,
      "step": 1200
    },
    {
      "epoch": 0.0023660319599312974,
      "grad_norm": 2.104069232940674,
      "learning_rate": 2.9999585618629363e-05,
      "loss": 1.8775,
      "step": 1215
    },
    {
      "epoch": 0.0023952422310415606,
      "grad_norm": 2.531449317932129,
      "learning_rate": 2.9999575323880894e-05,
      "loss": 2.0309,
      "step": 1230
    },
    {
      "epoch": 0.0024244525021518233,
      "grad_norm": 2.841078996658325,
      "learning_rate": 2.9999564902818894e-05,
      "loss": 2.0546,
      "step": 1245
    },
    {
      "epoch": 0.0024536627732620864,
      "grad_norm": 2.0627243518829346,
      "learning_rate": 2.9999554355443445e-05,
      "loss": 1.8833,
      "step": 1260
    },
    {
      "epoch": 0.002482873044372349,
      "grad_norm": 3.4014151096343994,
      "learning_rate": 2.9999543681754645e-05,
      "loss": 1.8935,
      "step": 1275
    },
    {
      "epoch": 0.0025120833154826123,
      "grad_norm": 3.507380962371826,
      "learning_rate": 2.999953288175257e-05,
      "loss": 1.9512,
      "step": 1290
    },
    {
      "epoch": 0.002541293586592875,
      "grad_norm": 5.4273362159729,
      "learning_rate": 2.9999521955437325e-05,
      "loss": 1.9794,
      "step": 1305
    },
    {
      "epoch": 0.002570503857703138,
      "grad_norm": 4.466185092926025,
      "learning_rate": 2.9999510902808993e-05,
      "loss": 1.8266,
      "step": 1320
    },
    {
      "epoch": 0.002599714128813401,
      "grad_norm": 2.4276440143585205,
      "learning_rate": 2.9999499723867672e-05,
      "loss": 2.0149,
      "step": 1335
    },
    {
      "epoch": 0.002628924399923664,
      "grad_norm": 3.589876413345337,
      "learning_rate": 2.9999488418613454e-05,
      "loss": 1.8865,
      "step": 1350
    },
    {
      "epoch": 0.0026581346710339267,
      "grad_norm": 3.750687599182129,
      "learning_rate": 2.9999476987046434e-05,
      "loss": 2.13,
      "step": 1365
    },
    {
      "epoch": 0.00268734494214419,
      "grad_norm": 3.638850212097168,
      "learning_rate": 2.999946542916671e-05,
      "loss": 1.981,
      "step": 1380
    },
    {
      "epoch": 0.0027165552132544526,
      "grad_norm": 4.610542297363281,
      "learning_rate": 2.9999453744974375e-05,
      "loss": 1.8887,
      "step": 1395
    },
    {
      "epoch": 0.0027457654843647157,
      "grad_norm": 2.3758933544158936,
      "learning_rate": 2.9999441934469534e-05,
      "loss": 1.7867,
      "step": 1410
    },
    {
      "epoch": 0.0027749757554749784,
      "grad_norm": 1.9892805814743042,
      "learning_rate": 2.999942999765228e-05,
      "loss": 1.8632,
      "step": 1425
    },
    {
      "epoch": 0.0028041860265852416,
      "grad_norm": 2.5921459197998047,
      "learning_rate": 2.999941793452272e-05,
      "loss": 1.9155,
      "step": 1440
    },
    {
      "epoch": 0.0028333962976955043,
      "grad_norm": 6.098895072937012,
      "learning_rate": 2.9999405745080948e-05,
      "loss": 2.0112,
      "step": 1455
    },
    {
      "epoch": 0.0028626065688057674,
      "grad_norm": 2.7144930362701416,
      "learning_rate": 2.9999393429327073e-05,
      "loss": 2.0019,
      "step": 1470
    },
    {
      "epoch": 0.00289181683991603,
      "grad_norm": 2.3964602947235107,
      "learning_rate": 2.9999380987261195e-05,
      "loss": 1.9524,
      "step": 1485
    },
    {
      "epoch": 0.0029210271110262933,
      "grad_norm": 3.2644741535186768,
      "learning_rate": 2.9999368418883422e-05,
      "loss": 1.881,
      "step": 1500
    },
    {
      "epoch": 0.002950237382136556,
      "grad_norm": 3.0930943489074707,
      "learning_rate": 2.9999355724193854e-05,
      "loss": 2.0001,
      "step": 1515
    },
    {
      "epoch": 0.002979447653246819,
      "grad_norm": 2.6339657306671143,
      "learning_rate": 2.9999342903192608e-05,
      "loss": 1.9344,
      "step": 1530
    },
    {
      "epoch": 0.003008657924357082,
      "grad_norm": 3.7254459857940674,
      "learning_rate": 2.999932995587978e-05,
      "loss": 2.0673,
      "step": 1545
    },
    {
      "epoch": 0.003037868195467345,
      "grad_norm": 2.7078473567962646,
      "learning_rate": 2.999931688225549e-05,
      "loss": 2.0223,
      "step": 1560
    },
    {
      "epoch": 0.0030670784665776077,
      "grad_norm": 2.7792718410491943,
      "learning_rate": 2.999930368231984e-05,
      "loss": 1.902,
      "step": 1575
    },
    {
      "epoch": 0.003096288737687871,
      "grad_norm": 4.607854843139648,
      "learning_rate": 2.999929035607294e-05,
      "loss": 1.9536,
      "step": 1590
    },
    {
      "epoch": 0.0031254990087981336,
      "grad_norm": 2.4656715393066406,
      "learning_rate": 2.999927690351491e-05,
      "loss": 2.1792,
      "step": 1605
    },
    {
      "epoch": 0.0031547092799083967,
      "grad_norm": 2.8231770992279053,
      "learning_rate": 2.9999263324645863e-05,
      "loss": 1.8908,
      "step": 1620
    },
    {
      "epoch": 0.0031839195510186594,
      "grad_norm": 3.5741395950317383,
      "learning_rate": 2.9999249619465904e-05,
      "loss": 1.9505,
      "step": 1635
    },
    {
      "epoch": 0.0032131298221289226,
      "grad_norm": 2.0066606998443604,
      "learning_rate": 2.999923578797516e-05,
      "loss": 2.0346,
      "step": 1650
    },
    {
      "epoch": 0.0032423400932391853,
      "grad_norm": 2.5830793380737305,
      "learning_rate": 2.999922183017374e-05,
      "loss": 1.9546,
      "step": 1665
    },
    {
      "epoch": 0.0032715503643494484,
      "grad_norm": 2.8811633586883545,
      "learning_rate": 2.999920774606176e-05,
      "loss": 1.8517,
      "step": 1680
    },
    {
      "epoch": 0.003300760635459711,
      "grad_norm": 3.0571601390838623,
      "learning_rate": 2.999919353563935e-05,
      "loss": 1.8746,
      "step": 1695
    },
    {
      "epoch": 0.0033299709065699743,
      "grad_norm": 2.927371025085449,
      "learning_rate": 2.9999179198906614e-05,
      "loss": 2.0249,
      "step": 1710
    },
    {
      "epoch": 0.003359181177680237,
      "grad_norm": 2.425579071044922,
      "learning_rate": 2.9999164735863685e-05,
      "loss": 1.9149,
      "step": 1725
    },
    {
      "epoch": 0.0033883914487905,
      "grad_norm": 3.0711071491241455,
      "learning_rate": 2.9999150146510678e-05,
      "loss": 1.8335,
      "step": 1740
    },
    {
      "epoch": 0.003417601719900763,
      "grad_norm": 4.509474277496338,
      "learning_rate": 2.9999135430847718e-05,
      "loss": 2.0002,
      "step": 1755
    },
    {
      "epoch": 0.003446811991011026,
      "grad_norm": 2.784403085708618,
      "learning_rate": 2.9999120588874927e-05,
      "loss": 1.985,
      "step": 1770
    },
    {
      "epoch": 0.0034760222621212887,
      "grad_norm": 4.556781768798828,
      "learning_rate": 2.9999105620592434e-05,
      "loss": 1.8832,
      "step": 1785
    },
    {
      "epoch": 0.003505232533231552,
      "grad_norm": 3.1926143169403076,
      "learning_rate": 2.9999090526000364e-05,
      "loss": 2.0136,
      "step": 1800
    },
    {
      "epoch": 0.0035344428043418146,
      "grad_norm": 3.627634286880493,
      "learning_rate": 2.9999075305098846e-05,
      "loss": 2.0413,
      "step": 1815
    },
    {
      "epoch": 0.0035636530754520777,
      "grad_norm": 4.250141620635986,
      "learning_rate": 2.9999059957887998e-05,
      "loss": 1.9076,
      "step": 1830
    },
    {
      "epoch": 0.0035928633465623404,
      "grad_norm": 4.263555526733398,
      "learning_rate": 2.999904448436796e-05,
      "loss": 1.9782,
      "step": 1845
    },
    {
      "epoch": 0.0036220736176726036,
      "grad_norm": 2.5871047973632812,
      "learning_rate": 2.999902888453886e-05,
      "loss": 1.9703,
      "step": 1860
    },
    {
      "epoch": 0.0036512838887828663,
      "grad_norm": 2.2850048542022705,
      "learning_rate": 2.9999013158400827e-05,
      "loss": 1.9571,
      "step": 1875
    },
    {
      "epoch": 0.0036804941598931294,
      "grad_norm": 2.419032096862793,
      "learning_rate": 2.9998997305953993e-05,
      "loss": 2.1583,
      "step": 1890
    },
    {
      "epoch": 0.003709704431003392,
      "grad_norm": 3.190084934234619,
      "learning_rate": 2.9998981327198497e-05,
      "loss": 1.9732,
      "step": 1905
    },
    {
      "epoch": 0.0037389147021136553,
      "grad_norm": 3.31540846824646,
      "learning_rate": 2.9998965222134468e-05,
      "loss": 1.8646,
      "step": 1920
    },
    {
      "epoch": 0.003768124973223918,
      "grad_norm": 2.423227310180664,
      "learning_rate": 2.9998948990762044e-05,
      "loss": 1.9282,
      "step": 1935
    },
    {
      "epoch": 0.003797335244334181,
      "grad_norm": 3.4635472297668457,
      "learning_rate": 2.999893263308136e-05,
      "loss": 1.9859,
      "step": 1950
    },
    {
      "epoch": 0.003826545515444444,
      "grad_norm": 2.474353790283203,
      "learning_rate": 2.9998916149092556e-05,
      "loss": 2.0055,
      "step": 1965
    },
    {
      "epoch": 0.003855755786554707,
      "grad_norm": 3.264435052871704,
      "learning_rate": 2.9998899538795766e-05,
      "loss": 1.9755,
      "step": 1980
    },
    {
      "epoch": 0.0038849660576649697,
      "grad_norm": 2.028916358947754,
      "learning_rate": 2.9998882802191137e-05,
      "loss": 1.9914,
      "step": 1995
    },
    {
      "epoch": 0.003914176328775233,
      "grad_norm": 1.9189203977584839,
      "learning_rate": 2.9998865939278805e-05,
      "loss": 1.8109,
      "step": 2010
    },
    {
      "epoch": 0.0039433865998854956,
      "grad_norm": 3.143857955932617,
      "learning_rate": 2.9998848950058913e-05,
      "loss": 1.8233,
      "step": 2025
    },
    {
      "epoch": 0.003972596870995758,
      "grad_norm": 2.662853479385376,
      "learning_rate": 2.9998831834531608e-05,
      "loss": 2.0015,
      "step": 2040
    },
    {
      "epoch": 0.004001807142106022,
      "grad_norm": 2.0305709838867188,
      "learning_rate": 2.9998814592697027e-05,
      "loss": 1.8817,
      "step": 2055
    },
    {
      "epoch": 0.0040310174132162846,
      "grad_norm": 4.386982440948486,
      "learning_rate": 2.999879722455532e-05,
      "loss": 1.8511,
      "step": 2070
    },
    {
      "epoch": 0.004060227684326547,
      "grad_norm": 3.3638317584991455,
      "learning_rate": 2.9998779730106633e-05,
      "loss": 1.9738,
      "step": 2085
    },
    {
      "epoch": 0.00408943795543681,
      "grad_norm": 3.6872494220733643,
      "learning_rate": 2.9998762109351107e-05,
      "loss": 1.8496,
      "step": 2100
    },
    {
      "epoch": 0.0041186482265470736,
      "grad_norm": 3.2654178142547607,
      "learning_rate": 2.9998744362288902e-05,
      "loss": 1.9003,
      "step": 2115
    },
    {
      "epoch": 0.004147858497657336,
      "grad_norm": 3.4839179515838623,
      "learning_rate": 2.9998726488920162e-05,
      "loss": 2.016,
      "step": 2130
    },
    {
      "epoch": 0.004177068768767599,
      "grad_norm": 2.537320613861084,
      "learning_rate": 2.9998708489245034e-05,
      "loss": 2.0073,
      "step": 2145
    },
    {
      "epoch": 0.004206279039877862,
      "grad_norm": 3.2071378231048584,
      "learning_rate": 2.999869036326367e-05,
      "loss": 1.9835,
      "step": 2160
    },
    {
      "epoch": 0.004235489310988125,
      "grad_norm": 2.7338645458221436,
      "learning_rate": 2.999867211097623e-05,
      "loss": 1.8995,
      "step": 2175
    },
    {
      "epoch": 0.004264699582098388,
      "grad_norm": 4.866488933563232,
      "learning_rate": 2.999865373238286e-05,
      "loss": 1.9656,
      "step": 2190
    },
    {
      "epoch": 0.004293909853208651,
      "grad_norm": 2.67726731300354,
      "learning_rate": 2.9998635227483715e-05,
      "loss": 2.0401,
      "step": 2205
    },
    {
      "epoch": 0.004323120124318914,
      "grad_norm": 3.1271047592163086,
      "learning_rate": 2.9998616596278955e-05,
      "loss": 2.0406,
      "step": 2220
    },
    {
      "epoch": 0.004352330395429177,
      "grad_norm": 5.673630237579346,
      "learning_rate": 2.9998597838768738e-05,
      "loss": 1.9733,
      "step": 2235
    },
    {
      "epoch": 0.00438154066653944,
      "grad_norm": 2.696476697921753,
      "learning_rate": 2.9998578954953216e-05,
      "loss": 2.16,
      "step": 2250
    },
    {
      "epoch": 0.004410750937649702,
      "grad_norm": 5.409951686859131,
      "learning_rate": 2.9998559944832553e-05,
      "loss": 1.9401,
      "step": 2265
    },
    {
      "epoch": 0.004439961208759966,
      "grad_norm": 2.183530330657959,
      "learning_rate": 2.9998540808406903e-05,
      "loss": 1.9405,
      "step": 2280
    },
    {
      "epoch": 0.004469171479870229,
      "grad_norm": 3.3076109886169434,
      "learning_rate": 2.9998521545676438e-05,
      "loss": 1.9617,
      "step": 2295
    },
    {
      "epoch": 0.004498381750980491,
      "grad_norm": 2.793837547302246,
      "learning_rate": 2.999850215664131e-05,
      "loss": 1.9463,
      "step": 2310
    },
    {
      "epoch": 0.004527592022090754,
      "grad_norm": 2.6732029914855957,
      "learning_rate": 2.9998482641301687e-05,
      "loss": 1.9114,
      "step": 2325
    },
    {
      "epoch": 0.004556802293201018,
      "grad_norm": 2.1354994773864746,
      "learning_rate": 2.999846299965773e-05,
      "loss": 2.0321,
      "step": 2340
    },
    {
      "epoch": 0.00458601256431128,
      "grad_norm": 3.889298439025879,
      "learning_rate": 2.9998443231709608e-05,
      "loss": 2.0722,
      "step": 2355
    },
    {
      "epoch": 0.004615222835421543,
      "grad_norm": 3.6255850791931152,
      "learning_rate": 2.9998423337457486e-05,
      "loss": 2.0355,
      "step": 2370
    },
    {
      "epoch": 0.004644433106531806,
      "grad_norm": 2.4627747535705566,
      "learning_rate": 2.9998403316901533e-05,
      "loss": 2.0076,
      "step": 2385
    },
    {
      "epoch": 0.004673643377642069,
      "grad_norm": 3.3934834003448486,
      "learning_rate": 2.9998383170041916e-05,
      "loss": 1.9877,
      "step": 2400
    },
    {
      "epoch": 0.004702853648752332,
      "grad_norm": 2.2475860118865967,
      "learning_rate": 2.99983628968788e-05,
      "loss": 1.8463,
      "step": 2415
    },
    {
      "epoch": 0.004732063919862595,
      "grad_norm": 2.8804898262023926,
      "learning_rate": 2.9998342497412365e-05,
      "loss": 2.0943,
      "step": 2430
    },
    {
      "epoch": 0.0047612741909728576,
      "grad_norm": 2.2566545009613037,
      "learning_rate": 2.999832197164278e-05,
      "loss": 1.8834,
      "step": 2445
    },
    {
      "epoch": 0.004790484462083121,
      "grad_norm": 2.948420524597168,
      "learning_rate": 2.9998301319570216e-05,
      "loss": 2.002,
      "step": 2460
    },
    {
      "epoch": 0.004819694733193384,
      "grad_norm": 3.6838879585266113,
      "learning_rate": 2.999828054119484e-05,
      "loss": 1.9071,
      "step": 2475
    },
    {
      "epoch": 0.0048489050043036466,
      "grad_norm": 2.2608137130737305,
      "learning_rate": 2.9998259636516845e-05,
      "loss": 1.9241,
      "step": 2490
    },
    {
      "epoch": 0.004878115275413909,
      "grad_norm": 2.5766665935516357,
      "learning_rate": 2.999823860553639e-05,
      "loss": 1.9269,
      "step": 2505
    },
    {
      "epoch": 0.004907325546524173,
      "grad_norm": 3.382236957550049,
      "learning_rate": 2.9998217448253658e-05,
      "loss": 2.0165,
      "step": 2520
    },
    {
      "epoch": 0.0049365358176344356,
      "grad_norm": 2.600278854370117,
      "learning_rate": 2.999819616466883e-05,
      "loss": 1.8852,
      "step": 2535
    },
    {
      "epoch": 0.004965746088744698,
      "grad_norm": 2.8558318614959717,
      "learning_rate": 2.999817475478208e-05,
      "loss": 1.9202,
      "step": 2550
    },
    {
      "epoch": 0.004994956359854961,
      "grad_norm": 2.5992493629455566,
      "learning_rate": 2.9998153218593594e-05,
      "loss": 2.0815,
      "step": 2565
    },
    {
      "epoch": 0.0050241666309652246,
      "grad_norm": 3.0813794136047363,
      "learning_rate": 2.9998131556103545e-05,
      "loss": 2.0847,
      "step": 2580
    },
    {
      "epoch": 0.005053376902075487,
      "grad_norm": 3.12187123298645,
      "learning_rate": 2.999810976731213e-05,
      "loss": 1.8592,
      "step": 2595
    },
    {
      "epoch": 0.00508258717318575,
      "grad_norm": 3.4060111045837402,
      "learning_rate": 2.9998087852219514e-05,
      "loss": 1.8904,
      "step": 2610
    },
    {
      "epoch": 0.005111797444296013,
      "grad_norm": 1.8444907665252686,
      "learning_rate": 2.9998065810825895e-05,
      "loss": 2.0493,
      "step": 2625
    },
    {
      "epoch": 0.005141007715406276,
      "grad_norm": 2.9746735095977783,
      "learning_rate": 2.999804364313145e-05,
      "loss": 2.0021,
      "step": 2640
    },
    {
      "epoch": 0.005170217986516539,
      "grad_norm": 2.150517463684082,
      "learning_rate": 2.9998021349136373e-05,
      "loss": 1.8938,
      "step": 2655
    },
    {
      "epoch": 0.005199428257626802,
      "grad_norm": 2.4695417881011963,
      "learning_rate": 2.9997998928840854e-05,
      "loss": 1.956,
      "step": 2670
    },
    {
      "epoch": 0.005228638528737064,
      "grad_norm": 3.365466356277466,
      "learning_rate": 2.999797638224507e-05,
      "loss": 1.8762,
      "step": 2685
    },
    {
      "epoch": 0.005257848799847328,
      "grad_norm": 2.206486940383911,
      "learning_rate": 2.999795370934922e-05,
      "loss": 2.0777,
      "step": 2700
    },
    {
      "epoch": 0.005287059070957591,
      "grad_norm": 2.5338962078094482,
      "learning_rate": 2.9997930910153492e-05,
      "loss": 1.9524,
      "step": 2715
    },
    {
      "epoch": 0.005316269342067853,
      "grad_norm": 2.7835092544555664,
      "learning_rate": 2.999790798465808e-05,
      "loss": 1.9849,
      "step": 2730
    },
    {
      "epoch": 0.005345479613178116,
      "grad_norm": 3.5604777336120605,
      "learning_rate": 2.999788493286317e-05,
      "loss": 1.9226,
      "step": 2745
    },
    {
      "epoch": 0.00537468988428838,
      "grad_norm": 2.7719836235046387,
      "learning_rate": 2.9997861754768965e-05,
      "loss": 2.0174,
      "step": 2760
    },
    {
      "epoch": 0.005403900155398642,
      "grad_norm": 2.0660643577575684,
      "learning_rate": 2.999783845037566e-05,
      "loss": 2.0959,
      "step": 2775
    },
    {
      "epoch": 0.005433110426508905,
      "grad_norm": 2.501246690750122,
      "learning_rate": 2.9997815019683443e-05,
      "loss": 1.9407,
      "step": 2790
    },
    {
      "epoch": 0.005462320697619168,
      "grad_norm": 3.6545281410217285,
      "learning_rate": 2.9997791462692518e-05,
      "loss": 2.0155,
      "step": 2805
    },
    {
      "epoch": 0.005491530968729431,
      "grad_norm": 2.2655766010284424,
      "learning_rate": 2.9997767779403085e-05,
      "loss": 1.7096,
      "step": 2820
    },
    {
      "epoch": 0.005520741239839694,
      "grad_norm": 1.9551692008972168,
      "learning_rate": 2.9997743969815337e-05,
      "loss": 2.0597,
      "step": 2835
    },
    {
      "epoch": 0.005549951510949957,
      "grad_norm": 4.887934684753418,
      "learning_rate": 2.999772003392948e-05,
      "loss": 2.0615,
      "step": 2850
    },
    {
      "epoch": 0.0055791617820602196,
      "grad_norm": 3.462581157684326,
      "learning_rate": 2.999769597174571e-05,
      "loss": 1.9574,
      "step": 2865
    },
    {
      "epoch": 0.005608372053170483,
      "grad_norm": 4.445065498352051,
      "learning_rate": 2.9997671783264234e-05,
      "loss": 1.9514,
      "step": 2880
    },
    {
      "epoch": 0.005637582324280746,
      "grad_norm": 2.81374454498291,
      "learning_rate": 2.9997647468485254e-05,
      "loss": 1.9163,
      "step": 2895
    },
    {
      "epoch": 0.0056667925953910086,
      "grad_norm": 3.4871456623077393,
      "learning_rate": 2.999762302740898e-05,
      "loss": 1.966,
      "step": 2910
    },
    {
      "epoch": 0.005696002866501271,
      "grad_norm": 2.006155490875244,
      "learning_rate": 2.9997598460035608e-05,
      "loss": 2.009,
      "step": 2925
    },
    {
      "epoch": 0.005725213137611535,
      "grad_norm": 2.104846477508545,
      "learning_rate": 2.9997573766365353e-05,
      "loss": 1.9507,
      "step": 2940
    },
    {
      "epoch": 0.0057544234087217976,
      "grad_norm": 3.3199880123138428,
      "learning_rate": 2.999754894639842e-05,
      "loss": 1.9102,
      "step": 2955
    },
    {
      "epoch": 0.00578363367983206,
      "grad_norm": 3.4421074390411377,
      "learning_rate": 2.9997524000135015e-05,
      "loss": 1.9148,
      "step": 2970
    },
    {
      "epoch": 0.005812843950942323,
      "grad_norm": 3.9583024978637695,
      "learning_rate": 2.9997498927575352e-05,
      "loss": 1.9872,
      "step": 2985
    },
    {
      "epoch": 0.0058420542220525866,
      "grad_norm": 2.90704607963562,
      "learning_rate": 2.999747372871964e-05,
      "loss": 1.8581,
      "step": 3000
    },
    {
      "epoch": 0.005871264493162849,
      "grad_norm": 2.327897548675537,
      "learning_rate": 2.99974484035681e-05,
      "loss": 2.155,
      "step": 3015
    },
    {
      "epoch": 0.005900474764273112,
      "grad_norm": 2.7539587020874023,
      "learning_rate": 2.999742295212093e-05,
      "loss": 2.0412,
      "step": 3030
    },
    {
      "epoch": 0.005929685035383375,
      "grad_norm": 3.2530441284179688,
      "learning_rate": 2.999739737437835e-05,
      "loss": 1.8964,
      "step": 3045
    },
    {
      "epoch": 0.005958895306493638,
      "grad_norm": 4.038791179656982,
      "learning_rate": 2.9997371670340583e-05,
      "loss": 1.8992,
      "step": 3060
    },
    {
      "epoch": 0.005988105577603901,
      "grad_norm": 3.810509443283081,
      "learning_rate": 2.999734584000784e-05,
      "loss": 2.0157,
      "step": 3075
    },
    {
      "epoch": 0.006017315848714164,
      "grad_norm": 2.6701083183288574,
      "learning_rate": 2.9997319883380334e-05,
      "loss": 1.9206,
      "step": 3090
    },
    {
      "epoch": 0.006046526119824426,
      "grad_norm": 2.0933597087860107,
      "learning_rate": 2.999729380045829e-05,
      "loss": 1.8986,
      "step": 3105
    },
    {
      "epoch": 0.00607573639093469,
      "grad_norm": 7.184755802154541,
      "learning_rate": 2.9997267591241924e-05,
      "loss": 2.0928,
      "step": 3120
    },
    {
      "epoch": 0.006104946662044953,
      "grad_norm": 4.04668664932251,
      "learning_rate": 2.9997241255731465e-05,
      "loss": 2.1005,
      "step": 3135
    },
    {
      "epoch": 0.006134156933155215,
      "grad_norm": 2.4879837036132812,
      "learning_rate": 2.9997214793927122e-05,
      "loss": 2.0177,
      "step": 3150
    },
    {
      "epoch": 0.006163367204265478,
      "grad_norm": 2.1891894340515137,
      "learning_rate": 2.9997188205829127e-05,
      "loss": 1.827,
      "step": 3165
    },
    {
      "epoch": 0.006192577475375742,
      "grad_norm": 2.6386845111846924,
      "learning_rate": 2.9997161491437696e-05,
      "loss": 2.0118,
      "step": 3180
    },
    {
      "epoch": 0.006221787746486004,
      "grad_norm": 2.792214870452881,
      "learning_rate": 2.9997134650753066e-05,
      "loss": 1.7433,
      "step": 3195
    },
    {
      "epoch": 0.006250998017596267,
      "grad_norm": 3.470515727996826,
      "learning_rate": 2.999710768377545e-05,
      "loss": 2.0895,
      "step": 3210
    },
    {
      "epoch": 0.00628020828870653,
      "grad_norm": 3.3043479919433594,
      "learning_rate": 2.9997080590505085e-05,
      "loss": 2.0023,
      "step": 3225
    },
    {
      "epoch": 0.006309418559816793,
      "grad_norm": 2.1124866008758545,
      "learning_rate": 2.9997053370942195e-05,
      "loss": 2.0457,
      "step": 3240
    },
    {
      "epoch": 0.006338628830927056,
      "grad_norm": 3.1867144107818604,
      "learning_rate": 2.9997026025087004e-05,
      "loss": 1.9503,
      "step": 3255
    },
    {
      "epoch": 0.006367839102037319,
      "grad_norm": 2.3057665824890137,
      "learning_rate": 2.999699855293975e-05,
      "loss": 1.8968,
      "step": 3270
    },
    {
      "epoch": 0.0063970493731475815,
      "grad_norm": 2.416084051132202,
      "learning_rate": 2.999697095450066e-05,
      "loss": 1.8304,
      "step": 3285
    },
    {
      "epoch": 0.006426259644257845,
      "grad_norm": 2.481602191925049,
      "learning_rate": 2.9996943229769977e-05,
      "loss": 2.1386,
      "step": 3300
    },
    {
      "epoch": 0.006455469915368108,
      "grad_norm": 2.4862520694732666,
      "learning_rate": 2.9996915378747918e-05,
      "loss": 1.9672,
      "step": 3315
    },
    {
      "epoch": 0.0064846801864783706,
      "grad_norm": 3.704164743423462,
      "learning_rate": 2.999688740143473e-05,
      "loss": 1.9671,
      "step": 3330
    },
    {
      "epoch": 0.006513890457588634,
      "grad_norm": 4.340814113616943,
      "learning_rate": 2.999685929783064e-05,
      "loss": 1.889,
      "step": 3345
    },
    {
      "epoch": 0.006543100728698897,
      "grad_norm": 2.8202598094940186,
      "learning_rate": 2.999683106793589e-05,
      "loss": 1.883,
      "step": 3360
    },
    {
      "epoch": 0.0065723109998091596,
      "grad_norm": 2.3717639446258545,
      "learning_rate": 2.9996802711750716e-05,
      "loss": 1.8932,
      "step": 3375
    },
    {
      "epoch": 0.006601521270919422,
      "grad_norm": 2.1720776557922363,
      "learning_rate": 2.999677422927536e-05,
      "loss": 1.8262,
      "step": 3390
    },
    {
      "epoch": 0.006630731542029686,
      "grad_norm": 6.15664005279541,
      "learning_rate": 2.9996745620510055e-05,
      "loss": 1.9139,
      "step": 3405
    },
    {
      "epoch": 0.0066599418131399486,
      "grad_norm": 3.080832004547119,
      "learning_rate": 2.9996716885455047e-05,
      "loss": 1.9838,
      "step": 3420
    },
    {
      "epoch": 0.006689152084250211,
      "grad_norm": 2.1853530406951904,
      "learning_rate": 2.9996688024110577e-05,
      "loss": 1.9353,
      "step": 3435
    },
    {
      "epoch": 0.006718362355360474,
      "grad_norm": 2.609475612640381,
      "learning_rate": 2.9996659036476886e-05,
      "loss": 1.9734,
      "step": 3450
    },
    {
      "epoch": 0.0067475726264707376,
      "grad_norm": 3.808016061782837,
      "learning_rate": 2.9996629922554225e-05,
      "loss": 1.9872,
      "step": 3465
    },
    {
      "epoch": 0.006776782897581,
      "grad_norm": 4.373661994934082,
      "learning_rate": 2.999660068234283e-05,
      "loss": 1.8084,
      "step": 3480
    },
    {
      "epoch": 0.006805993168691263,
      "grad_norm": 4.299337387084961,
      "learning_rate": 2.9996571315842954e-05,
      "loss": 2.0801,
      "step": 3495
    },
    {
      "epoch": 0.006835203439801526,
      "grad_norm": 4.49350118637085,
      "learning_rate": 2.999654182305484e-05,
      "loss": 1.8882,
      "step": 3510
    },
    {
      "epoch": 0.006864413710911789,
      "grad_norm": 2.319342613220215,
      "learning_rate": 2.999651220397874e-05,
      "loss": 1.9627,
      "step": 3525
    },
    {
      "epoch": 0.006893623982022052,
      "grad_norm": 3.2572271823883057,
      "learning_rate": 2.99964824586149e-05,
      "loss": 1.9482,
      "step": 3540
    },
    {
      "epoch": 0.006922834253132315,
      "grad_norm": 1.7764968872070312,
      "learning_rate": 2.9996452586963575e-05,
      "loss": 1.8938,
      "step": 3555
    },
    {
      "epoch": 0.006952044524242577,
      "grad_norm": 2.2628538608551025,
      "learning_rate": 2.9996422589025007e-05,
      "loss": 1.9644,
      "step": 3570
    },
    {
      "epoch": 0.006981254795352841,
      "grad_norm": 3.8122570514678955,
      "learning_rate": 2.999639246479946e-05,
      "loss": 1.7568,
      "step": 3585
    },
    {
      "epoch": 0.007010465066463104,
      "grad_norm": 3.868978977203369,
      "learning_rate": 2.999636221428718e-05,
      "loss": 2.077,
      "step": 3600
    },
    {
      "epoch": 0.007039675337573366,
      "grad_norm": 4.426783561706543,
      "learning_rate": 2.999633183748843e-05,
      "loss": 1.8161,
      "step": 3615
    },
    {
      "epoch": 0.007068885608683629,
      "grad_norm": 2.3112306594848633,
      "learning_rate": 2.9996301334403456e-05,
      "loss": 1.8738,
      "step": 3630
    },
    {
      "epoch": 0.007098095879793893,
      "grad_norm": 1.7659751176834106,
      "learning_rate": 2.9996270705032523e-05,
      "loss": 1.9714,
      "step": 3645
    },
    {
      "epoch": 0.007127306150904155,
      "grad_norm": 5.0493364334106445,
      "learning_rate": 2.9996239949375882e-05,
      "loss": 1.9451,
      "step": 3660
    },
    {
      "epoch": 0.007156516422014418,
      "grad_norm": 1.6928982734680176,
      "learning_rate": 2.9996209067433794e-05,
      "loss": 1.7872,
      "step": 3675
    },
    {
      "epoch": 0.007185726693124681,
      "grad_norm": 2.2954142093658447,
      "learning_rate": 2.9996178059206525e-05,
      "loss": 2.0124,
      "step": 3690
    },
    {
      "epoch": 0.007214936964234944,
      "grad_norm": 4.649162292480469,
      "learning_rate": 2.9996146924694327e-05,
      "loss": 1.823,
      "step": 3705
    },
    {
      "epoch": 0.007244147235345207,
      "grad_norm": 2.248623847961426,
      "learning_rate": 2.9996115663897468e-05,
      "loss": 1.9382,
      "step": 3720
    },
    {
      "epoch": 0.00727335750645547,
      "grad_norm": 3.033177375793457,
      "learning_rate": 2.999608427681621e-05,
      "loss": 2.0324,
      "step": 3735
    },
    {
      "epoch": 0.0073025677775657325,
      "grad_norm": 3.4453887939453125,
      "learning_rate": 2.9996052763450817e-05,
      "loss": 1.9293,
      "step": 3750
    },
    {
      "epoch": 0.007331778048675996,
      "grad_norm": 3.873504877090454,
      "learning_rate": 2.9996021123801556e-05,
      "loss": 2.0132,
      "step": 3765
    },
    {
      "epoch": 0.007360988319786259,
      "grad_norm": 3.7370665073394775,
      "learning_rate": 2.999598935786869e-05,
      "loss": 1.8681,
      "step": 3780
    },
    {
      "epoch": 0.0073901985908965215,
      "grad_norm": 5.5719523429870605,
      "learning_rate": 2.999595746565249e-05,
      "loss": 1.8799,
      "step": 3795
    },
    {
      "epoch": 0.007419408862006784,
      "grad_norm": 3.7253758907318115,
      "learning_rate": 2.9995925447153226e-05,
      "loss": 1.8633,
      "step": 3810
    },
    {
      "epoch": 0.007448619133117048,
      "grad_norm": 2.792862892150879,
      "learning_rate": 2.9995893302371158e-05,
      "loss": 1.9642,
      "step": 3825
    },
    {
      "epoch": 0.0074778294042273106,
      "grad_norm": 2.693080425262451,
      "learning_rate": 2.999586103130657e-05,
      "loss": 1.8963,
      "step": 3840
    },
    {
      "epoch": 0.007507039675337573,
      "grad_norm": 2.7681262493133545,
      "learning_rate": 2.9995828633959724e-05,
      "loss": 2.0581,
      "step": 3855
    },
    {
      "epoch": 0.007536249946447836,
      "grad_norm": 3.635828733444214,
      "learning_rate": 2.9995796110330894e-05,
      "loss": 1.7659,
      "step": 3870
    },
    {
      "epoch": 0.0075654602175580996,
      "grad_norm": 2.46408748626709,
      "learning_rate": 2.9995763460420358e-05,
      "loss": 1.7966,
      "step": 3885
    },
    {
      "epoch": 0.007594670488668362,
      "grad_norm": 2.2896721363067627,
      "learning_rate": 2.999573068422839e-05,
      "loss": 1.9918,
      "step": 3900
    },
    {
      "epoch": 0.007623880759778625,
      "grad_norm": 2.3533968925476074,
      "learning_rate": 2.9995697781755262e-05,
      "loss": 1.8725,
      "step": 3915
    },
    {
      "epoch": 0.007653091030888888,
      "grad_norm": 2.5543251037597656,
      "learning_rate": 2.999566475300125e-05,
      "loss": 1.8896,
      "step": 3930
    },
    {
      "epoch": 0.007682301301999151,
      "grad_norm": 3.2201671600341797,
      "learning_rate": 2.999563159796665e-05,
      "loss": 1.8444,
      "step": 3945
    },
    {
      "epoch": 0.007711511573109414,
      "grad_norm": 2.2435178756713867,
      "learning_rate": 2.9995598316651713e-05,
      "loss": 1.976,
      "step": 3960
    },
    {
      "epoch": 0.007740721844219677,
      "grad_norm": 3.219825506210327,
      "learning_rate": 2.999556490905674e-05,
      "loss": 1.8561,
      "step": 3975
    },
    {
      "epoch": 0.007769932115329939,
      "grad_norm": 2.4510769844055176,
      "learning_rate": 2.9995531375182008e-05,
      "loss": 2.068,
      "step": 3990
    },
    {
      "epoch": 0.007799142386440203,
      "grad_norm": 4.242166519165039,
      "learning_rate": 2.999549771502779e-05,
      "loss": 1.8497,
      "step": 4005
    },
    {
      "epoch": 0.007828352657550466,
      "grad_norm": 2.9145870208740234,
      "learning_rate": 2.9995463928594383e-05,
      "loss": 1.8896,
      "step": 4020
    },
    {
      "epoch": 0.007857562928660728,
      "grad_norm": 2.33396315574646,
      "learning_rate": 2.9995430015882064e-05,
      "loss": 1.9502,
      "step": 4035
    },
    {
      "epoch": 0.007886773199770991,
      "grad_norm": 3.1073617935180664,
      "learning_rate": 2.9995395976891118e-05,
      "loss": 1.9178,
      "step": 4050
    },
    {
      "epoch": 0.007915983470881254,
      "grad_norm": 3.444310188293457,
      "learning_rate": 2.9995361811621838e-05,
      "loss": 1.7477,
      "step": 4065
    },
    {
      "epoch": 0.007945193741991517,
      "grad_norm": 3.131169080734253,
      "learning_rate": 2.9995327520074504e-05,
      "loss": 1.8799,
      "step": 4080
    },
    {
      "epoch": 0.007974404013101781,
      "grad_norm": 2.9107017517089844,
      "learning_rate": 2.9995293102249408e-05,
      "loss": 1.8785,
      "step": 4095
    },
    {
      "epoch": 0.008003614284212044,
      "grad_norm": 2.795280694961548,
      "learning_rate": 2.9995258558146834e-05,
      "loss": 2.0386,
      "step": 4110
    },
    {
      "epoch": 0.008032824555322306,
      "grad_norm": 5.588748455047607,
      "learning_rate": 2.9995223887767087e-05,
      "loss": 1.9852,
      "step": 4125
    },
    {
      "epoch": 0.008062034826432569,
      "grad_norm": 2.659045934677124,
      "learning_rate": 2.999518909111045e-05,
      "loss": 1.8707,
      "step": 4140
    },
    {
      "epoch": 0.008091245097542832,
      "grad_norm": 2.9934346675872803,
      "learning_rate": 2.9995154168177214e-05,
      "loss": 1.9004,
      "step": 4155
    },
    {
      "epoch": 0.008120455368653095,
      "grad_norm": 2.6598739624023438,
      "learning_rate": 2.9995119118967674e-05,
      "loss": 1.8897,
      "step": 4170
    },
    {
      "epoch": 0.008149665639763357,
      "grad_norm": 3.6846165657043457,
      "learning_rate": 2.9995083943482126e-05,
      "loss": 1.9932,
      "step": 4185
    },
    {
      "epoch": 0.00817887591087362,
      "grad_norm": 3.6916184425354004,
      "learning_rate": 2.9995048641720873e-05,
      "loss": 1.7672,
      "step": 4200
    },
    {
      "epoch": 0.008208086181983884,
      "grad_norm": 3.650599956512451,
      "learning_rate": 2.9995013213684202e-05,
      "loss": 1.9271,
      "step": 4215
    },
    {
      "epoch": 0.008237296453094147,
      "grad_norm": 4.196152687072754,
      "learning_rate": 2.999497765937242e-05,
      "loss": 1.8244,
      "step": 4230
    },
    {
      "epoch": 0.00826650672420441,
      "grad_norm": 3.1078038215637207,
      "learning_rate": 2.9994941978785817e-05,
      "loss": 1.893,
      "step": 4245
    },
    {
      "epoch": 0.008295716995314673,
      "grad_norm": 4.401791095733643,
      "learning_rate": 2.9994906171924703e-05,
      "loss": 1.8844,
      "step": 4260
    },
    {
      "epoch": 0.008324927266424935,
      "grad_norm": 2.6136245727539062,
      "learning_rate": 2.999487023878937e-05,
      "loss": 1.8981,
      "step": 4275
    },
    {
      "epoch": 0.008354137537535198,
      "grad_norm": 3.334519624710083,
      "learning_rate": 2.9994834179380134e-05,
      "loss": 1.9983,
      "step": 4290
    },
    {
      "epoch": 0.00838334780864546,
      "grad_norm": 1.8010269403457642,
      "learning_rate": 2.9994797993697283e-05,
      "loss": 1.8192,
      "step": 4305
    },
    {
      "epoch": 0.008412558079755723,
      "grad_norm": 3.2746548652648926,
      "learning_rate": 2.9994761681741135e-05,
      "loss": 1.9696,
      "step": 4320
    },
    {
      "epoch": 0.008441768350865988,
      "grad_norm": 2.172431468963623,
      "learning_rate": 2.9994725243511982e-05,
      "loss": 1.9068,
      "step": 4335
    },
    {
      "epoch": 0.00847097862197625,
      "grad_norm": 3.5535871982574463,
      "learning_rate": 2.999468867901015e-05,
      "loss": 1.985,
      "step": 4350
    },
    {
      "epoch": 0.008500188893086513,
      "grad_norm": 3.866422176361084,
      "learning_rate": 2.9994651988235923e-05,
      "loss": 1.8057,
      "step": 4365
    },
    {
      "epoch": 0.008529399164196776,
      "grad_norm": 4.770716190338135,
      "learning_rate": 2.999461517118963e-05,
      "loss": 1.9949,
      "step": 4380
    },
    {
      "epoch": 0.008558609435307039,
      "grad_norm": 2.9273180961608887,
      "learning_rate": 2.999457822787157e-05,
      "loss": 1.8686,
      "step": 4395
    },
    {
      "epoch": 0.008587819706417301,
      "grad_norm": 3.192166328430176,
      "learning_rate": 2.9994541158282063e-05,
      "loss": 1.9539,
      "step": 4410
    },
    {
      "epoch": 0.008617029977527564,
      "grad_norm": 3.507930040359497,
      "learning_rate": 2.9994503962421417e-05,
      "loss": 1.7732,
      "step": 4425
    },
    {
      "epoch": 0.008646240248637829,
      "grad_norm": 2.666705369949341,
      "learning_rate": 2.9994466640289938e-05,
      "loss": 1.9615,
      "step": 4440
    },
    {
      "epoch": 0.008675450519748091,
      "grad_norm": 2.640362501144409,
      "learning_rate": 2.999442919188795e-05,
      "loss": 1.89,
      "step": 4455
    },
    {
      "epoch": 0.008704660790858354,
      "grad_norm": 2.524216890335083,
      "learning_rate": 2.9994391617215765e-05,
      "loss": 2.0473,
      "step": 4470
    },
    {
      "epoch": 0.008733871061968617,
      "grad_norm": 2.6444146633148193,
      "learning_rate": 2.9994353916273696e-05,
      "loss": 1.8889,
      "step": 4485
    },
    {
      "epoch": 0.00876308133307888,
      "grad_norm": 2.4632787704467773,
      "learning_rate": 2.9994316089062068e-05,
      "loss": 1.9616,
      "step": 4500
    },
    {
      "epoch": 0.008792291604189142,
      "grad_norm": 5.182243347167969,
      "learning_rate": 2.999427813558119e-05,
      "loss": 1.8216,
      "step": 4515
    },
    {
      "epoch": 0.008821501875299405,
      "grad_norm": 3.6622776985168457,
      "learning_rate": 2.9994240055831395e-05,
      "loss": 1.8692,
      "step": 4530
    },
    {
      "epoch": 0.008850712146409668,
      "grad_norm": 2.899912118911743,
      "learning_rate": 2.9994201849812988e-05,
      "loss": 1.9363,
      "step": 4545
    },
    {
      "epoch": 0.008879922417519932,
      "grad_norm": 1.8940974473953247,
      "learning_rate": 2.99941635175263e-05,
      "loss": 1.8404,
      "step": 4560
    },
    {
      "epoch": 0.008909132688630195,
      "grad_norm": 3.581655263900757,
      "learning_rate": 2.9994125058971657e-05,
      "loss": 1.9623,
      "step": 4575
    },
    {
      "epoch": 0.008938342959740457,
      "grad_norm": 2.6804749965667725,
      "learning_rate": 2.9994086474149375e-05,
      "loss": 2.0692,
      "step": 4590
    },
    {
      "epoch": 0.00896755323085072,
      "grad_norm": 4.302793502807617,
      "learning_rate": 2.999404776305978e-05,
      "loss": 1.9728,
      "step": 4605
    },
    {
      "epoch": 0.008996763501960983,
      "grad_norm": 2.143483877182007,
      "learning_rate": 2.9994008925703202e-05,
      "loss": 1.941,
      "step": 4620
    },
    {
      "epoch": 0.009025973773071246,
      "grad_norm": 2.2785732746124268,
      "learning_rate": 2.9993969962079964e-05,
      "loss": 1.9828,
      "step": 4635
    },
    {
      "epoch": 0.009055184044181508,
      "grad_norm": 3.918194055557251,
      "learning_rate": 2.9993930872190398e-05,
      "loss": 1.9211,
      "step": 4650
    },
    {
      "epoch": 0.009084394315291771,
      "grad_norm": 2.7863261699676514,
      "learning_rate": 2.999389165603483e-05,
      "loss": 1.8564,
      "step": 4665
    },
    {
      "epoch": 0.009113604586402035,
      "grad_norm": 2.655966281890869,
      "learning_rate": 2.9993852313613596e-05,
      "loss": 1.9327,
      "step": 4680
    },
    {
      "epoch": 0.009142814857512298,
      "grad_norm": 2.987030506134033,
      "learning_rate": 2.999381284492702e-05,
      "loss": 1.8325,
      "step": 4695
    },
    {
      "epoch": 0.00917202512862256,
      "grad_norm": 3.127544641494751,
      "learning_rate": 2.9993773249975435e-05,
      "loss": 1.7246,
      "step": 4710
    },
    {
      "epoch": 0.009201235399732824,
      "grad_norm": 2.395202398300171,
      "learning_rate": 2.999373352875918e-05,
      "loss": 1.9827,
      "step": 4725
    },
    {
      "epoch": 0.009230445670843086,
      "grad_norm": 4.163525104522705,
      "learning_rate": 2.9993693681278582e-05,
      "loss": 1.9441,
      "step": 4740
    },
    {
      "epoch": 0.009259655941953349,
      "grad_norm": 2.84067964553833,
      "learning_rate": 2.9993653707533985e-05,
      "loss": 2.0252,
      "step": 4755
    },
    {
      "epoch": 0.009288866213063612,
      "grad_norm": 1.949599266052246,
      "learning_rate": 2.9993613607525717e-05,
      "loss": 1.9497,
      "step": 4770
    },
    {
      "epoch": 0.009318076484173874,
      "grad_norm": 4.164729595184326,
      "learning_rate": 2.9993573381254124e-05,
      "loss": 2.1013,
      "step": 4785
    },
    {
      "epoch": 0.009347286755284139,
      "grad_norm": 1.7438952922821045,
      "learning_rate": 2.9993533028719537e-05,
      "loss": 1.9461,
      "step": 4800
    },
    {
      "epoch": 0.009376497026394402,
      "grad_norm": 1.9006491899490356,
      "learning_rate": 2.9993492549922302e-05,
      "loss": 1.9273,
      "step": 4815
    },
    {
      "epoch": 0.009405707297504664,
      "grad_norm": 3.693070411682129,
      "learning_rate": 2.9993451944862762e-05,
      "loss": 1.9791,
      "step": 4830
    },
    {
      "epoch": 0.009434917568614927,
      "grad_norm": 2.9572765827178955,
      "learning_rate": 2.9993411213541248e-05,
      "loss": 1.7725,
      "step": 4845
    },
    {
      "epoch": 0.00946412783972519,
      "grad_norm": 2.882349967956543,
      "learning_rate": 2.999337035595811e-05,
      "loss": 1.8094,
      "step": 4860
    },
    {
      "epoch": 0.009493338110835452,
      "grad_norm": 3.9489054679870605,
      "learning_rate": 2.9993329372113695e-05,
      "loss": 1.8678,
      "step": 4875
    },
    {
      "epoch": 0.009522548381945715,
      "grad_norm": 2.7020881175994873,
      "learning_rate": 2.999328826200834e-05,
      "loss": 2.0528,
      "step": 4890
    },
    {
      "epoch": 0.009551758653055978,
      "grad_norm": 3.121814489364624,
      "learning_rate": 2.99932470256424e-05,
      "loss": 1.9087,
      "step": 4905
    },
    {
      "epoch": 0.009580968924166242,
      "grad_norm": 2.713003158569336,
      "learning_rate": 2.9993205663016218e-05,
      "loss": 1.8721,
      "step": 4920
    },
    {
      "epoch": 0.009610179195276505,
      "grad_norm": 3.5016112327575684,
      "learning_rate": 2.9993164174130137e-05,
      "loss": 1.8573,
      "step": 4935
    },
    {
      "epoch": 0.009639389466386768,
      "grad_norm": 4.257192611694336,
      "learning_rate": 2.9993122558984516e-05,
      "loss": 1.9638,
      "step": 4950
    },
    {
      "epoch": 0.00966859973749703,
      "grad_norm": 3.4689440727233887,
      "learning_rate": 2.9993080817579702e-05,
      "loss": 1.801,
      "step": 4965
    },
    {
      "epoch": 0.009697810008607293,
      "grad_norm": 2.3255503177642822,
      "learning_rate": 2.999303894991605e-05,
      "loss": 1.6792,
      "step": 4980
    },
    {
      "epoch": 0.009727020279717556,
      "grad_norm": 2.5599734783172607,
      "learning_rate": 2.9992996955993898e-05,
      "loss": 2.0037,
      "step": 4995
    },
    {
      "epoch": 0.009756230550827819,
      "grad_norm": 2.528571605682373,
      "learning_rate": 2.9992954835813616e-05,
      "loss": 1.9778,
      "step": 5010
    },
    {
      "epoch": 0.009785440821938081,
      "grad_norm": 3.466859817504883,
      "learning_rate": 2.999291258937555e-05,
      "loss": 1.9507,
      "step": 5025
    },
    {
      "epoch": 0.009814651093048346,
      "grad_norm": 2.9515936374664307,
      "learning_rate": 2.999287021668006e-05,
      "loss": 1.9506,
      "step": 5040
    },
    {
      "epoch": 0.009843861364158608,
      "grad_norm": 2.116895914077759,
      "learning_rate": 2.99928277177275e-05,
      "loss": 2.0403,
      "step": 5055
    },
    {
      "epoch": 0.009873071635268871,
      "grad_norm": 2.0982749462127686,
      "learning_rate": 2.999278509251823e-05,
      "loss": 1.981,
      "step": 5070
    },
    {
      "epoch": 0.009902281906379134,
      "grad_norm": 2.1464314460754395,
      "learning_rate": 2.9992742341052612e-05,
      "loss": 2.0126,
      "step": 5085
    },
    {
      "epoch": 0.009931492177489397,
      "grad_norm": 5.098091125488281,
      "learning_rate": 2.9992699463330995e-05,
      "loss": 1.8511,
      "step": 5100
    },
    {
      "epoch": 0.00996070244859966,
      "grad_norm": 3.8748703002929688,
      "learning_rate": 2.999265645935375e-05,
      "loss": 1.9058,
      "step": 5115
    },
    {
      "epoch": 0.009989912719709922,
      "grad_norm": 3.9060122966766357,
      "learning_rate": 2.999261332912124e-05,
      "loss": 1.9298,
      "step": 5130
    },
    {
      "epoch": 0.010019122990820185,
      "grad_norm": 3.6169393062591553,
      "learning_rate": 2.999257007263382e-05,
      "loss": 1.6655,
      "step": 5145
    },
    {
      "epoch": 0.010048333261930449,
      "grad_norm": 2.0531461238861084,
      "learning_rate": 2.999252668989186e-05,
      "loss": 2.0188,
      "step": 5160
    },
    {
      "epoch": 0.010077543533040712,
      "grad_norm": 2.554202079772949,
      "learning_rate": 2.9992483180895725e-05,
      "loss": 1.8039,
      "step": 5175
    },
    {
      "epoch": 0.010106753804150975,
      "grad_norm": 3.2061095237731934,
      "learning_rate": 2.9992439545645778e-05,
      "loss": 1.8889,
      "step": 5190
    },
    {
      "epoch": 0.010135964075261237,
      "grad_norm": 2.8294739723205566,
      "learning_rate": 2.9992395784142395e-05,
      "loss": 1.9291,
      "step": 5205
    },
    {
      "epoch": 0.0101651743463715,
      "grad_norm": 2.8096394538879395,
      "learning_rate": 2.9992351896385932e-05,
      "loss": 1.924,
      "step": 5220
    },
    {
      "epoch": 0.010194384617481763,
      "grad_norm": 3.7793962955474854,
      "learning_rate": 2.999230788237677e-05,
      "loss": 1.933,
      "step": 5235
    },
    {
      "epoch": 0.010223594888592025,
      "grad_norm": 5.307252407073975,
      "learning_rate": 2.999226374211527e-05,
      "loss": 1.9296,
      "step": 5250
    },
    {
      "epoch": 0.010252805159702288,
      "grad_norm": 3.207782506942749,
      "learning_rate": 2.9992219475601806e-05,
      "loss": 2.0676,
      "step": 5265
    },
    {
      "epoch": 0.010282015430812553,
      "grad_norm": 2.7217934131622314,
      "learning_rate": 2.9992175082836765e-05,
      "loss": 1.8675,
      "step": 5280
    },
    {
      "epoch": 0.010311225701922815,
      "grad_norm": 3.455260753631592,
      "learning_rate": 2.9992130563820497e-05,
      "loss": 1.924,
      "step": 5295
    },
    {
      "epoch": 0.010340435973033078,
      "grad_norm": 2.638262987136841,
      "learning_rate": 2.9992085918553393e-05,
      "loss": 1.9263,
      "step": 5310
    },
    {
      "epoch": 0.01036964624414334,
      "grad_norm": 3.1709418296813965,
      "learning_rate": 2.9992041147035828e-05,
      "loss": 1.8848,
      "step": 5325
    },
    {
      "epoch": 0.010398856515253603,
      "grad_norm": 2.1629397869110107,
      "learning_rate": 2.9991996249268175e-05,
      "loss": 1.9271,
      "step": 5340
    },
    {
      "epoch": 0.010428066786363866,
      "grad_norm": 4.2046284675598145,
      "learning_rate": 2.999195122525081e-05,
      "loss": 1.9208,
      "step": 5355
    },
    {
      "epoch": 0.010457277057474129,
      "grad_norm": 4.966402053833008,
      "learning_rate": 2.9991906074984116e-05,
      "loss": 1.8059,
      "step": 5370
    },
    {
      "epoch": 0.010486487328584392,
      "grad_norm": 3.0322844982147217,
      "learning_rate": 2.9991860798468473e-05,
      "loss": 1.9608,
      "step": 5385
    },
    {
      "epoch": 0.010515697599694656,
      "grad_norm": 2.3690319061279297,
      "learning_rate": 2.9991815395704266e-05,
      "loss": 1.9424,
      "step": 5400
    },
    {
      "epoch": 0.010544907870804919,
      "grad_norm": 2.81915545463562,
      "learning_rate": 2.9991769866691865e-05,
      "loss": 1.8246,
      "step": 5415
    },
    {
      "epoch": 0.010574118141915181,
      "grad_norm": 3.064317464828491,
      "learning_rate": 2.9991724211431667e-05,
      "loss": 1.8012,
      "step": 5430
    },
    {
      "epoch": 0.010603328413025444,
      "grad_norm": 4.302711009979248,
      "learning_rate": 2.999167842992405e-05,
      "loss": 1.9098,
      "step": 5445
    },
    {
      "epoch": 0.010632538684135707,
      "grad_norm": 3.3137192726135254,
      "learning_rate": 2.9991632522169398e-05,
      "loss": 1.8922,
      "step": 5460
    },
    {
      "epoch": 0.01066174895524597,
      "grad_norm": 3.0742363929748535,
      "learning_rate": 2.9991586488168104e-05,
      "loss": 2.0403,
      "step": 5475
    },
    {
      "epoch": 0.010690959226356232,
      "grad_norm": 3.034343957901001,
      "learning_rate": 2.9991540327920547e-05,
      "loss": 1.8944,
      "step": 5490
    },
    {
      "epoch": 0.010720169497466495,
      "grad_norm": 4.131673812866211,
      "learning_rate": 2.9991494041427124e-05,
      "loss": 1.8897,
      "step": 5505
    },
    {
      "epoch": 0.01074937976857676,
      "grad_norm": 3.8295650482177734,
      "learning_rate": 2.999144762868822e-05,
      "loss": 1.8522,
      "step": 5520
    },
    {
      "epoch": 0.010778590039687022,
      "grad_norm": 4.337125301361084,
      "learning_rate": 2.999140108970423e-05,
      "loss": 1.8039,
      "step": 5535
    },
    {
      "epoch": 0.010807800310797285,
      "grad_norm": 4.047338008880615,
      "learning_rate": 2.999135442447554e-05,
      "loss": 1.964,
      "step": 5550
    },
    {
      "epoch": 0.010837010581907548,
      "grad_norm": 3.5203754901885986,
      "learning_rate": 2.9991307633002546e-05,
      "loss": 1.9198,
      "step": 5565
    },
    {
      "epoch": 0.01086622085301781,
      "grad_norm": 3.9507977962493896,
      "learning_rate": 2.9991260715285642e-05,
      "loss": 1.9079,
      "step": 5580
    },
    {
      "epoch": 0.010895431124128073,
      "grad_norm": 4.072928428649902,
      "learning_rate": 2.9991213671325223e-05,
      "loss": 2.032,
      "step": 5595
    },
    {
      "epoch": 0.010924641395238336,
      "grad_norm": 2.8438544273376465,
      "learning_rate": 2.9991166501121685e-05,
      "loss": 1.9995,
      "step": 5610
    },
    {
      "epoch": 0.0109538516663486,
      "grad_norm": 3.2885472774505615,
      "learning_rate": 2.9991119204675425e-05,
      "loss": 1.8653,
      "step": 5625
    },
    {
      "epoch": 0.010983061937458863,
      "grad_norm": 3.6044580936431885,
      "learning_rate": 2.9991071781986843e-05,
      "loss": 1.9103,
      "step": 5640
    },
    {
      "epoch": 0.011012272208569126,
      "grad_norm": 4.155179977416992,
      "learning_rate": 2.9991024233056335e-05,
      "loss": 1.8887,
      "step": 5655
    },
    {
      "epoch": 0.011041482479679388,
      "grad_norm": 4.087657451629639,
      "learning_rate": 2.9990976557884308e-05,
      "loss": 1.967,
      "step": 5670
    },
    {
      "epoch": 0.011070692750789651,
      "grad_norm": 4.724533557891846,
      "learning_rate": 2.999092875647116e-05,
      "loss": 1.8383,
      "step": 5685
    },
    {
      "epoch": 0.011099903021899914,
      "grad_norm": 7.630526065826416,
      "learning_rate": 2.9990880828817287e-05,
      "loss": 1.9142,
      "step": 5700
    },
    {
      "epoch": 0.011129113293010176,
      "grad_norm": 4.646370887756348,
      "learning_rate": 2.99908327749231e-05,
      "loss": 1.8862,
      "step": 5715
    },
    {
      "epoch": 0.011158323564120439,
      "grad_norm": 1.9703487157821655,
      "learning_rate": 2.9990784594789e-05,
      "loss": 1.8747,
      "step": 5730
    },
    {
      "epoch": 0.011187533835230704,
      "grad_norm": 2.7092244625091553,
      "learning_rate": 2.99907362884154e-05,
      "loss": 1.9593,
      "step": 5745
    },
    {
      "epoch": 0.011216744106340966,
      "grad_norm": 2.3020477294921875,
      "learning_rate": 2.9990687855802695e-05,
      "loss": 1.9351,
      "step": 5760
    },
    {
      "epoch": 0.011245954377451229,
      "grad_norm": 3.7924842834472656,
      "learning_rate": 2.9990639296951303e-05,
      "loss": 1.9888,
      "step": 5775
    },
    {
      "epoch": 0.011275164648561492,
      "grad_norm": 2.8485357761383057,
      "learning_rate": 2.9990590611861625e-05,
      "loss": 1.7921,
      "step": 5790
    },
    {
      "epoch": 0.011304374919671754,
      "grad_norm": 2.7510149478912354,
      "learning_rate": 2.999054180053408e-05,
      "loss": 1.9655,
      "step": 5805
    },
    {
      "epoch": 0.011333585190782017,
      "grad_norm": 4.522819519042969,
      "learning_rate": 2.999049286296907e-05,
      "loss": 1.9384,
      "step": 5820
    },
    {
      "epoch": 0.01136279546189228,
      "grad_norm": 2.325582265853882,
      "learning_rate": 2.9990443799167018e-05,
      "loss": 1.8342,
      "step": 5835
    },
    {
      "epoch": 0.011392005733002543,
      "grad_norm": 3.583799123764038,
      "learning_rate": 2.999039460912832e-05,
      "loss": 1.9671,
      "step": 5850
    },
    {
      "epoch": 0.011421216004112807,
      "grad_norm": 2.342571973800659,
      "learning_rate": 2.999034529285341e-05,
      "loss": 1.8169,
      "step": 5865
    },
    {
      "epoch": 0.01145042627522307,
      "grad_norm": 3.772407054901123,
      "learning_rate": 2.9990295850342694e-05,
      "loss": 1.8066,
      "step": 5880
    },
    {
      "epoch": 0.011479636546333332,
      "grad_norm": 2.550743341445923,
      "learning_rate": 2.9990246281596583e-05,
      "loss": 1.9611,
      "step": 5895
    },
    {
      "epoch": 0.011508846817443595,
      "grad_norm": 1.9916815757751465,
      "learning_rate": 2.9990196586615502e-05,
      "loss": 1.8308,
      "step": 5910
    },
    {
      "epoch": 0.011538057088553858,
      "grad_norm": 3.8781256675720215,
      "learning_rate": 2.9990146765399868e-05,
      "loss": 1.804,
      "step": 5925
    },
    {
      "epoch": 0.01156726735966412,
      "grad_norm": 1.9027587175369263,
      "learning_rate": 2.99900968179501e-05,
      "loss": 1.8449,
      "step": 5940
    },
    {
      "epoch": 0.011596477630774383,
      "grad_norm": 2.9147605895996094,
      "learning_rate": 2.9990046744266612e-05,
      "loss": 1.9283,
      "step": 5955
    },
    {
      "epoch": 0.011625687901884646,
      "grad_norm": 2.0023555755615234,
      "learning_rate": 2.9989996544349842e-05,
      "loss": 1.9464,
      "step": 5970
    },
    {
      "epoch": 0.01165489817299491,
      "grad_norm": 2.819077253341675,
      "learning_rate": 2.9989946218200195e-05,
      "loss": 1.7682,
      "step": 5985
    },
    {
      "epoch": 0.011684108444105173,
      "grad_norm": 4.221833229064941,
      "learning_rate": 2.998989576581811e-05,
      "loss": 1.6921,
      "step": 6000
    },
    {
      "epoch": 0.011713318715215436,
      "grad_norm": 2.9984993934631348,
      "learning_rate": 2.9989845187204e-05,
      "loss": 1.8921,
      "step": 6015
    },
    {
      "epoch": 0.011742528986325699,
      "grad_norm": 2.866452932357788,
      "learning_rate": 2.9989794482358293e-05,
      "loss": 1.9759,
      "step": 6030
    },
    {
      "epoch": 0.011771739257435961,
      "grad_norm": 2.398813486099243,
      "learning_rate": 2.9989743651281424e-05,
      "loss": 1.8948,
      "step": 6045
    },
    {
      "epoch": 0.011800949528546224,
      "grad_norm": 2.6533145904541016,
      "learning_rate": 2.998969269397381e-05,
      "loss": 1.8532,
      "step": 6060
    },
    {
      "epoch": 0.011830159799656487,
      "grad_norm": 3.8135955333709717,
      "learning_rate": 2.998964161043589e-05,
      "loss": 1.9564,
      "step": 6075
    },
    {
      "epoch": 0.01185937007076675,
      "grad_norm": 2.5285484790802,
      "learning_rate": 2.9989590400668086e-05,
      "loss": 1.8229,
      "step": 6090
    },
    {
      "epoch": 0.011888580341877014,
      "grad_norm": 2.0157272815704346,
      "learning_rate": 2.9989539064670838e-05,
      "loss": 1.9952,
      "step": 6105
    },
    {
      "epoch": 0.011917790612987277,
      "grad_norm": 1.8600102663040161,
      "learning_rate": 2.998948760244457e-05,
      "loss": 2.0176,
      "step": 6120
    },
    {
      "epoch": 0.01194700088409754,
      "grad_norm": 2.4452428817749023,
      "learning_rate": 2.9989436013989718e-05,
      "loss": 1.8965,
      "step": 6135
    },
    {
      "epoch": 0.011976211155207802,
      "grad_norm": 3.593918561935425,
      "learning_rate": 2.998938429930672e-05,
      "loss": 1.9629,
      "step": 6150
    },
    {
      "epoch": 0.012005421426318065,
      "grad_norm": 2.161616086959839,
      "learning_rate": 2.9989332458396005e-05,
      "loss": 2.0643,
      "step": 6165
    },
    {
      "epoch": 0.012034631697428327,
      "grad_norm": 4.1473493576049805,
      "learning_rate": 2.9989280491258015e-05,
      "loss": 1.8227,
      "step": 6180
    },
    {
      "epoch": 0.01206384196853859,
      "grad_norm": 3.4432153701782227,
      "learning_rate": 2.9989228397893186e-05,
      "loss": 1.9787,
      "step": 6195
    },
    {
      "epoch": 0.012093052239648853,
      "grad_norm": 3.591747283935547,
      "learning_rate": 2.9989176178301955e-05,
      "loss": 1.8248,
      "step": 6210
    },
    {
      "epoch": 0.012122262510759117,
      "grad_norm": 2.385715961456299,
      "learning_rate": 2.9989123832484767e-05,
      "loss": 2.0869,
      "step": 6225
    },
    {
      "epoch": 0.01215147278186938,
      "grad_norm": 3.128617763519287,
      "learning_rate": 2.9989071360442058e-05,
      "loss": 1.8622,
      "step": 6240
    },
    {
      "epoch": 0.012180683052979643,
      "grad_norm": 5.122420787811279,
      "learning_rate": 2.998901876217427e-05,
      "loss": 1.828,
      "step": 6255
    },
    {
      "epoch": 0.012209893324089905,
      "grad_norm": 2.451355218887329,
      "learning_rate": 2.9988966037681844e-05,
      "loss": 1.8766,
      "step": 6270
    },
    {
      "epoch": 0.012239103595200168,
      "grad_norm": 2.8492448329925537,
      "learning_rate": 2.9988913186965232e-05,
      "loss": 1.9675,
      "step": 6285
    },
    {
      "epoch": 0.01226831386631043,
      "grad_norm": 4.332848072052002,
      "learning_rate": 2.998886021002487e-05,
      "loss": 2.0243,
      "step": 6300
    },
    {
      "epoch": 0.012297524137420694,
      "grad_norm": 2.7769408226013184,
      "learning_rate": 2.9988807106861208e-05,
      "loss": 1.9015,
      "step": 6315
    },
    {
      "epoch": 0.012326734408530956,
      "grad_norm": 2.2205896377563477,
      "learning_rate": 2.9988753877474696e-05,
      "loss": 1.8927,
      "step": 6330
    },
    {
      "epoch": 0.01235594467964122,
      "grad_norm": 4.445504665374756,
      "learning_rate": 2.9988700521865777e-05,
      "loss": 1.9646,
      "step": 6345
    },
    {
      "epoch": 0.012385154950751483,
      "grad_norm": 2.951155662536621,
      "learning_rate": 2.9988647040034905e-05,
      "loss": 1.8487,
      "step": 6360
    },
    {
      "epoch": 0.012414365221861746,
      "grad_norm": 4.908761501312256,
      "learning_rate": 2.998859343198253e-05,
      "loss": 1.8704,
      "step": 6375
    },
    {
      "epoch": 0.012443575492972009,
      "grad_norm": 3.0668880939483643,
      "learning_rate": 2.9988539697709098e-05,
      "loss": 1.8632,
      "step": 6390
    },
    {
      "epoch": 0.012472785764082272,
      "grad_norm": 1.8523426055908203,
      "learning_rate": 2.9988485837215068e-05,
      "loss": 1.8434,
      "step": 6405
    },
    {
      "epoch": 0.012501996035192534,
      "grad_norm": 4.285421848297119,
      "learning_rate": 2.9988431850500887e-05,
      "loss": 1.8438,
      "step": 6420
    },
    {
      "epoch": 0.012531206306302797,
      "grad_norm": 1.9119669198989868,
      "learning_rate": 2.9988377737567013e-05,
      "loss": 1.8469,
      "step": 6435
    },
    {
      "epoch": 0.01256041657741306,
      "grad_norm": 2.430860757827759,
      "learning_rate": 2.9988323498413907e-05,
      "loss": 2.1108,
      "step": 6450
    },
    {
      "epoch": 0.012589626848523324,
      "grad_norm": 2.2205920219421387,
      "learning_rate": 2.9988269133042016e-05,
      "loss": 1.8481,
      "step": 6465
    },
    {
      "epoch": 0.012618837119633587,
      "grad_norm": 2.0588815212249756,
      "learning_rate": 2.9988214641451804e-05,
      "loss": 1.8007,
      "step": 6480
    },
    {
      "epoch": 0.01264804739074385,
      "grad_norm": 2.527836322784424,
      "learning_rate": 2.998816002364373e-05,
      "loss": 1.9974,
      "step": 6495
    },
    {
      "epoch": 0.012677257661854112,
      "grad_norm": 3.0486202239990234,
      "learning_rate": 2.9988105279618253e-05,
      "loss": 1.8459,
      "step": 6510
    },
    {
      "epoch": 0.012706467932964375,
      "grad_norm": 2.6629745960235596,
      "learning_rate": 2.998805040937583e-05,
      "loss": 1.9373,
      "step": 6525
    },
    {
      "epoch": 0.012735678204074638,
      "grad_norm": 3.2922544479370117,
      "learning_rate": 2.9987995412916928e-05,
      "loss": 1.8839,
      "step": 6540
    },
    {
      "epoch": 0.0127648884751849,
      "grad_norm": 4.026648044586182,
      "learning_rate": 2.998794029024201e-05,
      "loss": 2.1104,
      "step": 6555
    },
    {
      "epoch": 0.012794098746295163,
      "grad_norm": 2.473214864730835,
      "learning_rate": 2.998788504135154e-05,
      "loss": 1.664,
      "step": 6570
    },
    {
      "epoch": 0.012823309017405428,
      "grad_norm": 2.8834683895111084,
      "learning_rate": 2.998782966624598e-05,
      "loss": 1.9737,
      "step": 6585
    },
    {
      "epoch": 0.01285251928851569,
      "grad_norm": 3.123595952987671,
      "learning_rate": 2.99877741649258e-05,
      "loss": 2.1071,
      "step": 6600
    },
    {
      "epoch": 0.012881729559625953,
      "grad_norm": 2.3172380924224854,
      "learning_rate": 2.998771853739146e-05,
      "loss": 1.8585,
      "step": 6615
    },
    {
      "epoch": 0.012910939830736216,
      "grad_norm": 3.4954440593719482,
      "learning_rate": 2.998766278364344e-05,
      "loss": 1.8458,
      "step": 6630
    },
    {
      "epoch": 0.012940150101846478,
      "grad_norm": 2.48248553276062,
      "learning_rate": 2.9987606903682203e-05,
      "loss": 1.9398,
      "step": 6645
    },
    {
      "epoch": 0.012969360372956741,
      "grad_norm": 2.7575950622558594,
      "learning_rate": 2.998755089750822e-05,
      "loss": 1.8692,
      "step": 6660
    },
    {
      "epoch": 0.012998570644067004,
      "grad_norm": 2.821286678314209,
      "learning_rate": 2.9987494765121962e-05,
      "loss": 2.1104,
      "step": 6675
    },
    {
      "epoch": 0.013027780915177268,
      "grad_norm": 3.881669521331787,
      "learning_rate": 2.99874385065239e-05,
      "loss": 1.8453,
      "step": 6690
    },
    {
      "epoch": 0.013056991186287531,
      "grad_norm": 2.573246955871582,
      "learning_rate": 2.9987382121714516e-05,
      "loss": 1.9538,
      "step": 6705
    },
    {
      "epoch": 0.013086201457397794,
      "grad_norm": 4.5084052085876465,
      "learning_rate": 2.9987325610694277e-05,
      "loss": 1.9216,
      "step": 6720
    },
    {
      "epoch": 0.013115411728508056,
      "grad_norm": 3.882384777069092,
      "learning_rate": 2.9987268973463662e-05,
      "loss": 1.7694,
      "step": 6735
    },
    {
      "epoch": 0.013144621999618319,
      "grad_norm": 2.769113302230835,
      "learning_rate": 2.9987212210023147e-05,
      "loss": 1.9591,
      "step": 6750
    },
    {
      "epoch": 0.013173832270728582,
      "grad_norm": 3.9864275455474854,
      "learning_rate": 2.9987155320373207e-05,
      "loss": 1.9079,
      "step": 6765
    },
    {
      "epoch": 0.013203042541838845,
      "grad_norm": 1.9832093715667725,
      "learning_rate": 2.998709830451433e-05,
      "loss": 1.8263,
      "step": 6780
    },
    {
      "epoch": 0.013232252812949107,
      "grad_norm": 5.0461106300354,
      "learning_rate": 2.9987041162446985e-05,
      "loss": 1.8265,
      "step": 6795
    },
    {
      "epoch": 0.013261463084059372,
      "grad_norm": 2.5139777660369873,
      "learning_rate": 2.998698389417166e-05,
      "loss": 1.8644,
      "step": 6810
    },
    {
      "epoch": 0.013290673355169634,
      "grad_norm": 2.5377390384674072,
      "learning_rate": 2.998692649968884e-05,
      "loss": 1.9956,
      "step": 6825
    },
    {
      "epoch": 0.013319883626279897,
      "grad_norm": 2.68589186668396,
      "learning_rate": 2.9986868978998998e-05,
      "loss": 1.9298,
      "step": 6840
    },
    {
      "epoch": 0.01334909389739016,
      "grad_norm": 5.362221717834473,
      "learning_rate": 2.9986811332102624e-05,
      "loss": 1.8319,
      "step": 6855
    },
    {
      "epoch": 0.013378304168500423,
      "grad_norm": 3.440687417984009,
      "learning_rate": 2.9986753559000207e-05,
      "loss": 1.8854,
      "step": 6870
    },
    {
      "epoch": 0.013407514439610685,
      "grad_norm": 2.2173657417297363,
      "learning_rate": 2.9986695659692233e-05,
      "loss": 1.8672,
      "step": 6885
    },
    {
      "epoch": 0.013436724710720948,
      "grad_norm": 2.1564152240753174,
      "learning_rate": 2.998663763417918e-05,
      "loss": 1.9548,
      "step": 6900
    },
    {
      "epoch": 0.01346593498183121,
      "grad_norm": 2.339550018310547,
      "learning_rate": 2.9986579482461552e-05,
      "loss": 1.7549,
      "step": 6915
    },
    {
      "epoch": 0.013495145252941475,
      "grad_norm": 8.277188301086426,
      "learning_rate": 2.9986521204539824e-05,
      "loss": 1.8958,
      "step": 6930
    },
    {
      "epoch": 0.013524355524051738,
      "grad_norm": 3.044663190841675,
      "learning_rate": 2.9986462800414498e-05,
      "loss": 1.9019,
      "step": 6945
    },
    {
      "epoch": 0.013553565795162,
      "grad_norm": 2.622844696044922,
      "learning_rate": 2.9986404270086056e-05,
      "loss": 1.8349,
      "step": 6960
    },
    {
      "epoch": 0.013582776066272263,
      "grad_norm": 2.686244010925293,
      "learning_rate": 2.9986345613554998e-05,
      "loss": 1.9178,
      "step": 6975
    },
    {
      "epoch": 0.013611986337382526,
      "grad_norm": 3.487210512161255,
      "learning_rate": 2.9986286830821817e-05,
      "loss": 2.0353,
      "step": 6990
    },
    {
      "epoch": 0.013641196608492789,
      "grad_norm": 3.297060966491699,
      "learning_rate": 2.9986227921887005e-05,
      "loss": 2.0344,
      "step": 7005
    },
    {
      "epoch": 0.013670406879603051,
      "grad_norm": 6.4708571434021,
      "learning_rate": 2.9986168886751064e-05,
      "loss": 1.946,
      "step": 7020
    },
    {
      "epoch": 0.013699617150713314,
      "grad_norm": 3.04331636428833,
      "learning_rate": 2.9986109725414485e-05,
      "loss": 1.9229,
      "step": 7035
    },
    {
      "epoch": 0.013728827421823579,
      "grad_norm": 3.3614730834960938,
      "learning_rate": 2.9986050437877762e-05,
      "loss": 1.8943,
      "step": 7050
    },
    {
      "epoch": 0.013758037692933841,
      "grad_norm": 2.632763624191284,
      "learning_rate": 2.998599102414141e-05,
      "loss": 1.907,
      "step": 7065
    },
    {
      "epoch": 0.013787247964044104,
      "grad_norm": 2.441969871520996,
      "learning_rate": 2.998593148420592e-05,
      "loss": 1.9382,
      "step": 7080
    },
    {
      "epoch": 0.013816458235154367,
      "grad_norm": 3.3013856410980225,
      "learning_rate": 2.9985871818071784e-05,
      "loss": 1.7866,
      "step": 7095
    },
    {
      "epoch": 0.01384566850626463,
      "grad_norm": 3.12922739982605,
      "learning_rate": 2.9985812025739518e-05,
      "loss": 1.9155,
      "step": 7110
    },
    {
      "epoch": 0.013874878777374892,
      "grad_norm": 3.3649489879608154,
      "learning_rate": 2.998575210720962e-05,
      "loss": 1.7256,
      "step": 7125
    },
    {
      "epoch": 0.013904089048485155,
      "grad_norm": 2.381049871444702,
      "learning_rate": 2.9985692062482603e-05,
      "loss": 1.9324,
      "step": 7140
    },
    {
      "epoch": 0.013933299319595418,
      "grad_norm": 2.5328924655914307,
      "learning_rate": 2.998563189155896e-05,
      "loss": 2.0635,
      "step": 7155
    },
    {
      "epoch": 0.013962509590705682,
      "grad_norm": 3.5615475177764893,
      "learning_rate": 2.99855715944392e-05,
      "loss": 1.8377,
      "step": 7170
    },
    {
      "epoch": 0.013991719861815945,
      "grad_norm": 3.7230849266052246,
      "learning_rate": 2.998551117112384e-05,
      "loss": 1.839,
      "step": 7185
    },
    {
      "epoch": 0.014020930132926207,
      "grad_norm": 2.7863929271698,
      "learning_rate": 2.998545062161338e-05,
      "loss": 1.7265,
      "step": 7200
    },
    {
      "epoch": 0.01405014040403647,
      "grad_norm": 2.238398790359497,
      "learning_rate": 2.9985389945908332e-05,
      "loss": 1.9747,
      "step": 7215
    },
    {
      "epoch": 0.014079350675146733,
      "grad_norm": 2.3368685245513916,
      "learning_rate": 2.998532914400921e-05,
      "loss": 2.0277,
      "step": 7230
    },
    {
      "epoch": 0.014108560946256996,
      "grad_norm": 3.667415142059326,
      "learning_rate": 2.9985268215916523e-05,
      "loss": 1.8947,
      "step": 7245
    },
    {
      "epoch": 0.014137771217367258,
      "grad_norm": 5.902375221252441,
      "learning_rate": 2.9985207161630784e-05,
      "loss": 2.023,
      "step": 7260
    },
    {
      "epoch": 0.014166981488477521,
      "grad_norm": 3.5179100036621094,
      "learning_rate": 2.998514598115251e-05,
      "loss": 1.9654,
      "step": 7275
    },
    {
      "epoch": 0.014196191759587785,
      "grad_norm": 4.09523344039917,
      "learning_rate": 2.9985084674482207e-05,
      "loss": 1.9822,
      "step": 7290
    },
    {
      "epoch": 0.014225402030698048,
      "grad_norm": 3.9758706092834473,
      "learning_rate": 2.9985023241620405e-05,
      "loss": 1.7241,
      "step": 7305
    },
    {
      "epoch": 0.01425461230180831,
      "grad_norm": 4.463382720947266,
      "learning_rate": 2.9984961682567614e-05,
      "loss": 1.9208,
      "step": 7320
    },
    {
      "epoch": 0.014283822572918574,
      "grad_norm": 2.2591640949249268,
      "learning_rate": 2.9984899997324357e-05,
      "loss": 1.9372,
      "step": 7335
    },
    {
      "epoch": 0.014313032844028836,
      "grad_norm": 2.958768129348755,
      "learning_rate": 2.998483818589114e-05,
      "loss": 1.9092,
      "step": 7350
    },
    {
      "epoch": 0.014342243115139099,
      "grad_norm": 2.4651241302490234,
      "learning_rate": 2.99847762482685e-05,
      "loss": 1.7909,
      "step": 7365
    },
    {
      "epoch": 0.014371453386249362,
      "grad_norm": 4.785153388977051,
      "learning_rate": 2.9984714184456948e-05,
      "loss": 2.0846,
      "step": 7380
    },
    {
      "epoch": 0.014400663657359624,
      "grad_norm": 5.212136745452881,
      "learning_rate": 2.9984651994457013e-05,
      "loss": 1.7623,
      "step": 7395
    },
    {
      "epoch": 0.014429873928469889,
      "grad_norm": 3.5508365631103516,
      "learning_rate": 2.9984589678269216e-05,
      "loss": 1.8922,
      "step": 7410
    },
    {
      "epoch": 0.014459084199580152,
      "grad_norm": 4.388883113861084,
      "learning_rate": 2.998452723589408e-05,
      "loss": 1.9701,
      "step": 7425
    },
    {
      "epoch": 0.014488294470690414,
      "grad_norm": 2.4577128887176514,
      "learning_rate": 2.9984464667332135e-05,
      "loss": 1.8807,
      "step": 7440
    },
    {
      "epoch": 0.014517504741800677,
      "grad_norm": 3.046642541885376,
      "learning_rate": 2.99844019725839e-05,
      "loss": 2.0671,
      "step": 7455
    },
    {
      "epoch": 0.01454671501291094,
      "grad_norm": 2.639798879623413,
      "learning_rate": 2.9984339151649913e-05,
      "loss": 1.7981,
      "step": 7470
    },
    {
      "epoch": 0.014575925284021202,
      "grad_norm": 4.231945991516113,
      "learning_rate": 2.9984276204530702e-05,
      "loss": 1.822,
      "step": 7485
    },
    {
      "epoch": 0.014605135555131465,
      "grad_norm": 2.3419246673583984,
      "learning_rate": 2.9984213131226788e-05,
      "loss": 1.8931,
      "step": 7500
    },
    {
      "epoch": 0.014634345826241728,
      "grad_norm": 3.034775733947754,
      "learning_rate": 2.998414993173871e-05,
      "loss": 1.8925,
      "step": 7515
    },
    {
      "epoch": 0.014663556097351992,
      "grad_norm": 2.818657875061035,
      "learning_rate": 2.9984086606066997e-05,
      "loss": 1.8714,
      "step": 7530
    },
    {
      "epoch": 0.014692766368462255,
      "grad_norm": 2.0940043926239014,
      "learning_rate": 2.9984023154212183e-05,
      "loss": 1.9476,
      "step": 7545
    },
    {
      "epoch": 0.014721976639572518,
      "grad_norm": 1.9247641563415527,
      "learning_rate": 2.9983959576174807e-05,
      "loss": 1.8433,
      "step": 7560
    },
    {
      "epoch": 0.01475118691068278,
      "grad_norm": 4.496128559112549,
      "learning_rate": 2.9983895871955397e-05,
      "loss": 1.8137,
      "step": 7575
    },
    {
      "epoch": 0.014780397181793043,
      "grad_norm": 1.846083402633667,
      "learning_rate": 2.998383204155449e-05,
      "loss": 1.9474,
      "step": 7590
    },
    {
      "epoch": 0.014809607452903306,
      "grad_norm": 2.063615083694458,
      "learning_rate": 2.9983768084972626e-05,
      "loss": 1.9017,
      "step": 7605
    },
    {
      "epoch": 0.014838817724013569,
      "grad_norm": 3.480355978012085,
      "learning_rate": 2.9983704002210346e-05,
      "loss": 1.8554,
      "step": 7620
    },
    {
      "epoch": 0.014868027995123831,
      "grad_norm": 3.3823037147521973,
      "learning_rate": 2.9983639793268187e-05,
      "loss": 2.067,
      "step": 7635
    },
    {
      "epoch": 0.014897238266234096,
      "grad_norm": 4.111497402191162,
      "learning_rate": 2.998357545814669e-05,
      "loss": 1.8451,
      "step": 7650
    },
    {
      "epoch": 0.014926448537344358,
      "grad_norm": 2.51603102684021,
      "learning_rate": 2.9983510996846397e-05,
      "loss": 1.835,
      "step": 7665
    },
    {
      "epoch": 0.014955658808454621,
      "grad_norm": 2.576188087463379,
      "learning_rate": 2.9983446409367846e-05,
      "loss": 2.0444,
      "step": 7680
    },
    {
      "epoch": 0.014984869079564884,
      "grad_norm": 2.695582151412964,
      "learning_rate": 2.9983381695711595e-05,
      "loss": 1.9372,
      "step": 7695
    },
    {
      "epoch": 0.015014079350675147,
      "grad_norm": 2.580967664718628,
      "learning_rate": 2.9983316855878172e-05,
      "loss": 1.7984,
      "step": 7710
    },
    {
      "epoch": 0.01504328962178541,
      "grad_norm": 2.779932737350464,
      "learning_rate": 2.9983251889868133e-05,
      "loss": 2.0286,
      "step": 7725
    },
    {
      "epoch": 0.015072499892895672,
      "grad_norm": 4.4765849113464355,
      "learning_rate": 2.998318679768202e-05,
      "loss": 1.873,
      "step": 7740
    },
    {
      "epoch": 0.015101710164005935,
      "grad_norm": 5.089080810546875,
      "learning_rate": 2.9983121579320387e-05,
      "loss": 1.9452,
      "step": 7755
    },
    {
      "epoch": 0.015130920435116199,
      "grad_norm": 3.860607862472534,
      "learning_rate": 2.9983056234783774e-05,
      "loss": 1.9469,
      "step": 7770
    },
    {
      "epoch": 0.015160130706226462,
      "grad_norm": 3.014214277267456,
      "learning_rate": 2.998299076407274e-05,
      "loss": 2.0488,
      "step": 7785
    },
    {
      "epoch": 0.015189340977336725,
      "grad_norm": 3.1203925609588623,
      "learning_rate": 2.998292516718784e-05,
      "loss": 1.8334,
      "step": 7800
    },
    {
      "epoch": 0.015218551248446987,
      "grad_norm": 1.6761817932128906,
      "learning_rate": 2.998285944412961e-05,
      "loss": 2.0263,
      "step": 7815
    },
    {
      "epoch": 0.01524776151955725,
      "grad_norm": 2.3997247219085693,
      "learning_rate": 2.9982793594898623e-05,
      "loss": 1.9195,
      "step": 7830
    },
    {
      "epoch": 0.015276971790667513,
      "grad_norm": 2.0805671215057373,
      "learning_rate": 2.998272761949542e-05,
      "loss": 1.8461,
      "step": 7845
    },
    {
      "epoch": 0.015306182061777775,
      "grad_norm": 3.0058650970458984,
      "learning_rate": 2.998266151792056e-05,
      "loss": 2.0221,
      "step": 7860
    },
    {
      "epoch": 0.01533539233288804,
      "grad_norm": 4.376441955566406,
      "learning_rate": 2.99825952901746e-05,
      "loss": 1.8942,
      "step": 7875
    },
    {
      "epoch": 0.015364602603998303,
      "grad_norm": 2.509711503982544,
      "learning_rate": 2.9982528936258096e-05,
      "loss": 1.7979,
      "step": 7890
    },
    {
      "epoch": 0.015393812875108565,
      "grad_norm": 2.763103723526001,
      "learning_rate": 2.9982462456171605e-05,
      "loss": 1.9112,
      "step": 7905
    },
    {
      "epoch": 0.015423023146218828,
      "grad_norm": 2.519554615020752,
      "learning_rate": 2.9982395849915698e-05,
      "loss": 1.9375,
      "step": 7920
    },
    {
      "epoch": 0.01545223341732909,
      "grad_norm": 2.679543972015381,
      "learning_rate": 2.9982329117490926e-05,
      "loss": 1.8226,
      "step": 7935
    },
    {
      "epoch": 0.015481443688439353,
      "grad_norm": 3.619253158569336,
      "learning_rate": 2.9982262258897855e-05,
      "loss": 1.8639,
      "step": 7950
    },
    {
      "epoch": 0.015510653959549616,
      "grad_norm": 1.7691304683685303,
      "learning_rate": 2.9982195274137042e-05,
      "loss": 1.7112,
      "step": 7965
    },
    {
      "epoch": 0.015539864230659879,
      "grad_norm": 3.720994234085083,
      "learning_rate": 2.9982128163209058e-05,
      "loss": 1.8901,
      "step": 7980
    },
    {
      "epoch": 0.015569074501770143,
      "grad_norm": 3.499058723449707,
      "learning_rate": 2.9982060926114467e-05,
      "loss": 1.9527,
      "step": 7995
    },
    {
      "epoch": 0.015598284772880406,
      "grad_norm": 2.401658773422241,
      "learning_rate": 2.9981993562853833e-05,
      "loss": 1.9037,
      "step": 8010
    },
    {
      "epoch": 0.015627495043990667,
      "grad_norm": 3.7763659954071045,
      "learning_rate": 2.9981926073427724e-05,
      "loss": 2.0314,
      "step": 8025
    },
    {
      "epoch": 0.01565670531510093,
      "grad_norm": 3.2521162033081055,
      "learning_rate": 2.9981858457836707e-05,
      "loss": 1.7982,
      "step": 8040
    },
    {
      "epoch": 0.015685915586211196,
      "grad_norm": 3.2938101291656494,
      "learning_rate": 2.9981790716081353e-05,
      "loss": 1.9225,
      "step": 8055
    },
    {
      "epoch": 0.015715125857321457,
      "grad_norm": 2.9045004844665527,
      "learning_rate": 2.9981722848162233e-05,
      "loss": 2.04,
      "step": 8070
    },
    {
      "epoch": 0.01574433612843172,
      "grad_norm": 2.2528388500213623,
      "learning_rate": 2.9981654854079918e-05,
      "loss": 1.792,
      "step": 8085
    },
    {
      "epoch": 0.015773546399541982,
      "grad_norm": 2.9156363010406494,
      "learning_rate": 2.998158673383498e-05,
      "loss": 1.9291,
      "step": 8100
    },
    {
      "epoch": 0.015802756670652247,
      "grad_norm": 2.375291347503662,
      "learning_rate": 2.9981518487427996e-05,
      "loss": 1.8216,
      "step": 8115
    },
    {
      "epoch": 0.015831966941762508,
      "grad_norm": 4.628626823425293,
      "learning_rate": 2.9981450114859532e-05,
      "loss": 1.7895,
      "step": 8130
    },
    {
      "epoch": 0.015861177212872772,
      "grad_norm": 2.200885534286499,
      "learning_rate": 2.9981381616130172e-05,
      "loss": 1.9302,
      "step": 8145
    },
    {
      "epoch": 0.015890387483983033,
      "grad_norm": 3.7578535079956055,
      "learning_rate": 2.998131299124049e-05,
      "loss": 1.9202,
      "step": 8160
    },
    {
      "epoch": 0.015919597755093298,
      "grad_norm": 2.235351800918579,
      "learning_rate": 2.9981244240191063e-05,
      "loss": 1.9151,
      "step": 8175
    },
    {
      "epoch": 0.015948808026203562,
      "grad_norm": 1.905808925628662,
      "learning_rate": 2.9981175362982473e-05,
      "loss": 2.0942,
      "step": 8190
    },
    {
      "epoch": 0.015978018297313823,
      "grad_norm": 2.797210454940796,
      "learning_rate": 2.99811063596153e-05,
      "loss": 2.0168,
      "step": 8205
    },
    {
      "epoch": 0.016007228568424087,
      "grad_norm": 2.7250447273254395,
      "learning_rate": 2.9981037230090125e-05,
      "loss": 1.8033,
      "step": 8220
    },
    {
      "epoch": 0.01603643883953435,
      "grad_norm": 2.6646409034729004,
      "learning_rate": 2.9980967974407525e-05,
      "loss": 1.8914,
      "step": 8235
    },
    {
      "epoch": 0.016065649110644613,
      "grad_norm": 4.621725082397461,
      "learning_rate": 2.9980898592568086e-05,
      "loss": 1.9805,
      "step": 8250
    },
    {
      "epoch": 0.016094859381754874,
      "grad_norm": 7.544195175170898,
      "learning_rate": 2.9980829084572393e-05,
      "loss": 1.8876,
      "step": 8265
    },
    {
      "epoch": 0.016124069652865138,
      "grad_norm": 3.1491341590881348,
      "learning_rate": 2.9980759450421032e-05,
      "loss": 1.8096,
      "step": 8280
    },
    {
      "epoch": 0.016153279923975403,
      "grad_norm": 3.049743175506592,
      "learning_rate": 2.998068969011459e-05,
      "loss": 1.9014,
      "step": 8295
    },
    {
      "epoch": 0.016182490195085664,
      "grad_norm": 3.4776172637939453,
      "learning_rate": 2.998061980365365e-05,
      "loss": 1.9011,
      "step": 8310
    },
    {
      "epoch": 0.016211700466195928,
      "grad_norm": 3.4769961833953857,
      "learning_rate": 2.9980549791038804e-05,
      "loss": 2.2367,
      "step": 8325
    },
    {
      "epoch": 0.01624091073730619,
      "grad_norm": 2.252976894378662,
      "learning_rate": 2.9980479652270645e-05,
      "loss": 1.9017,
      "step": 8340
    },
    {
      "epoch": 0.016270121008416454,
      "grad_norm": 2.3866465091705322,
      "learning_rate": 2.998040938734976e-05,
      "loss": 1.8849,
      "step": 8355
    },
    {
      "epoch": 0.016299331279526715,
      "grad_norm": 4.388668060302734,
      "learning_rate": 2.998033899627674e-05,
      "loss": 1.8601,
      "step": 8370
    },
    {
      "epoch": 0.01632854155063698,
      "grad_norm": 3.2678322792053223,
      "learning_rate": 2.9980268479052173e-05,
      "loss": 1.8799,
      "step": 8385
    },
    {
      "epoch": 0.01635775182174724,
      "grad_norm": 3.3072402477264404,
      "learning_rate": 2.9980197835676665e-05,
      "loss": 1.8995,
      "step": 8400
    },
    {
      "epoch": 0.016386962092857504,
      "grad_norm": 2.1432266235351562,
      "learning_rate": 2.99801270661508e-05,
      "loss": 1.9894,
      "step": 8415
    },
    {
      "epoch": 0.01641617236396777,
      "grad_norm": 4.359271049499512,
      "learning_rate": 2.998005617047518e-05,
      "loss": 1.8252,
      "step": 8430
    },
    {
      "epoch": 0.01644538263507803,
      "grad_norm": 4.683668613433838,
      "learning_rate": 2.99799851486504e-05,
      "loss": 1.8865,
      "step": 8445
    },
    {
      "epoch": 0.016474592906188294,
      "grad_norm": 2.4480679035186768,
      "learning_rate": 2.997991400067706e-05,
      "loss": 1.8999,
      "step": 8460
    },
    {
      "epoch": 0.016503803177298555,
      "grad_norm": 4.169287204742432,
      "learning_rate": 2.9979842726555753e-05,
      "loss": 1.8878,
      "step": 8475
    },
    {
      "epoch": 0.01653301344840882,
      "grad_norm": 2.4626035690307617,
      "learning_rate": 2.9979771326287084e-05,
      "loss": 2.0727,
      "step": 8490
    },
    {
      "epoch": 0.01656222371951908,
      "grad_norm": 3.7514021396636963,
      "learning_rate": 2.9979699799871658e-05,
      "loss": 1.9559,
      "step": 8505
    },
    {
      "epoch": 0.016591433990629345,
      "grad_norm": 3.0994958877563477,
      "learning_rate": 2.9979628147310068e-05,
      "loss": 1.8535,
      "step": 8520
    },
    {
      "epoch": 0.01662064426173961,
      "grad_norm": 5.931899070739746,
      "learning_rate": 2.9979556368602924e-05,
      "loss": 1.9244,
      "step": 8535
    },
    {
      "epoch": 0.01664985453284987,
      "grad_norm": 2.016737699508667,
      "learning_rate": 2.9979484463750833e-05,
      "loss": 1.9107,
      "step": 8550
    },
    {
      "epoch": 0.016679064803960135,
      "grad_norm": 2.3796162605285645,
      "learning_rate": 2.9979412432754394e-05,
      "loss": 2.0904,
      "step": 8565
    },
    {
      "epoch": 0.016708275075070396,
      "grad_norm": 1.870884656906128,
      "learning_rate": 2.9979340275614217e-05,
      "loss": 1.926,
      "step": 8580
    },
    {
      "epoch": 0.01673748534618066,
      "grad_norm": 2.833564281463623,
      "learning_rate": 2.997926799233091e-05,
      "loss": 1.871,
      "step": 8595
    },
    {
      "epoch": 0.01676669561729092,
      "grad_norm": 3.7916762828826904,
      "learning_rate": 2.9979195582905075e-05,
      "loss": 1.8166,
      "step": 8610
    },
    {
      "epoch": 0.016795905888401186,
      "grad_norm": 2.3426475524902344,
      "learning_rate": 2.997912304733733e-05,
      "loss": 1.8514,
      "step": 8625
    },
    {
      "epoch": 0.016825116159511447,
      "grad_norm": 4.586437225341797,
      "learning_rate": 2.9979050385628286e-05,
      "loss": 1.8544,
      "step": 8640
    },
    {
      "epoch": 0.01685432643062171,
      "grad_norm": 4.281703472137451,
      "learning_rate": 2.997897759777855e-05,
      "loss": 1.8978,
      "step": 8655
    },
    {
      "epoch": 0.016883536701731976,
      "grad_norm": 2.1048526763916016,
      "learning_rate": 2.9978904683788735e-05,
      "loss": 1.9104,
      "step": 8670
    },
    {
      "epoch": 0.016912746972842237,
      "grad_norm": 4.526645660400391,
      "learning_rate": 2.9978831643659462e-05,
      "loss": 1.9814,
      "step": 8685
    },
    {
      "epoch": 0.0169419572439525,
      "grad_norm": 1.9656120538711548,
      "learning_rate": 2.9978758477391334e-05,
      "loss": 1.878,
      "step": 8700
    },
    {
      "epoch": 0.016971167515062762,
      "grad_norm": 2.5091657638549805,
      "learning_rate": 2.997868518498498e-05,
      "loss": 1.8586,
      "step": 8715
    },
    {
      "epoch": 0.017000377786173027,
      "grad_norm": 3.0821568965911865,
      "learning_rate": 2.997861176644101e-05,
      "loss": 2.1301,
      "step": 8730
    },
    {
      "epoch": 0.017029588057283288,
      "grad_norm": 2.465061902999878,
      "learning_rate": 2.997853822176004e-05,
      "loss": 1.9435,
      "step": 8745
    },
    {
      "epoch": 0.017058798328393552,
      "grad_norm": 4.7108306884765625,
      "learning_rate": 2.9978464550942697e-05,
      "loss": 1.8857,
      "step": 8760
    },
    {
      "epoch": 0.017088008599503816,
      "grad_norm": 2.838949203491211,
      "learning_rate": 2.9978390753989597e-05,
      "loss": 1.9367,
      "step": 8775
    },
    {
      "epoch": 0.017117218870614077,
      "grad_norm": 6.954312324523926,
      "learning_rate": 2.9978316830901358e-05,
      "loss": 1.6954,
      "step": 8790
    },
    {
      "epoch": 0.017146429141724342,
      "grad_norm": 3.075137138366699,
      "learning_rate": 2.997824278167861e-05,
      "loss": 1.8787,
      "step": 8805
    },
    {
      "epoch": 0.017175639412834603,
      "grad_norm": 2.7041006088256836,
      "learning_rate": 2.9978168606321975e-05,
      "loss": 1.8062,
      "step": 8820
    },
    {
      "epoch": 0.017204849683944867,
      "grad_norm": 2.9021966457366943,
      "learning_rate": 2.997809430483207e-05,
      "loss": 1.9083,
      "step": 8835
    },
    {
      "epoch": 0.017234059955055128,
      "grad_norm": 3.350419521331787,
      "learning_rate": 2.9978019877209528e-05,
      "loss": 2.0403,
      "step": 8850
    },
    {
      "epoch": 0.017263270226165393,
      "grad_norm": 4.803377628326416,
      "learning_rate": 2.9977945323454977e-05,
      "loss": 1.8116,
      "step": 8865
    },
    {
      "epoch": 0.017292480497275657,
      "grad_norm": 4.362671375274658,
      "learning_rate": 2.997787064356904e-05,
      "loss": 1.8089,
      "step": 8880
    },
    {
      "epoch": 0.017321690768385918,
      "grad_norm": 3.1585206985473633,
      "learning_rate": 2.9977795837552347e-05,
      "loss": 2.1531,
      "step": 8895
    },
    {
      "epoch": 0.017350901039496183,
      "grad_norm": 2.9386544227600098,
      "learning_rate": 2.997772090540553e-05,
      "loss": 1.8995,
      "step": 8910
    },
    {
      "epoch": 0.017380111310606444,
      "grad_norm": 4.238663673400879,
      "learning_rate": 2.9977645847129216e-05,
      "loss": 1.9374,
      "step": 8925
    },
    {
      "epoch": 0.017409321581716708,
      "grad_norm": 3.94399356842041,
      "learning_rate": 2.9977570662724047e-05,
      "loss": 1.9782,
      "step": 8940
    },
    {
      "epoch": 0.01743853185282697,
      "grad_norm": 3.2048075199127197,
      "learning_rate": 2.9977495352190643e-05,
      "loss": 1.8984,
      "step": 8955
    },
    {
      "epoch": 0.017467742123937233,
      "grad_norm": 2.4807546138763428,
      "learning_rate": 2.9977419915529646e-05,
      "loss": 1.8613,
      "step": 8970
    },
    {
      "epoch": 0.017496952395047494,
      "grad_norm": 2.518021583557129,
      "learning_rate": 2.9977344352741686e-05,
      "loss": 2.0321,
      "step": 8985
    },
    {
      "epoch": 0.01752616266615776,
      "grad_norm": 4.450172424316406,
      "learning_rate": 2.9977268663827403e-05,
      "loss": 1.9419,
      "step": 9000
    },
    {
      "epoch": 0.017555372937268023,
      "grad_norm": 3.8519856929779053,
      "learning_rate": 2.9977192848787437e-05,
      "loss": 1.9083,
      "step": 9015
    },
    {
      "epoch": 0.017584583208378284,
      "grad_norm": 5.37404727935791,
      "learning_rate": 2.9977116907622422e-05,
      "loss": 1.9904,
      "step": 9030
    },
    {
      "epoch": 0.01761379347948855,
      "grad_norm": 2.9285478591918945,
      "learning_rate": 2.9977040840333e-05,
      "loss": 1.7262,
      "step": 9045
    },
    {
      "epoch": 0.01764300375059881,
      "grad_norm": 3.8375890254974365,
      "learning_rate": 2.9976964646919814e-05,
      "loss": 1.945,
      "step": 9060
    },
    {
      "epoch": 0.017672214021709074,
      "grad_norm": 2.987417459487915,
      "learning_rate": 2.9976888327383497e-05,
      "loss": 2.1196,
      "step": 9075
    },
    {
      "epoch": 0.017701424292819335,
      "grad_norm": 3.6940131187438965,
      "learning_rate": 2.99768118817247e-05,
      "loss": 2.002,
      "step": 9090
    },
    {
      "epoch": 0.0177306345639296,
      "grad_norm": 1.8791909217834473,
      "learning_rate": 2.997673530994406e-05,
      "loss": 1.9082,
      "step": 9105
    },
    {
      "epoch": 0.017759844835039864,
      "grad_norm": 5.242600440979004,
      "learning_rate": 2.997665861204223e-05,
      "loss": 1.8396,
      "step": 9120
    },
    {
      "epoch": 0.017789055106150125,
      "grad_norm": 3.7789740562438965,
      "learning_rate": 2.997658178801985e-05,
      "loss": 1.9959,
      "step": 9135
    },
    {
      "epoch": 0.01781826537726039,
      "grad_norm": 2.6715869903564453,
      "learning_rate": 2.9976504837877566e-05,
      "loss": 1.8346,
      "step": 9150
    },
    {
      "epoch": 0.01784747564837065,
      "grad_norm": 3.229962110519409,
      "learning_rate": 2.997642776161603e-05,
      "loss": 2.0002,
      "step": 9165
    },
    {
      "epoch": 0.017876685919480915,
      "grad_norm": 2.602320671081543,
      "learning_rate": 2.997635055923589e-05,
      "loss": 2.0765,
      "step": 9180
    },
    {
      "epoch": 0.017905896190591176,
      "grad_norm": 3.277393102645874,
      "learning_rate": 2.9976273230737795e-05,
      "loss": 1.8345,
      "step": 9195
    },
    {
      "epoch": 0.01793510646170144,
      "grad_norm": 2.1936373710632324,
      "learning_rate": 2.9976195776122397e-05,
      "loss": 1.9265,
      "step": 9210
    },
    {
      "epoch": 0.0179643167328117,
      "grad_norm": 4.018658638000488,
      "learning_rate": 2.997611819539035e-05,
      "loss": 1.9237,
      "step": 9225
    },
    {
      "epoch": 0.017993527003921966,
      "grad_norm": 3.5876455307006836,
      "learning_rate": 2.9976040488542304e-05,
      "loss": 1.8761,
      "step": 9240
    },
    {
      "epoch": 0.01802273727503223,
      "grad_norm": 2.9479000568389893,
      "learning_rate": 2.9975962655578915e-05,
      "loss": 1.9062,
      "step": 9255
    },
    {
      "epoch": 0.01805194754614249,
      "grad_norm": 3.185248613357544,
      "learning_rate": 2.9975884696500835e-05,
      "loss": 1.9958,
      "step": 9270
    },
    {
      "epoch": 0.018081157817252756,
      "grad_norm": 2.5612637996673584,
      "learning_rate": 2.9975806611308725e-05,
      "loss": 2.0438,
      "step": 9285
    },
    {
      "epoch": 0.018110368088363017,
      "grad_norm": 3.9929094314575195,
      "learning_rate": 2.9975728400003244e-05,
      "loss": 1.8404,
      "step": 9300
    },
    {
      "epoch": 0.01813957835947328,
      "grad_norm": 2.15783953666687,
      "learning_rate": 2.9975650062585043e-05,
      "loss": 1.8677,
      "step": 9315
    },
    {
      "epoch": 0.018168788630583542,
      "grad_norm": 3.7899887561798096,
      "learning_rate": 2.997557159905479e-05,
      "loss": 1.9269,
      "step": 9330
    },
    {
      "epoch": 0.018197998901693806,
      "grad_norm": 4.0455145835876465,
      "learning_rate": 2.9975493009413144e-05,
      "loss": 1.8552,
      "step": 9345
    },
    {
      "epoch": 0.01822720917280407,
      "grad_norm": 2.837963581085205,
      "learning_rate": 2.9975414293660766e-05,
      "loss": 2.1462,
      "step": 9360
    },
    {
      "epoch": 0.018256419443914332,
      "grad_norm": 3.8562536239624023,
      "learning_rate": 2.9975335451798317e-05,
      "loss": 1.8625,
      "step": 9375
    },
    {
      "epoch": 0.018285629715024596,
      "grad_norm": 2.4235100746154785,
      "learning_rate": 2.9975256483826453e-05,
      "loss": 1.8512,
      "step": 9390
    },
    {
      "epoch": 0.018314839986134857,
      "grad_norm": 5.756357192993164,
      "learning_rate": 2.997517738974586e-05,
      "loss": 1.84,
      "step": 9405
    },
    {
      "epoch": 0.01834405025724512,
      "grad_norm": 2.3832759857177734,
      "learning_rate": 2.9975098169557187e-05,
      "loss": 1.8721,
      "step": 9420
    },
    {
      "epoch": 0.018373260528355383,
      "grad_norm": 2.0849883556365967,
      "learning_rate": 2.9975018823261106e-05,
      "loss": 1.914,
      "step": 9435
    },
    {
      "epoch": 0.018402470799465647,
      "grad_norm": 2.3563778400421143,
      "learning_rate": 2.997493935085829e-05,
      "loss": 1.7283,
      "step": 9450
    },
    {
      "epoch": 0.018431681070575908,
      "grad_norm": 2.018721580505371,
      "learning_rate": 2.9974859752349396e-05,
      "loss": 1.7748,
      "step": 9465
    },
    {
      "epoch": 0.018460891341686173,
      "grad_norm": 2.725719451904297,
      "learning_rate": 2.9974780027735103e-05,
      "loss": 1.9287,
      "step": 9480
    },
    {
      "epoch": 0.018490101612796437,
      "grad_norm": 3.2846550941467285,
      "learning_rate": 2.9974700177016082e-05,
      "loss": 1.8214,
      "step": 9495
    },
    {
      "epoch": 0.018519311883906698,
      "grad_norm": 4.153242111206055,
      "learning_rate": 2.997462020019301e-05,
      "loss": 1.9377,
      "step": 9510
    },
    {
      "epoch": 0.018548522155016962,
      "grad_norm": 2.386509656906128,
      "learning_rate": 2.997454009726655e-05,
      "loss": 1.7513,
      "step": 9525
    },
    {
      "epoch": 0.018577732426127223,
      "grad_norm": 3.9288957118988037,
      "learning_rate": 2.9974459868237384e-05,
      "loss": 1.8623,
      "step": 9540
    },
    {
      "epoch": 0.018606942697237488,
      "grad_norm": 2.2002370357513428,
      "learning_rate": 2.9974379513106184e-05,
      "loss": 2.0153,
      "step": 9555
    },
    {
      "epoch": 0.01863615296834775,
      "grad_norm": 2.4271867275238037,
      "learning_rate": 2.9974299031873625e-05,
      "loss": 1.8772,
      "step": 9570
    },
    {
      "epoch": 0.018665363239458013,
      "grad_norm": 2.355729341506958,
      "learning_rate": 2.9974218424540395e-05,
      "loss": 1.9947,
      "step": 9585
    },
    {
      "epoch": 0.018694573510568278,
      "grad_norm": 4.210724830627441,
      "learning_rate": 2.9974137691107164e-05,
      "loss": 1.8113,
      "step": 9600
    },
    {
      "epoch": 0.01872378378167854,
      "grad_norm": 3.092832088470459,
      "learning_rate": 2.997405683157461e-05,
      "loss": 1.8457,
      "step": 9615
    },
    {
      "epoch": 0.018752994052788803,
      "grad_norm": 3.154505491256714,
      "learning_rate": 2.997397584594342e-05,
      "loss": 1.9223,
      "step": 9630
    },
    {
      "epoch": 0.018782204323899064,
      "grad_norm": 2.661449909210205,
      "learning_rate": 2.997389473421427e-05,
      "loss": 1.8362,
      "step": 9645
    },
    {
      "epoch": 0.01881141459500933,
      "grad_norm": 4.499241352081299,
      "learning_rate": 2.997381349638785e-05,
      "loss": 1.974,
      "step": 9660
    },
    {
      "epoch": 0.01884062486611959,
      "grad_norm": 8.712298393249512,
      "learning_rate": 2.9973732132464838e-05,
      "loss": 1.793,
      "step": 9675
    },
    {
      "epoch": 0.018869835137229854,
      "grad_norm": 3.771261215209961,
      "learning_rate": 2.9973650642445926e-05,
      "loss": 1.8158,
      "step": 9690
    },
    {
      "epoch": 0.018899045408340115,
      "grad_norm": 2.3334624767303467,
      "learning_rate": 2.997356902633179e-05,
      "loss": 1.9791,
      "step": 9705
    },
    {
      "epoch": 0.01892825567945038,
      "grad_norm": 3.4070332050323486,
      "learning_rate": 2.997348728412313e-05,
      "loss": 1.8621,
      "step": 9720
    },
    {
      "epoch": 0.018957465950560644,
      "grad_norm": 2.4055604934692383,
      "learning_rate": 2.997340541582062e-05,
      "loss": 1.8424,
      "step": 9735
    },
    {
      "epoch": 0.018986676221670905,
      "grad_norm": 3.209122896194458,
      "learning_rate": 2.9973323421424962e-05,
      "loss": 1.8791,
      "step": 9750
    },
    {
      "epoch": 0.01901588649278117,
      "grad_norm": 2.720518112182617,
      "learning_rate": 2.9973241300936842e-05,
      "loss": 1.9752,
      "step": 9765
    },
    {
      "epoch": 0.01904509676389143,
      "grad_norm": 1.9057116508483887,
      "learning_rate": 2.9973159054356948e-05,
      "loss": 1.9658,
      "step": 9780
    },
    {
      "epoch": 0.019074307035001695,
      "grad_norm": 3.4243197441101074,
      "learning_rate": 2.9973076681685977e-05,
      "loss": 1.9394,
      "step": 9795
    },
    {
      "epoch": 0.019103517306111956,
      "grad_norm": 3.557957410812378,
      "learning_rate": 2.997299418292462e-05,
      "loss": 1.8854,
      "step": 9810
    },
    {
      "epoch": 0.01913272757722222,
      "grad_norm": 2.576314926147461,
      "learning_rate": 2.9972911558073575e-05,
      "loss": 1.8975,
      "step": 9825
    },
    {
      "epoch": 0.019161937848332485,
      "grad_norm": 2.4958183765411377,
      "learning_rate": 2.9972828807133537e-05,
      "loss": 1.765,
      "step": 9840
    },
    {
      "epoch": 0.019191148119442746,
      "grad_norm": 3.1889865398406982,
      "learning_rate": 2.99727459301052e-05,
      "loss": 1.9785,
      "step": 9855
    },
    {
      "epoch": 0.01922035839055301,
      "grad_norm": 4.607937335968018,
      "learning_rate": 2.9972662926989267e-05,
      "loss": 1.7931,
      "step": 9870
    },
    {
      "epoch": 0.01924956866166327,
      "grad_norm": 3.380537271499634,
      "learning_rate": 2.997257979778643e-05,
      "loss": 1.8366,
      "step": 9885
    },
    {
      "epoch": 0.019278778932773535,
      "grad_norm": 1.7773466110229492,
      "learning_rate": 2.9972496542497393e-05,
      "loss": 1.841,
      "step": 9900
    },
    {
      "epoch": 0.019307989203883796,
      "grad_norm": 3.43685245513916,
      "learning_rate": 2.9972413161122858e-05,
      "loss": 1.8255,
      "step": 9915
    },
    {
      "epoch": 0.01933719947499406,
      "grad_norm": 4.238219261169434,
      "learning_rate": 2.9972329653663525e-05,
      "loss": 1.8403,
      "step": 9930
    },
    {
      "epoch": 0.019366409746104322,
      "grad_norm": 3.8438355922698975,
      "learning_rate": 2.99722460201201e-05,
      "loss": 1.8005,
      "step": 9945
    },
    {
      "epoch": 0.019395620017214586,
      "grad_norm": 4.286600589752197,
      "learning_rate": 2.997216226049328e-05,
      "loss": 1.8684,
      "step": 9960
    },
    {
      "epoch": 0.01942483028832485,
      "grad_norm": 1.7342430353164673,
      "learning_rate": 2.997207837478378e-05,
      "loss": 1.9657,
      "step": 9975
    },
    {
      "epoch": 0.01945404055943511,
      "grad_norm": 3.3315911293029785,
      "learning_rate": 2.9971994362992304e-05,
      "loss": 1.9156,
      "step": 9990
    },
    {
      "epoch": 0.019483250830545376,
      "grad_norm": 4.346848011016846,
      "learning_rate": 2.9971910225119556e-05,
      "loss": 1.9114,
      "step": 10005
    },
    {
      "epoch": 0.019512461101655637,
      "grad_norm": 2.8670308589935303,
      "learning_rate": 2.9971825961166248e-05,
      "loss": 1.8471,
      "step": 10020
    },
    {
      "epoch": 0.0195416713727659,
      "grad_norm": 2.1465935707092285,
      "learning_rate": 2.9971741571133085e-05,
      "loss": 1.8608,
      "step": 10035
    },
    {
      "epoch": 0.019570881643876162,
      "grad_norm": 3.3292319774627686,
      "learning_rate": 2.9971657055020782e-05,
      "loss": 1.877,
      "step": 10050
    },
    {
      "epoch": 0.019600091914986427,
      "grad_norm": 2.3058934211730957,
      "learning_rate": 2.9971572412830045e-05,
      "loss": 1.9125,
      "step": 10065
    },
    {
      "epoch": 0.01962930218609669,
      "grad_norm": 4.684175968170166,
      "learning_rate": 2.9971487644561597e-05,
      "loss": 1.9237,
      "step": 10080
    },
    {
      "epoch": 0.019658512457206952,
      "grad_norm": 3.6633639335632324,
      "learning_rate": 2.9971402750216144e-05,
      "loss": 1.856,
      "step": 10095
    },
    {
      "epoch": 0.019687722728317217,
      "grad_norm": 3.837944269180298,
      "learning_rate": 2.9971317729794404e-05,
      "loss": 2.0146,
      "step": 10110
    },
    {
      "epoch": 0.019716932999427478,
      "grad_norm": 4.053643226623535,
      "learning_rate": 2.997123258329709e-05,
      "loss": 1.9125,
      "step": 10125
    },
    {
      "epoch": 0.019746143270537742,
      "grad_norm": 1.9680250883102417,
      "learning_rate": 2.9971147310724923e-05,
      "loss": 1.9431,
      "step": 10140
    },
    {
      "epoch": 0.019775353541648003,
      "grad_norm": 1.941953420639038,
      "learning_rate": 2.9971061912078615e-05,
      "loss": 1.8638,
      "step": 10155
    },
    {
      "epoch": 0.019804563812758268,
      "grad_norm": 3.1855714321136475,
      "learning_rate": 2.997097638735889e-05,
      "loss": 1.6231,
      "step": 10170
    },
    {
      "epoch": 0.019833774083868532,
      "grad_norm": 4.022531509399414,
      "learning_rate": 2.997089073656647e-05,
      "loss": 1.8637,
      "step": 10185
    },
    {
      "epoch": 0.019862984354978793,
      "grad_norm": 4.320540904998779,
      "learning_rate": 2.997080495970207e-05,
      "loss": 1.8383,
      "step": 10200
    },
    {
      "epoch": 0.019892194626089058,
      "grad_norm": 3.0604958534240723,
      "learning_rate": 2.997071905676642e-05,
      "loss": 1.8843,
      "step": 10215
    },
    {
      "epoch": 0.01992140489719932,
      "grad_norm": 3.2216265201568604,
      "learning_rate": 2.9970633027760235e-05,
      "loss": 2.004,
      "step": 10230
    },
    {
      "epoch": 0.019950615168309583,
      "grad_norm": 2.4354753494262695,
      "learning_rate": 2.997054687268425e-05,
      "loss": 1.9255,
      "step": 10245
    },
    {
      "epoch": 0.019979825439419844,
      "grad_norm": 2.3556060791015625,
      "learning_rate": 2.9970460591539175e-05,
      "loss": 2.0436,
      "step": 10260
    },
    {
      "epoch": 0.02000903571053011,
      "grad_norm": 3.6977896690368652,
      "learning_rate": 2.9970374184325753e-05,
      "loss": 1.9181,
      "step": 10275
    },
    {
      "epoch": 0.02003824598164037,
      "grad_norm": 1.909177541732788,
      "learning_rate": 2.99702876510447e-05,
      "loss": 1.7653,
      "step": 10290
    },
    {
      "epoch": 0.020067456252750634,
      "grad_norm": 2.486943006515503,
      "learning_rate": 2.997020099169675e-05,
      "loss": 1.9137,
      "step": 10305
    },
    {
      "epoch": 0.020096666523860898,
      "grad_norm": 3.0627150535583496,
      "learning_rate": 2.9970114206282634e-05,
      "loss": 1.8442,
      "step": 10320
    },
    {
      "epoch": 0.02012587679497116,
      "grad_norm": 2.7819271087646484,
      "learning_rate": 2.997002729480308e-05,
      "loss": 2.1325,
      "step": 10335
    },
    {
      "epoch": 0.020155087066081424,
      "grad_norm": 2.097712755203247,
      "learning_rate": 2.9969940257258823e-05,
      "loss": 1.8952,
      "step": 10350
    },
    {
      "epoch": 0.020184297337191685,
      "grad_norm": 2.317915439605713,
      "learning_rate": 2.9969853093650592e-05,
      "loss": 1.8896,
      "step": 10365
    },
    {
      "epoch": 0.02021350760830195,
      "grad_norm": 3.3799221515655518,
      "learning_rate": 2.996976580397912e-05,
      "loss": 1.9032,
      "step": 10380
    },
    {
      "epoch": 0.02024271787941221,
      "grad_norm": 4.226128578186035,
      "learning_rate": 2.996967838824515e-05,
      "loss": 1.9812,
      "step": 10395
    },
    {
      "epoch": 0.020271928150522475,
      "grad_norm": 2.872182607650757,
      "learning_rate": 2.996959084644941e-05,
      "loss": 1.7335,
      "step": 10410
    },
    {
      "epoch": 0.02030113842163274,
      "grad_norm": 1.931477427482605,
      "learning_rate": 2.9969503178592638e-05,
      "loss": 1.8178,
      "step": 10425
    },
    {
      "epoch": 0.020330348692743,
      "grad_norm": 1.9570348262786865,
      "learning_rate": 2.9969415384675577e-05,
      "loss": 1.9652,
      "step": 10440
    },
    {
      "epoch": 0.020359558963853264,
      "grad_norm": 2.1548566818237305,
      "learning_rate": 2.996932746469896e-05,
      "loss": 1.9105,
      "step": 10455
    },
    {
      "epoch": 0.020388769234963525,
      "grad_norm": 2.138561248779297,
      "learning_rate": 2.9969239418663538e-05,
      "loss": 2.2595,
      "step": 10470
    },
    {
      "epoch": 0.02041797950607379,
      "grad_norm": 3.682020425796509,
      "learning_rate": 2.9969151246570038e-05,
      "loss": 1.8828,
      "step": 10485
    },
    {
      "epoch": 0.02044718977718405,
      "grad_norm": 1.9086365699768066,
      "learning_rate": 2.9969062948419213e-05,
      "loss": 1.8974,
      "step": 10500
    },
    {
      "epoch": 0.020476400048294315,
      "grad_norm": 3.4526236057281494,
      "learning_rate": 2.9968974524211807e-05,
      "loss": 1.8972,
      "step": 10515
    },
    {
      "epoch": 0.020505610319404576,
      "grad_norm": 5.137340068817139,
      "learning_rate": 2.996888597394856e-05,
      "loss": 1.7885,
      "step": 10530
    },
    {
      "epoch": 0.02053482059051484,
      "grad_norm": 2.410789966583252,
      "learning_rate": 2.9968797297630215e-05,
      "loss": 1.9339,
      "step": 10545
    },
    {
      "epoch": 0.020564030861625105,
      "grad_norm": 2.756986618041992,
      "learning_rate": 2.9968708495257527e-05,
      "loss": 1.9682,
      "step": 10560
    },
    {
      "epoch": 0.020593241132735366,
      "grad_norm": 4.116410255432129,
      "learning_rate": 2.9968619566831238e-05,
      "loss": 1.8814,
      "step": 10575
    },
    {
      "epoch": 0.02062245140384563,
      "grad_norm": 3.892730712890625,
      "learning_rate": 2.9968530512352098e-05,
      "loss": 1.9744,
      "step": 10590
    },
    {
      "epoch": 0.02065166167495589,
      "grad_norm": 3.5503933429718018,
      "learning_rate": 2.9968441331820856e-05,
      "loss": 2.0472,
      "step": 10605
    },
    {
      "epoch": 0.020680871946066156,
      "grad_norm": 4.085498809814453,
      "learning_rate": 2.9968352025238263e-05,
      "loss": 1.9428,
      "step": 10620
    },
    {
      "epoch": 0.020710082217176417,
      "grad_norm": 3.806868553161621,
      "learning_rate": 2.996826259260508e-05,
      "loss": 1.8212,
      "step": 10635
    },
    {
      "epoch": 0.02073929248828668,
      "grad_norm": 2.24172306060791,
      "learning_rate": 2.9968173033922045e-05,
      "loss": 1.7338,
      "step": 10650
    },
    {
      "epoch": 0.020768502759396946,
      "grad_norm": 2.516962766647339,
      "learning_rate": 2.996808334918992e-05,
      "loss": 1.9185,
      "step": 10665
    },
    {
      "epoch": 0.020797713030507207,
      "grad_norm": 3.8772926330566406,
      "learning_rate": 2.9967993538409465e-05,
      "loss": 1.8053,
      "step": 10680
    },
    {
      "epoch": 0.02082692330161747,
      "grad_norm": 5.146918296813965,
      "learning_rate": 2.9967903601581427e-05,
      "loss": 1.8488,
      "step": 10695
    },
    {
      "epoch": 0.020856133572727732,
      "grad_norm": 6.216543197631836,
      "learning_rate": 2.9967813538706568e-05,
      "loss": 1.9446,
      "step": 10710
    },
    {
      "epoch": 0.020885343843837997,
      "grad_norm": 3.0393502712249756,
      "learning_rate": 2.9967723349785648e-05,
      "loss": 1.8881,
      "step": 10725
    },
    {
      "epoch": 0.020914554114948258,
      "grad_norm": 2.7038638591766357,
      "learning_rate": 2.996763303481942e-05,
      "loss": 1.9454,
      "step": 10740
    },
    {
      "epoch": 0.020943764386058522,
      "grad_norm": 2.4057178497314453,
      "learning_rate": 2.9967542593808655e-05,
      "loss": 2.0256,
      "step": 10755
    },
    {
      "epoch": 0.020972974657168783,
      "grad_norm": 2.2479588985443115,
      "learning_rate": 2.9967452026754104e-05,
      "loss": 1.8835,
      "step": 10770
    },
    {
      "epoch": 0.021002184928279048,
      "grad_norm": 2.2106525897979736,
      "learning_rate": 2.996736133365654e-05,
      "loss": 1.8594,
      "step": 10785
    },
    {
      "epoch": 0.021031395199389312,
      "grad_norm": 2.273165225982666,
      "learning_rate": 2.9967270514516718e-05,
      "loss": 1.772,
      "step": 10800
    },
    {
      "epoch": 0.021060605470499573,
      "grad_norm": 4.13364315032959,
      "learning_rate": 2.9967179569335407e-05,
      "loss": 1.8364,
      "step": 10815
    },
    {
      "epoch": 0.021089815741609837,
      "grad_norm": 4.664285182952881,
      "learning_rate": 2.9967088498113368e-05,
      "loss": 1.7126,
      "step": 10830
    },
    {
      "epoch": 0.0211190260127201,
      "grad_norm": 3.9353389739990234,
      "learning_rate": 2.9966997300851376e-05,
      "loss": 1.8366,
      "step": 10845
    },
    {
      "epoch": 0.021148236283830363,
      "grad_norm": 3.433561086654663,
      "learning_rate": 2.996690597755019e-05,
      "loss": 1.7921,
      "step": 10860
    },
    {
      "epoch": 0.021177446554940624,
      "grad_norm": 2.0325567722320557,
      "learning_rate": 2.996681452821059e-05,
      "loss": 1.88,
      "step": 10875
    },
    {
      "epoch": 0.021206656826050888,
      "grad_norm": 3.6523592472076416,
      "learning_rate": 2.9966722952833335e-05,
      "loss": 1.8464,
      "step": 10890
    },
    {
      "epoch": 0.021235867097161153,
      "grad_norm": 4.08983039855957,
      "learning_rate": 2.99666312514192e-05,
      "loss": 2.0643,
      "step": 10905
    },
    {
      "epoch": 0.021265077368271414,
      "grad_norm": 4.714212894439697,
      "learning_rate": 2.9966539423968964e-05,
      "loss": 1.9458,
      "step": 10920
    },
    {
      "epoch": 0.021294287639381678,
      "grad_norm": 3.2425897121429443,
      "learning_rate": 2.996644747048339e-05,
      "loss": 1.9802,
      "step": 10935
    },
    {
      "epoch": 0.02132349791049194,
      "grad_norm": 5.025219440460205,
      "learning_rate": 2.9966355390963258e-05,
      "loss": 1.8679,
      "step": 10950
    },
    {
      "epoch": 0.021352708181602204,
      "grad_norm": 3.669241428375244,
      "learning_rate": 2.9966263185409343e-05,
      "loss": 1.9634,
      "step": 10965
    },
    {
      "epoch": 0.021381918452712464,
      "grad_norm": 3.5736284255981445,
      "learning_rate": 2.996617085382242e-05,
      "loss": 1.9348,
      "step": 10980
    },
    {
      "epoch": 0.02141112872382273,
      "grad_norm": 2.8263094425201416,
      "learning_rate": 2.996607839620327e-05,
      "loss": 1.7897,
      "step": 10995
    },
    {
      "epoch": 0.02144033899493299,
      "grad_norm": 4.033946990966797,
      "learning_rate": 2.996598581255267e-05,
      "loss": 1.9587,
      "step": 11010
    },
    {
      "epoch": 0.021469549266043254,
      "grad_norm": 3.7127420902252197,
      "learning_rate": 2.996589310287139e-05,
      "loss": 1.874,
      "step": 11025
    },
    {
      "epoch": 0.02149875953715352,
      "grad_norm": 4.083348751068115,
      "learning_rate": 2.9965800267160223e-05,
      "loss": 1.7831,
      "step": 11040
    },
    {
      "epoch": 0.02152796980826378,
      "grad_norm": 4.666345596313477,
      "learning_rate": 2.996570730541995e-05,
      "loss": 1.8516,
      "step": 11055
    },
    {
      "epoch": 0.021557180079374044,
      "grad_norm": 2.3497631549835205,
      "learning_rate": 2.996561421765135e-05,
      "loss": 1.9213,
      "step": 11070
    },
    {
      "epoch": 0.021586390350484305,
      "grad_norm": 4.355146408081055,
      "learning_rate": 2.996552100385521e-05,
      "loss": 1.8072,
      "step": 11085
    },
    {
      "epoch": 0.02161560062159457,
      "grad_norm": 3.6673879623413086,
      "learning_rate": 2.996542766403231e-05,
      "loss": 1.8541,
      "step": 11100
    },
    {
      "epoch": 0.02164481089270483,
      "grad_norm": 2.2446086406707764,
      "learning_rate": 2.996533419818344e-05,
      "loss": 1.9764,
      "step": 11115
    },
    {
      "epoch": 0.021674021163815095,
      "grad_norm": 3.8175010681152344,
      "learning_rate": 2.996524060630938e-05,
      "loss": 1.9664,
      "step": 11130
    },
    {
      "epoch": 0.02170323143492536,
      "grad_norm": 4.360842704772949,
      "learning_rate": 2.996514688841093e-05,
      "loss": 1.9145,
      "step": 11145
    },
    {
      "epoch": 0.02173244170603562,
      "grad_norm": 3.110860586166382,
      "learning_rate": 2.996505304448887e-05,
      "loss": 1.8939,
      "step": 11160
    },
    {
      "epoch": 0.021761651977145885,
      "grad_norm": 2.5407495498657227,
      "learning_rate": 2.9964959074544e-05,
      "loss": 1.8649,
      "step": 11175
    },
    {
      "epoch": 0.021790862248256146,
      "grad_norm": 2.933225393295288,
      "learning_rate": 2.9964864978577103e-05,
      "loss": 1.8987,
      "step": 11190
    },
    {
      "epoch": 0.02182007251936641,
      "grad_norm": 3.003664493560791,
      "learning_rate": 2.996477075658897e-05,
      "loss": 1.936,
      "step": 11205
    },
    {
      "epoch": 0.02184928279047667,
      "grad_norm": 3.119703531265259,
      "learning_rate": 2.99646764085804e-05,
      "loss": 1.7183,
      "step": 11220
    },
    {
      "epoch": 0.021878493061586936,
      "grad_norm": 2.6627697944641113,
      "learning_rate": 2.9964581934552182e-05,
      "loss": 1.9218,
      "step": 11235
    },
    {
      "epoch": 0.0219077033326972,
      "grad_norm": 4.310539245605469,
      "learning_rate": 2.9964487334505114e-05,
      "loss": 2.0666,
      "step": 11250
    },
    {
      "epoch": 0.02193691360380746,
      "grad_norm": 2.611443519592285,
      "learning_rate": 2.9964392608439997e-05,
      "loss": 1.8154,
      "step": 11265
    },
    {
      "epoch": 0.021966123874917726,
      "grad_norm": 3.391406774520874,
      "learning_rate": 2.996429775635763e-05,
      "loss": 1.9308,
      "step": 11280
    },
    {
      "epoch": 0.021995334146027987,
      "grad_norm": 3.1492297649383545,
      "learning_rate": 2.9964202778258797e-05,
      "loss": 1.939,
      "step": 11295
    },
    {
      "epoch": 0.02202454441713825,
      "grad_norm": 3.680859088897705,
      "learning_rate": 2.9964107674144313e-05,
      "loss": 1.9048,
      "step": 11310
    },
    {
      "epoch": 0.022053754688248512,
      "grad_norm": 2.0522656440734863,
      "learning_rate": 2.9964012444014972e-05,
      "loss": 1.9477,
      "step": 11325
    },
    {
      "epoch": 0.022082964959358777,
      "grad_norm": 3.265316963195801,
      "learning_rate": 2.996391708787158e-05,
      "loss": 1.8116,
      "step": 11340
    },
    {
      "epoch": 0.022112175230469037,
      "grad_norm": 2.0570802688598633,
      "learning_rate": 2.9963821605714934e-05,
      "loss": 1.8493,
      "step": 11355
    },
    {
      "epoch": 0.022141385501579302,
      "grad_norm": 4.1805877685546875,
      "learning_rate": 2.9963725997545844e-05,
      "loss": 1.8909,
      "step": 11370
    },
    {
      "epoch": 0.022170595772689566,
      "grad_norm": 2.6846070289611816,
      "learning_rate": 2.9963630263365116e-05,
      "loss": 1.8424,
      "step": 11385
    },
    {
      "epoch": 0.022199806043799827,
      "grad_norm": 4.480174541473389,
      "learning_rate": 2.996353440317355e-05,
      "loss": 1.919,
      "step": 11400
    },
    {
      "epoch": 0.022229016314910092,
      "grad_norm": 2.179137706756592,
      "learning_rate": 2.996343841697195e-05,
      "loss": 1.8023,
      "step": 11415
    },
    {
      "epoch": 0.022258226586020353,
      "grad_norm": 4.341340065002441,
      "learning_rate": 2.996334230476114e-05,
      "loss": 1.8263,
      "step": 11430
    },
    {
      "epoch": 0.022287436857130617,
      "grad_norm": 1.6666501760482788,
      "learning_rate": 2.9963246066541913e-05,
      "loss": 1.8854,
      "step": 11445
    },
    {
      "epoch": 0.022316647128240878,
      "grad_norm": 2.9793460369110107,
      "learning_rate": 2.9963149702315093e-05,
      "loss": 1.9214,
      "step": 11460
    },
    {
      "epoch": 0.022345857399351143,
      "grad_norm": 4.338296413421631,
      "learning_rate": 2.996305321208148e-05,
      "loss": 1.9517,
      "step": 11475
    },
    {
      "epoch": 0.022375067670461407,
      "grad_norm": 4.449549674987793,
      "learning_rate": 2.99629565958419e-05,
      "loss": 2.0761,
      "step": 11490
    },
    {
      "epoch": 0.022404277941571668,
      "grad_norm": 2.6661598682403564,
      "learning_rate": 2.9962859853597146e-05,
      "loss": 1.9029,
      "step": 11505
    },
    {
      "epoch": 0.022433488212681933,
      "grad_norm": 2.5066778659820557,
      "learning_rate": 2.996276298534805e-05,
      "loss": 1.7934,
      "step": 11520
    },
    {
      "epoch": 0.022462698483792193,
      "grad_norm": 2.2089765071868896,
      "learning_rate": 2.9962665991095424e-05,
      "loss": 2.0755,
      "step": 11535
    },
    {
      "epoch": 0.022491908754902458,
      "grad_norm": 3.543388605117798,
      "learning_rate": 2.9962568870840078e-05,
      "loss": 1.8682,
      "step": 11550
    },
    {
      "epoch": 0.02252111902601272,
      "grad_norm": 3.1596784591674805,
      "learning_rate": 2.9962471624582838e-05,
      "loss": 2.0225,
      "step": 11565
    },
    {
      "epoch": 0.022550329297122983,
      "grad_norm": 3.338447332382202,
      "learning_rate": 2.9962374252324524e-05,
      "loss": 1.7248,
      "step": 11580
    },
    {
      "epoch": 0.022579539568233244,
      "grad_norm": 2.6090617179870605,
      "learning_rate": 2.996227675406595e-05,
      "loss": 2.1125,
      "step": 11595
    },
    {
      "epoch": 0.02260874983934351,
      "grad_norm": 3.618283271789551,
      "learning_rate": 2.9962179129807936e-05,
      "loss": 2.0012,
      "step": 11610
    },
    {
      "epoch": 0.022637960110453773,
      "grad_norm": 2.161893129348755,
      "learning_rate": 2.996208137955131e-05,
      "loss": 1.8541,
      "step": 11625
    },
    {
      "epoch": 0.022667170381564034,
      "grad_norm": 4.789167881011963,
      "learning_rate": 2.996198350329689e-05,
      "loss": 1.7935,
      "step": 11640
    },
    {
      "epoch": 0.0226963806526743,
      "grad_norm": 2.7407031059265137,
      "learning_rate": 2.9961885501045505e-05,
      "loss": 1.8944,
      "step": 11655
    },
    {
      "epoch": 0.02272559092378456,
      "grad_norm": 2.0403008460998535,
      "learning_rate": 2.9961787372797977e-05,
      "loss": 1.9245,
      "step": 11670
    },
    {
      "epoch": 0.022754801194894824,
      "grad_norm": 2.2842493057250977,
      "learning_rate": 2.996168911855513e-05,
      "loss": 1.8867,
      "step": 11685
    },
    {
      "epoch": 0.022784011466005085,
      "grad_norm": 3.4035165309906006,
      "learning_rate": 2.99615907383178e-05,
      "loss": 1.8029,
      "step": 11700
    },
    {
      "epoch": 0.02281322173711535,
      "grad_norm": 4.04712438583374,
      "learning_rate": 2.996149223208681e-05,
      "loss": 1.8933,
      "step": 11715
    },
    {
      "epoch": 0.022842432008225614,
      "grad_norm": 3.2550394535064697,
      "learning_rate": 2.996139359986299e-05,
      "loss": 1.8503,
      "step": 11730
    },
    {
      "epoch": 0.022871642279335875,
      "grad_norm": 2.3266172409057617,
      "learning_rate": 2.9961294841647164e-05,
      "loss": 1.939,
      "step": 11745
    },
    {
      "epoch": 0.02290085255044614,
      "grad_norm": 1.7769925594329834,
      "learning_rate": 2.9961195957440172e-05,
      "loss": 2.0939,
      "step": 11760
    },
    {
      "epoch": 0.0229300628215564,
      "grad_norm": 3.4839985370635986,
      "learning_rate": 2.9961096947242846e-05,
      "loss": 1.8933,
      "step": 11775
    },
    {
      "epoch": 0.022959273092666665,
      "grad_norm": 4.051612377166748,
      "learning_rate": 2.9960997811056017e-05,
      "loss": 1.8464,
      "step": 11790
    },
    {
      "epoch": 0.022988483363776926,
      "grad_norm": 5.2628703117370605,
      "learning_rate": 2.9960898548880525e-05,
      "loss": 1.9296,
      "step": 11805
    },
    {
      "epoch": 0.02301769363488719,
      "grad_norm": 4.6693434715271,
      "learning_rate": 2.99607991607172e-05,
      "loss": 2.0126,
      "step": 11820
    },
    {
      "epoch": 0.02304690390599745,
      "grad_norm": 2.2805821895599365,
      "learning_rate": 2.996069964656688e-05,
      "loss": 1.7084,
      "step": 11835
    },
    {
      "epoch": 0.023076114177107716,
      "grad_norm": 5.130448818206787,
      "learning_rate": 2.996060000643041e-05,
      "loss": 1.8437,
      "step": 11850
    },
    {
      "epoch": 0.02310532444821798,
      "grad_norm": 3.6057186126708984,
      "learning_rate": 2.9960500240308616e-05,
      "loss": 2.0847,
      "step": 11865
    },
    {
      "epoch": 0.02313453471932824,
      "grad_norm": 3.6711442470550537,
      "learning_rate": 2.9960400348202348e-05,
      "loss": 1.9817,
      "step": 11880
    },
    {
      "epoch": 0.023163744990438506,
      "grad_norm": 2.514784097671509,
      "learning_rate": 2.9960300330112445e-05,
      "loss": 1.7633,
      "step": 11895
    },
    {
      "epoch": 0.023192955261548766,
      "grad_norm": 2.1372082233428955,
      "learning_rate": 2.996020018603975e-05,
      "loss": 1.7941,
      "step": 11910
    },
    {
      "epoch": 0.02322216553265903,
      "grad_norm": 2.8542985916137695,
      "learning_rate": 2.9960099915985104e-05,
      "loss": 1.7778,
      "step": 11925
    },
    {
      "epoch": 0.023251375803769292,
      "grad_norm": 2.1213855743408203,
      "learning_rate": 2.9959999519949354e-05,
      "loss": 1.8966,
      "step": 11940
    },
    {
      "epoch": 0.023280586074879556,
      "grad_norm": 2.751647710800171,
      "learning_rate": 2.995989899793334e-05,
      "loss": 1.7919,
      "step": 11955
    },
    {
      "epoch": 0.02330979634598982,
      "grad_norm": 2.2502388954162598,
      "learning_rate": 2.9959798349937915e-05,
      "loss": 1.8062,
      "step": 11970
    },
    {
      "epoch": 0.023339006617100082,
      "grad_norm": 2.5316834449768066,
      "learning_rate": 2.995969757596392e-05,
      "loss": 1.9685,
      "step": 11985
    },
    {
      "epoch": 0.023368216888210346,
      "grad_norm": 2.8897788524627686,
      "learning_rate": 2.995959667601221e-05,
      "loss": 1.9659,
      "step": 12000
    },
    {
      "epoch": 0.023397427159320607,
      "grad_norm": 3.8147826194763184,
      "learning_rate": 2.9959495650083634e-05,
      "loss": 1.7452,
      "step": 12015
    },
    {
      "epoch": 0.02342663743043087,
      "grad_norm": 1.9649070501327515,
      "learning_rate": 2.9959394498179043e-05,
      "loss": 1.7987,
      "step": 12030
    },
    {
      "epoch": 0.023455847701541133,
      "grad_norm": 2.2376818656921387,
      "learning_rate": 2.9959293220299287e-05,
      "loss": 1.87,
      "step": 12045
    },
    {
      "epoch": 0.023485057972651397,
      "grad_norm": 2.3533129692077637,
      "learning_rate": 2.9959191816445217e-05,
      "loss": 2.006,
      "step": 12060
    },
    {
      "epoch": 0.023514268243761658,
      "grad_norm": 2.5106008052825928,
      "learning_rate": 2.9959090286617686e-05,
      "loss": 1.8053,
      "step": 12075
    },
    {
      "epoch": 0.023543478514871923,
      "grad_norm": 2.843824863433838,
      "learning_rate": 2.9958988630817555e-05,
      "loss": 2.0062,
      "step": 12090
    },
    {
      "epoch": 0.023572688785982187,
      "grad_norm": 3.826493501663208,
      "learning_rate": 2.9958886849045678e-05,
      "loss": 1.8213,
      "step": 12105
    },
    {
      "epoch": 0.023601899057092448,
      "grad_norm": 2.669509172439575,
      "learning_rate": 2.9958784941302908e-05,
      "loss": 2.0031,
      "step": 12120
    },
    {
      "epoch": 0.023631109328202712,
      "grad_norm": 5.227283477783203,
      "learning_rate": 2.995868290759011e-05,
      "loss": 1.9724,
      "step": 12135
    },
    {
      "epoch": 0.023660319599312973,
      "grad_norm": 2.67545485496521,
      "learning_rate": 2.9958580747908134e-05,
      "loss": 1.7393,
      "step": 12150
    },
    {
      "epoch": 0.023689529870423238,
      "grad_norm": 4.294065952301025,
      "learning_rate": 2.9958478462257847e-05,
      "loss": 2.0262,
      "step": 12165
    },
    {
      "epoch": 0.0237187401415335,
      "grad_norm": 4.024529933929443,
      "learning_rate": 2.9958376050640114e-05,
      "loss": 1.8878,
      "step": 12180
    },
    {
      "epoch": 0.023747950412643763,
      "grad_norm": 3.88948130607605,
      "learning_rate": 2.9958273513055785e-05,
      "loss": 1.9365,
      "step": 12195
    },
    {
      "epoch": 0.023777160683754028,
      "grad_norm": 2.3247134685516357,
      "learning_rate": 2.9958170849505736e-05,
      "loss": 1.9275,
      "step": 12210
    },
    {
      "epoch": 0.02380637095486429,
      "grad_norm": 2.8689277172088623,
      "learning_rate": 2.9958068059990827e-05,
      "loss": 1.9015,
      "step": 12225
    },
    {
      "epoch": 0.023835581225974553,
      "grad_norm": 2.650768756866455,
      "learning_rate": 2.995796514451192e-05,
      "loss": 2.0453,
      "step": 12240
    },
    {
      "epoch": 0.023864791497084814,
      "grad_norm": 3.2005455493927,
      "learning_rate": 2.9957862103069886e-05,
      "loss": 1.8641,
      "step": 12255
    },
    {
      "epoch": 0.02389400176819508,
      "grad_norm": 3.2274022102355957,
      "learning_rate": 2.9957758935665592e-05,
      "loss": 2.0379,
      "step": 12270
    },
    {
      "epoch": 0.02392321203930534,
      "grad_norm": 2.933011054992676,
      "learning_rate": 2.9957655642299903e-05,
      "loss": 2.0304,
      "step": 12285
    },
    {
      "epoch": 0.023952422310415604,
      "grad_norm": 2.051677942276001,
      "learning_rate": 2.9957552222973696e-05,
      "loss": 1.9656,
      "step": 12300
    },
    {
      "epoch": 0.02398163258152587,
      "grad_norm": 2.488805055618286,
      "learning_rate": 2.995744867768784e-05,
      "loss": 1.8997,
      "step": 12315
    },
    {
      "epoch": 0.02401084285263613,
      "grad_norm": 2.1613032817840576,
      "learning_rate": 2.99573450064432e-05,
      "loss": 1.9733,
      "step": 12330
    },
    {
      "epoch": 0.024040053123746394,
      "grad_norm": 3.817368745803833,
      "learning_rate": 2.9957241209240656e-05,
      "loss": 1.74,
      "step": 12345
    },
    {
      "epoch": 0.024069263394856655,
      "grad_norm": 2.048835515975952,
      "learning_rate": 2.995713728608108e-05,
      "loss": 2.0201,
      "step": 12360
    },
    {
      "epoch": 0.02409847366596692,
      "grad_norm": 3.2480578422546387,
      "learning_rate": 2.995703323696535e-05,
      "loss": 1.8652,
      "step": 12375
    },
    {
      "epoch": 0.02412768393707718,
      "grad_norm": 2.2918214797973633,
      "learning_rate": 2.9956929061894334e-05,
      "loss": 1.9089,
      "step": 12390
    },
    {
      "epoch": 0.024156894208187445,
      "grad_norm": 2.4332637786865234,
      "learning_rate": 2.995682476086892e-05,
      "loss": 1.8168,
      "step": 12405
    },
    {
      "epoch": 0.024186104479297706,
      "grad_norm": 3.9398860931396484,
      "learning_rate": 2.9956720333889978e-05,
      "loss": 1.8114,
      "step": 12420
    },
    {
      "epoch": 0.02421531475040797,
      "grad_norm": 1.3877410888671875,
      "learning_rate": 2.995661578095839e-05,
      "loss": 1.7609,
      "step": 12435
    },
    {
      "epoch": 0.024244525021518235,
      "grad_norm": 3.1245439052581787,
      "learning_rate": 2.9956511102075043e-05,
      "loss": 1.9191,
      "step": 12450
    },
    {
      "epoch": 0.024273735292628495,
      "grad_norm": 4.245307445526123,
      "learning_rate": 2.9956406297240805e-05,
      "loss": 1.9551,
      "step": 12465
    },
    {
      "epoch": 0.02430294556373876,
      "grad_norm": 3.033841371536255,
      "learning_rate": 2.995630136645657e-05,
      "loss": 2.063,
      "step": 12480
    },
    {
      "epoch": 0.02433215583484902,
      "grad_norm": 2.9336037635803223,
      "learning_rate": 2.9956196309723217e-05,
      "loss": 1.8383,
      "step": 12495
    },
    {
      "epoch": 0.024361366105959285,
      "grad_norm": 2.946824312210083,
      "learning_rate": 2.9956091127041628e-05,
      "loss": 2.012,
      "step": 12510
    },
    {
      "epoch": 0.024390576377069546,
      "grad_norm": 1.9165891408920288,
      "learning_rate": 2.9955985818412695e-05,
      "loss": 1.8422,
      "step": 12525
    },
    {
      "epoch": 0.02441978664817981,
      "grad_norm": 2.4142305850982666,
      "learning_rate": 2.9955880383837304e-05,
      "loss": 2.027,
      "step": 12540
    },
    {
      "epoch": 0.024448996919290075,
      "grad_norm": 3.1490895748138428,
      "learning_rate": 2.9955774823316337e-05,
      "loss": 1.8485,
      "step": 12555
    },
    {
      "epoch": 0.024478207190400336,
      "grad_norm": 3.187546968460083,
      "learning_rate": 2.995566913685069e-05,
      "loss": 1.8955,
      "step": 12570
    },
    {
      "epoch": 0.0245074174615106,
      "grad_norm": 4.743402481079102,
      "learning_rate": 2.9955563324441246e-05,
      "loss": 1.8438,
      "step": 12585
    },
    {
      "epoch": 0.02453662773262086,
      "grad_norm": 4.372682094573975,
      "learning_rate": 2.9955457386088904e-05,
      "loss": 1.6814,
      "step": 12600
    },
    {
      "epoch": 0.024565838003731126,
      "grad_norm": 4.228031158447266,
      "learning_rate": 2.995535132179455e-05,
      "loss": 1.8358,
      "step": 12615
    },
    {
      "epoch": 0.024595048274841387,
      "grad_norm": 1.7856305837631226,
      "learning_rate": 2.9955245131559078e-05,
      "loss": 1.9314,
      "step": 12630
    },
    {
      "epoch": 0.02462425854595165,
      "grad_norm": 2.232226610183716,
      "learning_rate": 2.9955138815383383e-05,
      "loss": 1.7662,
      "step": 12645
    },
    {
      "epoch": 0.024653468817061912,
      "grad_norm": 2.812988758087158,
      "learning_rate": 2.9955032373268366e-05,
      "loss": 1.8475,
      "step": 12660
    },
    {
      "epoch": 0.024682679088172177,
      "grad_norm": 4.58977746963501,
      "learning_rate": 2.995492580521491e-05,
      "loss": 1.8952,
      "step": 12675
    },
    {
      "epoch": 0.02471188935928244,
      "grad_norm": 2.619033098220825,
      "learning_rate": 2.995481911122393e-05,
      "loss": 1.7343,
      "step": 12690
    },
    {
      "epoch": 0.024741099630392702,
      "grad_norm": 2.20595645904541,
      "learning_rate": 2.9954712291296303e-05,
      "loss": 1.8573,
      "step": 12705
    },
    {
      "epoch": 0.024770309901502967,
      "grad_norm": 4.261920928955078,
      "learning_rate": 2.9954605345432948e-05,
      "loss": 1.9771,
      "step": 12720
    },
    {
      "epoch": 0.024799520172613228,
      "grad_norm": 4.171009063720703,
      "learning_rate": 2.995449827363476e-05,
      "loss": 1.8328,
      "step": 12735
    },
    {
      "epoch": 0.024828730443723492,
      "grad_norm": 1.9221203327178955,
      "learning_rate": 2.9954391075902634e-05,
      "loss": 2.0122,
      "step": 12750
    },
    {
      "epoch": 0.024857940714833753,
      "grad_norm": 4.801123142242432,
      "learning_rate": 2.9954283752237478e-05,
      "loss": 1.9288,
      "step": 12765
    },
    {
      "epoch": 0.024887150985944018,
      "grad_norm": 5.146918296813965,
      "learning_rate": 2.99541763026402e-05,
      "loss": 1.8986,
      "step": 12780
    },
    {
      "epoch": 0.024916361257054282,
      "grad_norm": 3.8306210041046143,
      "learning_rate": 2.9954068727111694e-05,
      "loss": 1.9432,
      "step": 12795
    },
    {
      "epoch": 0.024945571528164543,
      "grad_norm": 3.569969892501831,
      "learning_rate": 2.9953961025652875e-05,
      "loss": 1.802,
      "step": 12810
    },
    {
      "epoch": 0.024974781799274808,
      "grad_norm": 2.11238956451416,
      "learning_rate": 2.995385319826465e-05,
      "loss": 1.7489,
      "step": 12825
    },
    {
      "epoch": 0.02500399207038507,
      "grad_norm": 2.0586655139923096,
      "learning_rate": 2.995374524494792e-05,
      "loss": 1.8308,
      "step": 12840
    },
    {
      "epoch": 0.025033202341495333,
      "grad_norm": 2.133302688598633,
      "learning_rate": 2.9953637165703597e-05,
      "loss": 1.8642,
      "step": 12855
    },
    {
      "epoch": 0.025062412612605594,
      "grad_norm": 6.014092922210693,
      "learning_rate": 2.9953528960532594e-05,
      "loss": 1.838,
      "step": 12870
    },
    {
      "epoch": 0.02509162288371586,
      "grad_norm": 4.038995742797852,
      "learning_rate": 2.9953420629435823e-05,
      "loss": 1.9485,
      "step": 12885
    },
    {
      "epoch": 0.02512083315482612,
      "grad_norm": 2.016037940979004,
      "learning_rate": 2.995331217241419e-05,
      "loss": 1.8532,
      "step": 12900
    },
    {
      "epoch": 0.025150043425936384,
      "grad_norm": 2.268634796142578,
      "learning_rate": 2.9953203589468617e-05,
      "loss": 1.7899,
      "step": 12915
    },
    {
      "epoch": 0.025179253697046648,
      "grad_norm": 4.676908016204834,
      "learning_rate": 2.995309488060001e-05,
      "loss": 1.875,
      "step": 12930
    },
    {
      "epoch": 0.02520846396815691,
      "grad_norm": 2.0081729888916016,
      "learning_rate": 2.9952986045809284e-05,
      "loss": 1.9863,
      "step": 12945
    },
    {
      "epoch": 0.025237674239267174,
      "grad_norm": 2.9928226470947266,
      "learning_rate": 2.9952877085097364e-05,
      "loss": 1.8073,
      "step": 12960
    },
    {
      "epoch": 0.025266884510377435,
      "grad_norm": 3.514173746109009,
      "learning_rate": 2.9952767998465164e-05,
      "loss": 1.849,
      "step": 12975
    },
    {
      "epoch": 0.0252960947814877,
      "grad_norm": 1.9361449480056763,
      "learning_rate": 2.99526587859136e-05,
      "loss": 1.8925,
      "step": 12990
    },
    {
      "epoch": 0.02532530505259796,
      "grad_norm": 4.396752834320068,
      "learning_rate": 2.9952549447443595e-05,
      "loss": 1.8844,
      "step": 13005
    },
    {
      "epoch": 0.025354515323708225,
      "grad_norm": 2.9122262001037598,
      "learning_rate": 2.9952439983056066e-05,
      "loss": 1.9564,
      "step": 13020
    },
    {
      "epoch": 0.02538372559481849,
      "grad_norm": 2.6632208824157715,
      "learning_rate": 2.9952330392751935e-05,
      "loss": 1.8514,
      "step": 13035
    },
    {
      "epoch": 0.02541293586592875,
      "grad_norm": 2.678126573562622,
      "learning_rate": 2.995222067653213e-05,
      "loss": 1.8146,
      "step": 13050
    },
    {
      "epoch": 0.025442146137039014,
      "grad_norm": 3.901380777359009,
      "learning_rate": 2.9952110834397572e-05,
      "loss": 2.0268,
      "step": 13065
    },
    {
      "epoch": 0.025471356408149275,
      "grad_norm": 2.7066125869750977,
      "learning_rate": 2.9952000866349185e-05,
      "loss": 1.8731,
      "step": 13080
    },
    {
      "epoch": 0.02550056667925954,
      "grad_norm": 4.8974456787109375,
      "learning_rate": 2.9951890772387897e-05,
      "loss": 1.908,
      "step": 13095
    },
    {
      "epoch": 0.0255297769503698,
      "grad_norm": 3.9215996265411377,
      "learning_rate": 2.995178055251463e-05,
      "loss": 2.0428,
      "step": 13110
    },
    {
      "epoch": 0.025558987221480065,
      "grad_norm": 2.519742250442505,
      "learning_rate": 2.9951670206730318e-05,
      "loss": 1.7794,
      "step": 13125
    },
    {
      "epoch": 0.025588197492590326,
      "grad_norm": 2.1564877033233643,
      "learning_rate": 2.995155973503589e-05,
      "loss": 1.6714,
      "step": 13140
    },
    {
      "epoch": 0.02561740776370059,
      "grad_norm": 4.537586688995361,
      "learning_rate": 2.9951449137432275e-05,
      "loss": 2.0812,
      "step": 13155
    },
    {
      "epoch": 0.025646618034810855,
      "grad_norm": 3.659740686416626,
      "learning_rate": 2.9951338413920403e-05,
      "loss": 1.995,
      "step": 13170
    },
    {
      "epoch": 0.025675828305921116,
      "grad_norm": 2.057332992553711,
      "learning_rate": 2.9951227564501207e-05,
      "loss": 1.8206,
      "step": 13185
    },
    {
      "epoch": 0.02570503857703138,
      "grad_norm": 3.4197821617126465,
      "learning_rate": 2.995111658917562e-05,
      "loss": 1.9727,
      "step": 13200
    },
    {
      "epoch": 0.02573424884814164,
      "grad_norm": 2.212815761566162,
      "learning_rate": 2.995100548794458e-05,
      "loss": 1.959,
      "step": 13215
    },
    {
      "epoch": 0.025763459119251906,
      "grad_norm": 2.1396286487579346,
      "learning_rate": 2.9950894260809015e-05,
      "loss": 1.9048,
      "step": 13230
    },
    {
      "epoch": 0.025792669390362167,
      "grad_norm": 2.823305606842041,
      "learning_rate": 2.995078290776987e-05,
      "loss": 1.9381,
      "step": 13245
    },
    {
      "epoch": 0.02582187966147243,
      "grad_norm": 3.0127432346343994,
      "learning_rate": 2.9950671428828083e-05,
      "loss": 1.959,
      "step": 13260
    },
    {
      "epoch": 0.025851089932582696,
      "grad_norm": 2.1536638736724854,
      "learning_rate": 2.9950559823984583e-05,
      "loss": 1.7713,
      "step": 13275
    },
    {
      "epoch": 0.025880300203692957,
      "grad_norm": 2.784940242767334,
      "learning_rate": 2.9950448093240318e-05,
      "loss": 1.8586,
      "step": 13290
    },
    {
      "epoch": 0.02590951047480322,
      "grad_norm": 2.103855848312378,
      "learning_rate": 2.9950336236596226e-05,
      "loss": 1.8167,
      "step": 13305
    },
    {
      "epoch": 0.025938720745913482,
      "grad_norm": 2.465585947036743,
      "learning_rate": 2.9950224254053254e-05,
      "loss": 1.9863,
      "step": 13320
    },
    {
      "epoch": 0.025967931017023747,
      "grad_norm": 3.2060279846191406,
      "learning_rate": 2.9950112145612335e-05,
      "loss": 1.838,
      "step": 13335
    },
    {
      "epoch": 0.025997141288134008,
      "grad_norm": 3.0637335777282715,
      "learning_rate": 2.9949999911274427e-05,
      "loss": 1.9325,
      "step": 13350
    },
    {
      "epoch": 0.026026351559244272,
      "grad_norm": 2.8039963245391846,
      "learning_rate": 2.994988755104046e-05,
      "loss": 1.8671,
      "step": 13365
    },
    {
      "epoch": 0.026055561830354537,
      "grad_norm": 4.239367485046387,
      "learning_rate": 2.9949775064911388e-05,
      "loss": 1.9305,
      "step": 13380
    },
    {
      "epoch": 0.026084772101464797,
      "grad_norm": 4.794821262359619,
      "learning_rate": 2.994966245288816e-05,
      "loss": 1.8946,
      "step": 13395
    },
    {
      "epoch": 0.026113982372575062,
      "grad_norm": 3.5099828243255615,
      "learning_rate": 2.994954971497172e-05,
      "loss": 1.919,
      "step": 13410
    },
    {
      "epoch": 0.026143192643685323,
      "grad_norm": 4.3362603187561035,
      "learning_rate": 2.994943685116302e-05,
      "loss": 1.9266,
      "step": 13425
    },
    {
      "epoch": 0.026172402914795587,
      "grad_norm": 4.010773658752441,
      "learning_rate": 2.994932386146301e-05,
      "loss": 1.8712,
      "step": 13440
    },
    {
      "epoch": 0.02620161318590585,
      "grad_norm": 2.9033546447753906,
      "learning_rate": 2.9949210745872638e-05,
      "loss": 1.8179,
      "step": 13455
    },
    {
      "epoch": 0.026230823457016113,
      "grad_norm": 2.215955972671509,
      "learning_rate": 2.9949097504392866e-05,
      "loss": 1.8573,
      "step": 13470
    },
    {
      "epoch": 0.026260033728126374,
      "grad_norm": 4.64263391494751,
      "learning_rate": 2.994898413702464e-05,
      "loss": 1.8504,
      "step": 13485
    },
    {
      "epoch": 0.026289243999236638,
      "grad_norm": 2.8851144313812256,
      "learning_rate": 2.9948870643768915e-05,
      "loss": 1.7891,
      "step": 13500
    },
    {
      "epoch": 0.026318454270346903,
      "grad_norm": 5.705179214477539,
      "learning_rate": 2.9948757024626645e-05,
      "loss": 1.8502,
      "step": 13515
    },
    {
      "epoch": 0.026347664541457164,
      "grad_norm": 4.427610397338867,
      "learning_rate": 2.994864327959879e-05,
      "loss": 1.746,
      "step": 13530
    },
    {
      "epoch": 0.026376874812567428,
      "grad_norm": 2.9682793617248535,
      "learning_rate": 2.994852940868631e-05,
      "loss": 1.8766,
      "step": 13545
    },
    {
      "epoch": 0.02640608508367769,
      "grad_norm": 2.8406543731689453,
      "learning_rate": 2.9948415411890164e-05,
      "loss": 1.8637,
      "step": 13560
    },
    {
      "epoch": 0.026435295354787954,
      "grad_norm": 2.9661149978637695,
      "learning_rate": 2.9948301289211308e-05,
      "loss": 1.7703,
      "step": 13575
    },
    {
      "epoch": 0.026464505625898214,
      "grad_norm": 2.961155652999878,
      "learning_rate": 2.99481870406507e-05,
      "loss": 1.8123,
      "step": 13590
    },
    {
      "epoch": 0.02649371589700848,
      "grad_norm": 3.7241668701171875,
      "learning_rate": 2.9948072666209308e-05,
      "loss": 2.018,
      "step": 13605
    },
    {
      "epoch": 0.026522926168118743,
      "grad_norm": 2.8102498054504395,
      "learning_rate": 2.9947958165888096e-05,
      "loss": 1.8577,
      "step": 13620
    },
    {
      "epoch": 0.026552136439229004,
      "grad_norm": 2.061007022857666,
      "learning_rate": 2.9947843539688027e-05,
      "loss": 1.9684,
      "step": 13635
    },
    {
      "epoch": 0.02658134671033927,
      "grad_norm": 4.699859619140625,
      "learning_rate": 2.994772878761006e-05,
      "loss": 1.9438,
      "step": 13650
    },
    {
      "epoch": 0.02661055698144953,
      "grad_norm": 5.8805952072143555,
      "learning_rate": 2.994761390965517e-05,
      "loss": 1.8862,
      "step": 13665
    },
    {
      "epoch": 0.026639767252559794,
      "grad_norm": 3.6178531646728516,
      "learning_rate": 2.994749890582432e-05,
      "loss": 1.9754,
      "step": 13680
    },
    {
      "epoch": 0.026668977523670055,
      "grad_norm": 2.891448497772217,
      "learning_rate": 2.9947383776118482e-05,
      "loss": 1.8838,
      "step": 13695
    },
    {
      "epoch": 0.02669818779478032,
      "grad_norm": 2.5380797386169434,
      "learning_rate": 2.994726852053862e-05,
      "loss": 2.0006,
      "step": 13710
    },
    {
      "epoch": 0.02672739806589058,
      "grad_norm": 3.083801031112671,
      "learning_rate": 2.994715313908571e-05,
      "loss": 1.9287,
      "step": 13725
    },
    {
      "epoch": 0.026756608337000845,
      "grad_norm": 3.9220306873321533,
      "learning_rate": 2.9947037631760717e-05,
      "loss": 2.0063,
      "step": 13740
    },
    {
      "epoch": 0.02678581860811111,
      "grad_norm": 2.41329288482666,
      "learning_rate": 2.994692199856462e-05,
      "loss": 1.7779,
      "step": 13755
    },
    {
      "epoch": 0.02681502887922137,
      "grad_norm": 3.137281656265259,
      "learning_rate": 2.9946806239498392e-05,
      "loss": 1.7686,
      "step": 13770
    },
    {
      "epoch": 0.026844239150331635,
      "grad_norm": 3.8897507190704346,
      "learning_rate": 2.994669035456301e-05,
      "loss": 1.9879,
      "step": 13785
    },
    {
      "epoch": 0.026873449421441896,
      "grad_norm": 2.888145685195923,
      "learning_rate": 2.994657434375944e-05,
      "loss": 2.0012,
      "step": 13800
    },
    {
      "epoch": 0.02690265969255216,
      "grad_norm": 2.683145523071289,
      "learning_rate": 2.9946458207088667e-05,
      "loss": 1.8579,
      "step": 13815
    },
    {
      "epoch": 0.02693186996366242,
      "grad_norm": 2.5023186206817627,
      "learning_rate": 2.9946341944551668e-05,
      "loss": 1.8899,
      "step": 13830
    },
    {
      "epoch": 0.026961080234772686,
      "grad_norm": 4.522122383117676,
      "learning_rate": 2.994622555614942e-05,
      "loss": 1.8373,
      "step": 13845
    },
    {
      "epoch": 0.02699029050588295,
      "grad_norm": 1.9197810888290405,
      "learning_rate": 2.9946109041882902e-05,
      "loss": 1.874,
      "step": 13860
    },
    {
      "epoch": 0.02701950077699321,
      "grad_norm": 2.0907135009765625,
      "learning_rate": 2.9945992401753103e-05,
      "loss": 1.9878,
      "step": 13875
    },
    {
      "epoch": 0.027048711048103476,
      "grad_norm": 3.0691592693328857,
      "learning_rate": 2.9945875635761e-05,
      "loss": 1.8859,
      "step": 13890
    },
    {
      "epoch": 0.027077921319213737,
      "grad_norm": 2.0707552433013916,
      "learning_rate": 2.9945758743907573e-05,
      "loss": 1.7612,
      "step": 13905
    },
    {
      "epoch": 0.027107131590324,
      "grad_norm": 2.2770462036132812,
      "learning_rate": 2.994564172619381e-05,
      "loss": 1.8028,
      "step": 13920
    },
    {
      "epoch": 0.027136341861434262,
      "grad_norm": 2.681814193725586,
      "learning_rate": 2.9945524582620695e-05,
      "loss": 1.7967,
      "step": 13935
    },
    {
      "epoch": 0.027165552132544526,
      "grad_norm": 3.0529186725616455,
      "learning_rate": 2.994540731318922e-05,
      "loss": 1.7972,
      "step": 13950
    },
    {
      "epoch": 0.027194762403654787,
      "grad_norm": 3.369091033935547,
      "learning_rate": 2.9945289917900368e-05,
      "loss": 1.8092,
      "step": 13965
    },
    {
      "epoch": 0.027223972674765052,
      "grad_norm": 2.190134048461914,
      "learning_rate": 2.9945172396755124e-05,
      "loss": 2.0228,
      "step": 13980
    },
    {
      "epoch": 0.027253182945875316,
      "grad_norm": 2.805100202560425,
      "learning_rate": 2.9945054749754483e-05,
      "loss": 1.9312,
      "step": 13995
    },
    {
      "epoch": 0.027282393216985577,
      "grad_norm": 2.195697546005249,
      "learning_rate": 2.9944936976899433e-05,
      "loss": 1.9791,
      "step": 14010
    },
    {
      "epoch": 0.027311603488095842,
      "grad_norm": 1.723713755607605,
      "learning_rate": 2.9944819078190967e-05,
      "loss": 1.8542,
      "step": 14025
    },
    {
      "epoch": 0.027340813759206103,
      "grad_norm": 2.633101463317871,
      "learning_rate": 2.9944701053630075e-05,
      "loss": 1.8127,
      "step": 14040
    },
    {
      "epoch": 0.027370024030316367,
      "grad_norm": 1.9390171766281128,
      "learning_rate": 2.9944582903217756e-05,
      "loss": 1.9183,
      "step": 14055
    },
    {
      "epoch": 0.027399234301426628,
      "grad_norm": 3.9491968154907227,
      "learning_rate": 2.9944464626955003e-05,
      "loss": 2.0849,
      "step": 14070
    },
    {
      "epoch": 0.027428444572536893,
      "grad_norm": 2.4679179191589355,
      "learning_rate": 2.9944346224842812e-05,
      "loss": 1.9285,
      "step": 14085
    },
    {
      "epoch": 0.027457654843647157,
      "grad_norm": 2.999509334564209,
      "learning_rate": 2.994422769688218e-05,
      "loss": 1.9523,
      "step": 14100
    },
    {
      "epoch": 0.027486865114757418,
      "grad_norm": 3.8798091411590576,
      "learning_rate": 2.9944109043074104e-05,
      "loss": 1.9014,
      "step": 14115
    },
    {
      "epoch": 0.027516075385867683,
      "grad_norm": 2.5288240909576416,
      "learning_rate": 2.9943990263419582e-05,
      "loss": 2.1135,
      "step": 14130
    },
    {
      "epoch": 0.027545285656977943,
      "grad_norm": 2.2120304107666016,
      "learning_rate": 2.994387135791962e-05,
      "loss": 1.7418,
      "step": 14145
    },
    {
      "epoch": 0.027574495928088208,
      "grad_norm": 2.805328607559204,
      "learning_rate": 2.994375232657521e-05,
      "loss": 1.8776,
      "step": 14160
    },
    {
      "epoch": 0.02760370619919847,
      "grad_norm": 2.8841097354888916,
      "learning_rate": 2.9943633169387365e-05,
      "loss": 1.9106,
      "step": 14175
    },
    {
      "epoch": 0.027632916470308733,
      "grad_norm": 1.8887025117874146,
      "learning_rate": 2.994351388635708e-05,
      "loss": 1.8916,
      "step": 14190
    },
    {
      "epoch": 0.027662126741418994,
      "grad_norm": 2.8623757362365723,
      "learning_rate": 2.9943394477485363e-05,
      "loss": 1.8735,
      "step": 14205
    },
    {
      "epoch": 0.02769133701252926,
      "grad_norm": 3.1046249866485596,
      "learning_rate": 2.994327494277322e-05,
      "loss": 1.9124,
      "step": 14220
    },
    {
      "epoch": 0.027720547283639523,
      "grad_norm": 2.653933525085449,
      "learning_rate": 2.9943155282221663e-05,
      "loss": 1.9387,
      "step": 14235
    },
    {
      "epoch": 0.027749757554749784,
      "grad_norm": 2.975820779800415,
      "learning_rate": 2.9943035495831688e-05,
      "loss": 1.8232,
      "step": 14250
    },
    {
      "epoch": 0.02777896782586005,
      "grad_norm": 5.906015396118164,
      "learning_rate": 2.9942915583604307e-05,
      "loss": 1.9167,
      "step": 14265
    },
    {
      "epoch": 0.02780817809697031,
      "grad_norm": 2.592456102371216,
      "learning_rate": 2.994279554554054e-05,
      "loss": 1.7433,
      "step": 14280
    },
    {
      "epoch": 0.027837388368080574,
      "grad_norm": 5.042680263519287,
      "learning_rate": 2.994267538164138e-05,
      "loss": 1.7878,
      "step": 14295
    },
    {
      "epoch": 0.027866598639190835,
      "grad_norm": 4.092184066772461,
      "learning_rate": 2.9942555091907853e-05,
      "loss": 1.6955,
      "step": 14310
    },
    {
      "epoch": 0.0278958089103011,
      "grad_norm": 4.623755931854248,
      "learning_rate": 2.994243467634097e-05,
      "loss": 1.7866,
      "step": 14325
    },
    {
      "epoch": 0.027925019181411364,
      "grad_norm": 3.042306661605835,
      "learning_rate": 2.994231413494174e-05,
      "loss": 1.8891,
      "step": 14340
    },
    {
      "epoch": 0.027954229452521625,
      "grad_norm": 2.784275531768799,
      "learning_rate": 2.9942193467711184e-05,
      "loss": 2.0112,
      "step": 14355
    },
    {
      "epoch": 0.02798343972363189,
      "grad_norm": 1.9308695793151855,
      "learning_rate": 2.9942072674650317e-05,
      "loss": 1.7964,
      "step": 14370
    },
    {
      "epoch": 0.02801264999474215,
      "grad_norm": 3.7377004623413086,
      "learning_rate": 2.994195175576015e-05,
      "loss": 1.8661,
      "step": 14385
    },
    {
      "epoch": 0.028041860265852415,
      "grad_norm": 2.484870195388794,
      "learning_rate": 2.994183071104171e-05,
      "loss": 1.7358,
      "step": 14400
    },
    {
      "epoch": 0.028071070536962676,
      "grad_norm": 2.6344974040985107,
      "learning_rate": 2.9941709540496013e-05,
      "loss": 1.9183,
      "step": 14415
    },
    {
      "epoch": 0.02810028080807294,
      "grad_norm": 2.168701410293579,
      "learning_rate": 2.9941588244124072e-05,
      "loss": 1.999,
      "step": 14430
    },
    {
      "epoch": 0.028129491079183205,
      "grad_norm": 2.986727476119995,
      "learning_rate": 2.994146682192692e-05,
      "loss": 1.8344,
      "step": 14445
    },
    {
      "epoch": 0.028158701350293466,
      "grad_norm": 3.3715713024139404,
      "learning_rate": 2.9941345273905573e-05,
      "loss": 2.0468,
      "step": 14460
    },
    {
      "epoch": 0.02818791162140373,
      "grad_norm": 2.2077038288116455,
      "learning_rate": 2.9941223600061054e-05,
      "loss": 2.0255,
      "step": 14475
    },
    {
      "epoch": 0.02821712189251399,
      "grad_norm": 3.4651224613189697,
      "learning_rate": 2.994110180039439e-05,
      "loss": 1.8604,
      "step": 14490
    },
    {
      "epoch": 0.028246332163624256,
      "grad_norm": 2.0584287643432617,
      "learning_rate": 2.994097987490661e-05,
      "loss": 2.0636,
      "step": 14505
    },
    {
      "epoch": 0.028275542434734516,
      "grad_norm": 3.1285014152526855,
      "learning_rate": 2.9940857823598736e-05,
      "loss": 1.7656,
      "step": 14520
    },
    {
      "epoch": 0.02830475270584478,
      "grad_norm": 2.541280746459961,
      "learning_rate": 2.9940735646471793e-05,
      "loss": 1.8682,
      "step": 14535
    },
    {
      "epoch": 0.028333962976955042,
      "grad_norm": 2.4748847484588623,
      "learning_rate": 2.9940613343526817e-05,
      "loss": 2.0047,
      "step": 14550
    },
    {
      "epoch": 0.028363173248065306,
      "grad_norm": 3.2105560302734375,
      "learning_rate": 2.9940490914764834e-05,
      "loss": 1.7709,
      "step": 14565
    },
    {
      "epoch": 0.02839238351917557,
      "grad_norm": 3.491591215133667,
      "learning_rate": 2.9940368360186878e-05,
      "loss": 1.786,
      "step": 14580
    },
    {
      "epoch": 0.028421593790285832,
      "grad_norm": 3.315342903137207,
      "learning_rate": 2.9940245679793978e-05,
      "loss": 1.917,
      "step": 14595
    },
    {
      "epoch": 0.028450804061396096,
      "grad_norm": 1.7594997882843018,
      "learning_rate": 2.9940122873587164e-05,
      "loss": 1.877,
      "step": 14610
    },
    {
      "epoch": 0.028480014332506357,
      "grad_norm": 2.442725896835327,
      "learning_rate": 2.9939999941567474e-05,
      "loss": 1.9577,
      "step": 14625
    },
    {
      "epoch": 0.02850922460361662,
      "grad_norm": 3.146977663040161,
      "learning_rate": 2.993987688373595e-05,
      "loss": 1.8722,
      "step": 14640
    },
    {
      "epoch": 0.028538434874726883,
      "grad_norm": 2.6678929328918457,
      "learning_rate": 2.9939753700093618e-05,
      "loss": 1.659,
      "step": 14655
    },
    {
      "epoch": 0.028567645145837147,
      "grad_norm": 2.4769906997680664,
      "learning_rate": 2.9939630390641518e-05,
      "loss": 1.8257,
      "step": 14670
    },
    {
      "epoch": 0.02859685541694741,
      "grad_norm": 2.9314770698547363,
      "learning_rate": 2.993950695538069e-05,
      "loss": 1.8004,
      "step": 14685
    },
    {
      "epoch": 0.028626065688057672,
      "grad_norm": 3.2279980182647705,
      "learning_rate": 2.993938339431217e-05,
      "loss": 1.9438,
      "step": 14700
    },
    {
      "epoch": 0.028655275959167937,
      "grad_norm": 2.8929495811462402,
      "learning_rate": 2.9939259707437002e-05,
      "loss": 1.7995,
      "step": 14715
    },
    {
      "epoch": 0.028684486230278198,
      "grad_norm": 4.861998558044434,
      "learning_rate": 2.9939135894756232e-05,
      "loss": 1.8188,
      "step": 14730
    },
    {
      "epoch": 0.028713696501388462,
      "grad_norm": 5.37394905090332,
      "learning_rate": 2.9939011956270893e-05,
      "loss": 2.061,
      "step": 14745
    },
    {
      "epoch": 0.028742906772498723,
      "grad_norm": 2.2253520488739014,
      "learning_rate": 2.9938887891982035e-05,
      "loss": 1.963,
      "step": 14760
    },
    {
      "epoch": 0.028772117043608988,
      "grad_norm": 3.424954414367676,
      "learning_rate": 2.99387637018907e-05,
      "loss": 2.0077,
      "step": 14775
    },
    {
      "epoch": 0.02880132731471925,
      "grad_norm": 2.8398706912994385,
      "learning_rate": 2.9938639385997934e-05,
      "loss": 2.0516,
      "step": 14790
    },
    {
      "epoch": 0.028830537585829513,
      "grad_norm": 2.371492385864258,
      "learning_rate": 2.9938514944304788e-05,
      "loss": 1.8057,
      "step": 14805
    },
    {
      "epoch": 0.028859747856939778,
      "grad_norm": 1.870301365852356,
      "learning_rate": 2.9938390376812304e-05,
      "loss": 1.8335,
      "step": 14820
    },
    {
      "epoch": 0.02888895812805004,
      "grad_norm": 3.1508800983428955,
      "learning_rate": 2.9938265683521533e-05,
      "loss": 2.0272,
      "step": 14835
    },
    {
      "epoch": 0.028918168399160303,
      "grad_norm": 2.8456640243530273,
      "learning_rate": 2.9938140864433528e-05,
      "loss": 1.889,
      "step": 14850
    },
    {
      "epoch": 0.028947378670270564,
      "grad_norm": 2.3040804862976074,
      "learning_rate": 2.9938015919549337e-05,
      "loss": 1.9274,
      "step": 14865
    },
    {
      "epoch": 0.02897658894138083,
      "grad_norm": 3.075559139251709,
      "learning_rate": 2.9937890848870012e-05,
      "loss": 1.9239,
      "step": 14880
    },
    {
      "epoch": 0.02900579921249109,
      "grad_norm": 2.4840190410614014,
      "learning_rate": 2.9937765652396608e-05,
      "loss": 1.9836,
      "step": 14895
    },
    {
      "epoch": 0.029035009483601354,
      "grad_norm": 2.915515422821045,
      "learning_rate": 2.9937640330130182e-05,
      "loss": 2.0196,
      "step": 14910
    },
    {
      "epoch": 0.02906421975471162,
      "grad_norm": 2.684401035308838,
      "learning_rate": 2.993751488207178e-05,
      "loss": 1.9699,
      "step": 14925
    },
    {
      "epoch": 0.02909343002582188,
      "grad_norm": 1.7906841039657593,
      "learning_rate": 2.9937389308222468e-05,
      "loss": 1.9435,
      "step": 14940
    },
    {
      "epoch": 0.029122640296932144,
      "grad_norm": 2.7629384994506836,
      "learning_rate": 2.9937263608583297e-05,
      "loss": 1.9266,
      "step": 14955
    },
    {
      "epoch": 0.029151850568042405,
      "grad_norm": 3.65447735786438,
      "learning_rate": 2.9937137783155326e-05,
      "loss": 1.8818,
      "step": 14970
    },
    {
      "epoch": 0.02918106083915267,
      "grad_norm": 2.684885025024414,
      "learning_rate": 2.993701183193962e-05,
      "loss": 1.744,
      "step": 14985
    },
    {
      "epoch": 0.02921027111026293,
      "grad_norm": 2.995678424835205,
      "learning_rate": 2.9936885754937237e-05,
      "loss": 1.868,
      "step": 15000
    },
    {
      "epoch": 0.029239481381373195,
      "grad_norm": 1.9314979314804077,
      "learning_rate": 2.993675955214924e-05,
      "loss": 1.8982,
      "step": 15015
    },
    {
      "epoch": 0.029268691652483456,
      "grad_norm": 3.8655660152435303,
      "learning_rate": 2.993663322357669e-05,
      "loss": 2.103,
      "step": 15030
    },
    {
      "epoch": 0.02929790192359372,
      "grad_norm": 3.2133545875549316,
      "learning_rate": 2.993650676922065e-05,
      "loss": 1.867,
      "step": 15045
    },
    {
      "epoch": 0.029327112194703985,
      "grad_norm": 1.7590382099151611,
      "learning_rate": 2.9936380189082184e-05,
      "loss": 1.9596,
      "step": 15060
    },
    {
      "epoch": 0.029356322465814245,
      "grad_norm": 3.5091662406921387,
      "learning_rate": 2.993625348316236e-05,
      "loss": 2.0665,
      "step": 15075
    },
    {
      "epoch": 0.02938553273692451,
      "grad_norm": 2.493689775466919,
      "learning_rate": 2.9936126651462246e-05,
      "loss": 1.8168,
      "step": 15090
    },
    {
      "epoch": 0.02941474300803477,
      "grad_norm": 2.158205509185791,
      "learning_rate": 2.993599969398291e-05,
      "loss": 1.9542,
      "step": 15105
    },
    {
      "epoch": 0.029443953279145035,
      "grad_norm": 1.6943566799163818,
      "learning_rate": 2.9935872610725415e-05,
      "loss": 1.9762,
      "step": 15120
    },
    {
      "epoch": 0.029473163550255296,
      "grad_norm": 4.148080825805664,
      "learning_rate": 2.993574540169084e-05,
      "loss": 1.8338,
      "step": 15135
    },
    {
      "epoch": 0.02950237382136556,
      "grad_norm": 3.9066367149353027,
      "learning_rate": 2.993561806688025e-05,
      "loss": 1.875,
      "step": 15150
    },
    {
      "epoch": 0.029531584092475825,
      "grad_norm": 4.32364559173584,
      "learning_rate": 2.9935490606294726e-05,
      "loss": 1.7329,
      "step": 15165
    },
    {
      "epoch": 0.029560794363586086,
      "grad_norm": 3.791557788848877,
      "learning_rate": 2.9935363019935327e-05,
      "loss": 1.8515,
      "step": 15180
    },
    {
      "epoch": 0.02959000463469635,
      "grad_norm": 2.274827241897583,
      "learning_rate": 2.9935235307803137e-05,
      "loss": 1.7744,
      "step": 15195
    },
    {
      "epoch": 0.02961921490580661,
      "grad_norm": 3.765348196029663,
      "learning_rate": 2.9935107469899235e-05,
      "loss": 1.7427,
      "step": 15210
    },
    {
      "epoch": 0.029648425176916876,
      "grad_norm": 1.8818484544754028,
      "learning_rate": 2.9934979506224687e-05,
      "loss": 1.912,
      "step": 15225
    },
    {
      "epoch": 0.029677635448027137,
      "grad_norm": 3.2676870822906494,
      "learning_rate": 2.993485141678058e-05,
      "loss": 2.022,
      "step": 15240
    },
    {
      "epoch": 0.0297068457191374,
      "grad_norm": 3.329846143722534,
      "learning_rate": 2.9934723201567986e-05,
      "loss": 1.9317,
      "step": 15255
    },
    {
      "epoch": 0.029736055990247662,
      "grad_norm": 4.478529453277588,
      "learning_rate": 2.993459486058799e-05,
      "loss": 1.8172,
      "step": 15270
    },
    {
      "epoch": 0.029765266261357927,
      "grad_norm": 5.912947654724121,
      "learning_rate": 2.9934466393841667e-05,
      "loss": 1.8828,
      "step": 15285
    },
    {
      "epoch": 0.02979447653246819,
      "grad_norm": 3.079585552215576,
      "learning_rate": 2.9934337801330102e-05,
      "loss": 1.946,
      "step": 15300
    },
    {
      "epoch": 0.029823686803578452,
      "grad_norm": 2.770911455154419,
      "learning_rate": 2.993420908305438e-05,
      "loss": 1.884,
      "step": 15315
    },
    {
      "epoch": 0.029852897074688717,
      "grad_norm": 2.8034188747406006,
      "learning_rate": 2.993408023901558e-05,
      "loss": 1.7856,
      "step": 15330
    },
    {
      "epoch": 0.029882107345798978,
      "grad_norm": 4.039456367492676,
      "learning_rate": 2.9933951269214793e-05,
      "loss": 1.9302,
      "step": 15345
    },
    {
      "epoch": 0.029911317616909242,
      "grad_norm": 3.329083204269409,
      "learning_rate": 2.99338221736531e-05,
      "loss": 1.9541,
      "step": 15360
    },
    {
      "epoch": 0.029940527888019503,
      "grad_norm": 2.9838755130767822,
      "learning_rate": 2.9933692952331593e-05,
      "loss": 1.8633,
      "step": 15375
    },
    {
      "epoch": 0.029969738159129768,
      "grad_norm": 3.2724435329437256,
      "learning_rate": 2.9933563605251356e-05,
      "loss": 1.9572,
      "step": 15390
    },
    {
      "epoch": 0.029998948430240032,
      "grad_norm": 1.9584968090057373,
      "learning_rate": 2.993343413241348e-05,
      "loss": 1.7946,
      "step": 15405
    },
    {
      "epoch": 0.030028158701350293,
      "grad_norm": 3.7978711128234863,
      "learning_rate": 2.9933304533819053e-05,
      "loss": 1.7179,
      "step": 15420
    },
    {
      "epoch": 0.030057368972460557,
      "grad_norm": 3.708948850631714,
      "learning_rate": 2.993317480946917e-05,
      "loss": 1.7777,
      "step": 15435
    },
    {
      "epoch": 0.03008657924357082,
      "grad_norm": 2.204556941986084,
      "learning_rate": 2.993304495936492e-05,
      "loss": 2.0804,
      "step": 15450
    },
    {
      "epoch": 0.030115789514681083,
      "grad_norm": 4.719995021820068,
      "learning_rate": 2.9932914983507398e-05,
      "loss": 1.8505,
      "step": 15465
    },
    {
      "epoch": 0.030144999785791344,
      "grad_norm": 2.958409547805786,
      "learning_rate": 2.9932784881897703e-05,
      "loss": 1.8833,
      "step": 15480
    },
    {
      "epoch": 0.03017421005690161,
      "grad_norm": 2.398984909057617,
      "learning_rate": 2.993265465453692e-05,
      "loss": 1.6463,
      "step": 15495
    },
    {
      "epoch": 0.03020342032801187,
      "grad_norm": 2.104973077774048,
      "learning_rate": 2.9932524301426155e-05,
      "loss": 1.7249,
      "step": 15510
    },
    {
      "epoch": 0.030232630599122134,
      "grad_norm": 5.241917610168457,
      "learning_rate": 2.99323938225665e-05,
      "loss": 1.7852,
      "step": 15525
    },
    {
      "epoch": 0.030261840870232398,
      "grad_norm": 2.9521865844726562,
      "learning_rate": 2.9932263217959064e-05,
      "loss": 1.8941,
      "step": 15540
    },
    {
      "epoch": 0.03029105114134266,
      "grad_norm": 3.14367413520813,
      "learning_rate": 2.9932132487604936e-05,
      "loss": 1.9962,
      "step": 15555
    },
    {
      "epoch": 0.030320261412452924,
      "grad_norm": 1.763987421989441,
      "learning_rate": 2.9932001631505217e-05,
      "loss": 1.9057,
      "step": 15570
    },
    {
      "epoch": 0.030349471683563185,
      "grad_norm": 4.6951446533203125,
      "learning_rate": 2.993187064966101e-05,
      "loss": 1.8781,
      "step": 15585
    },
    {
      "epoch": 0.03037868195467345,
      "grad_norm": 2.424650192260742,
      "learning_rate": 2.993173954207343e-05,
      "loss": 1.8808,
      "step": 15600
    },
    {
      "epoch": 0.03040789222578371,
      "grad_norm": 2.743579387664795,
      "learning_rate": 2.9931608308743562e-05,
      "loss": 1.899,
      "step": 15615
    },
    {
      "epoch": 0.030437102496893974,
      "grad_norm": 4.443767070770264,
      "learning_rate": 2.9931476949672524e-05,
      "loss": 1.8727,
      "step": 15630
    },
    {
      "epoch": 0.03046631276800424,
      "grad_norm": 2.6599271297454834,
      "learning_rate": 2.9931345464861418e-05,
      "loss": 1.8466,
      "step": 15645
    },
    {
      "epoch": 0.0304955230391145,
      "grad_norm": 2.598816394805908,
      "learning_rate": 2.993121385431135e-05,
      "loss": 2.0863,
      "step": 15660
    },
    {
      "epoch": 0.030524733310224764,
      "grad_norm": 3.126720666885376,
      "learning_rate": 2.9931082118023432e-05,
      "loss": 1.7731,
      "step": 15675
    },
    {
      "epoch": 0.030553943581335025,
      "grad_norm": 3.7353076934814453,
      "learning_rate": 2.9930950255998773e-05,
      "loss": 2.1104,
      "step": 15690
    },
    {
      "epoch": 0.03058315385244529,
      "grad_norm": 4.932044982910156,
      "learning_rate": 2.9930818268238483e-05,
      "loss": 1.8693,
      "step": 15705
    },
    {
      "epoch": 0.03061236412355555,
      "grad_norm": 2.6772072315216064,
      "learning_rate": 2.9930686154743666e-05,
      "loss": 1.8159,
      "step": 15720
    },
    {
      "epoch": 0.030641574394665815,
      "grad_norm": 3.73004412651062,
      "learning_rate": 2.9930553915515445e-05,
      "loss": 1.6705,
      "step": 15735
    },
    {
      "epoch": 0.03067078466577608,
      "grad_norm": 2.2986795902252197,
      "learning_rate": 2.993042155055493e-05,
      "loss": 2.05,
      "step": 15750
    },
    {
      "epoch": 0.03069999493688634,
      "grad_norm": 2.9536030292510986,
      "learning_rate": 2.9930289059863234e-05,
      "loss": 1.9297,
      "step": 15765
    },
    {
      "epoch": 0.030729205207996605,
      "grad_norm": 1.8823219537734985,
      "learning_rate": 2.9930156443441477e-05,
      "loss": 1.9812,
      "step": 15780
    },
    {
      "epoch": 0.030758415479106866,
      "grad_norm": 2.553921699523926,
      "learning_rate": 2.993002370129077e-05,
      "loss": 2.0723,
      "step": 15795
    },
    {
      "epoch": 0.03078762575021713,
      "grad_norm": 2.391080617904663,
      "learning_rate": 2.9929890833412233e-05,
      "loss": 1.8342,
      "step": 15810
    },
    {
      "epoch": 0.03081683602132739,
      "grad_norm": 1.898431420326233,
      "learning_rate": 2.9929757839806985e-05,
      "loss": 1.8282,
      "step": 15825
    },
    {
      "epoch": 0.030846046292437656,
      "grad_norm": 3.6443545818328857,
      "learning_rate": 2.9929624720476153e-05,
      "loss": 1.7823,
      "step": 15840
    },
    {
      "epoch": 0.030875256563547917,
      "grad_norm": 2.1495354175567627,
      "learning_rate": 2.9929491475420844e-05,
      "loss": 1.9565,
      "step": 15855
    },
    {
      "epoch": 0.03090446683465818,
      "grad_norm": 4.317218780517578,
      "learning_rate": 2.992935810464219e-05,
      "loss": 1.9516,
      "step": 15870
    },
    {
      "epoch": 0.030933677105768446,
      "grad_norm": 2.2863664627075195,
      "learning_rate": 2.992922460814131e-05,
      "loss": 1.9399,
      "step": 15885
    },
    {
      "epoch": 0.030962887376878707,
      "grad_norm": 3.2142059803009033,
      "learning_rate": 2.9929090985919334e-05,
      "loss": 2.0122,
      "step": 15900
    },
    {
      "epoch": 0.03099209764798897,
      "grad_norm": 3.0333306789398193,
      "learning_rate": 2.992895723797738e-05,
      "loss": 1.7733,
      "step": 15915
    },
    {
      "epoch": 0.031021307919099232,
      "grad_norm": 2.6775388717651367,
      "learning_rate": 2.9928823364316575e-05,
      "loss": 1.8205,
      "step": 15930
    },
    {
      "epoch": 0.031050518190209497,
      "grad_norm": 2.1693944931030273,
      "learning_rate": 2.9928689364938057e-05,
      "loss": 1.895,
      "step": 15945
    },
    {
      "epoch": 0.031079728461319758,
      "grad_norm": 2.754948377609253,
      "learning_rate": 2.992855523984294e-05,
      "loss": 1.8623,
      "step": 15960
    },
    {
      "epoch": 0.031108938732430022,
      "grad_norm": 2.659349203109741,
      "learning_rate": 2.9928420989032357e-05,
      "loss": 1.9873,
      "step": 15975
    },
    {
      "epoch": 0.031138149003540287,
      "grad_norm": 3.5560247898101807,
      "learning_rate": 2.9928286612507445e-05,
      "loss": 1.9364,
      "step": 15990
    },
    {
      "epoch": 0.031167359274650547,
      "grad_norm": 4.824398994445801,
      "learning_rate": 2.9928152110269335e-05,
      "loss": 1.8708,
      "step": 16005
    },
    {
      "epoch": 0.031196569545760812,
      "grad_norm": 2.5657832622528076,
      "learning_rate": 2.992801748231915e-05,
      "loss": 2.0086,
      "step": 16020
    },
    {
      "epoch": 0.031225779816871073,
      "grad_norm": 5.117823123931885,
      "learning_rate": 2.9927882728658036e-05,
      "loss": 1.7608,
      "step": 16035
    },
    {
      "epoch": 0.031254990087981334,
      "grad_norm": 1.9517539739608765,
      "learning_rate": 2.992774784928712e-05,
      "loss": 1.9367,
      "step": 16050
    },
    {
      "epoch": 0.0312842003590916,
      "grad_norm": 3.202143669128418,
      "learning_rate": 2.9927612844207537e-05,
      "loss": 1.8725,
      "step": 16065
    },
    {
      "epoch": 0.03131341063020186,
      "grad_norm": 2.5217974185943604,
      "learning_rate": 2.992747771342043e-05,
      "loss": 1.9029,
      "step": 16080
    },
    {
      "epoch": 0.031342620901312124,
      "grad_norm": 1.6550703048706055,
      "learning_rate": 2.992734245692693e-05,
      "loss": 1.9016,
      "step": 16095
    },
    {
      "epoch": 0.03137183117242239,
      "grad_norm": 2.7231898307800293,
      "learning_rate": 2.9927207074728187e-05,
      "loss": 1.9735,
      "step": 16110
    },
    {
      "epoch": 0.03140104144353265,
      "grad_norm": 3.855612277984619,
      "learning_rate": 2.9927071566825328e-05,
      "loss": 1.7641,
      "step": 16125
    },
    {
      "epoch": 0.031430251714642914,
      "grad_norm": 3.016885757446289,
      "learning_rate": 2.99269359332195e-05,
      "loss": 1.9789,
      "step": 16140
    },
    {
      "epoch": 0.031459461985753175,
      "grad_norm": 2.0709967613220215,
      "learning_rate": 2.9926800173911845e-05,
      "loss": 1.922,
      "step": 16155
    },
    {
      "epoch": 0.03148867225686344,
      "grad_norm": 5.23029088973999,
      "learning_rate": 2.992666428890351e-05,
      "loss": 1.9681,
      "step": 16170
    },
    {
      "epoch": 0.031517882527973703,
      "grad_norm": 2.7961294651031494,
      "learning_rate": 2.9926528278195634e-05,
      "loss": 1.7172,
      "step": 16185
    },
    {
      "epoch": 0.031547092799083964,
      "grad_norm": 2.5741660594940186,
      "learning_rate": 2.992639214178936e-05,
      "loss": 1.7638,
      "step": 16200
    },
    {
      "epoch": 0.031576303070194225,
      "grad_norm": 1.9811028242111206,
      "learning_rate": 2.9926255879685846e-05,
      "loss": 2.0028,
      "step": 16215
    },
    {
      "epoch": 0.03160551334130449,
      "grad_norm": 4.054990768432617,
      "learning_rate": 2.992611949188623e-05,
      "loss": 1.9305,
      "step": 16230
    },
    {
      "epoch": 0.031634723612414754,
      "grad_norm": 2.189107894897461,
      "learning_rate": 2.992598297839166e-05,
      "loss": 1.7656,
      "step": 16245
    },
    {
      "epoch": 0.031663933883525015,
      "grad_norm": 2.2795944213867188,
      "learning_rate": 2.9925846339203285e-05,
      "loss": 1.8474,
      "step": 16260
    },
    {
      "epoch": 0.03169314415463528,
      "grad_norm": 5.294840335845947,
      "learning_rate": 2.9925709574322262e-05,
      "loss": 1.8983,
      "step": 16275
    },
    {
      "epoch": 0.031722354425745544,
      "grad_norm": 3.4388887882232666,
      "learning_rate": 2.9925572683749742e-05,
      "loss": 1.7128,
      "step": 16290
    },
    {
      "epoch": 0.031751564696855805,
      "grad_norm": 4.72010612487793,
      "learning_rate": 2.992543566748687e-05,
      "loss": 1.9095,
      "step": 16305
    },
    {
      "epoch": 0.031780774967966066,
      "grad_norm": 3.0153849124908447,
      "learning_rate": 2.9925298525534807e-05,
      "loss": 1.8908,
      "step": 16320
    },
    {
      "epoch": 0.031809985239076334,
      "grad_norm": 2.4210598468780518,
      "learning_rate": 2.992516125789471e-05,
      "loss": 2.0611,
      "step": 16335
    },
    {
      "epoch": 0.031839195510186595,
      "grad_norm": 2.9562766551971436,
      "learning_rate": 2.9925023864567727e-05,
      "loss": 1.7784,
      "step": 16350
    },
    {
      "epoch": 0.031868405781296856,
      "grad_norm": 2.944288969039917,
      "learning_rate": 2.992488634555502e-05,
      "loss": 1.8543,
      "step": 16365
    },
    {
      "epoch": 0.031897616052407124,
      "grad_norm": 3.5835225582122803,
      "learning_rate": 2.9924748700857747e-05,
      "loss": 1.9654,
      "step": 16380
    },
    {
      "epoch": 0.031926826323517385,
      "grad_norm": 3.5023677349090576,
      "learning_rate": 2.9924610930477062e-05,
      "loss": 1.7496,
      "step": 16395
    },
    {
      "epoch": 0.031956036594627646,
      "grad_norm": 2.08420467376709,
      "learning_rate": 2.9924473034414136e-05,
      "loss": 1.8234,
      "step": 16410
    },
    {
      "epoch": 0.03198524686573791,
      "grad_norm": 4.309360504150391,
      "learning_rate": 2.992433501267012e-05,
      "loss": 1.9149,
      "step": 16425
    },
    {
      "epoch": 0.032014457136848175,
      "grad_norm": 3.0357537269592285,
      "learning_rate": 2.9924196865246175e-05,
      "loss": 1.9551,
      "step": 16440
    },
    {
      "epoch": 0.032043667407958436,
      "grad_norm": 4.165322780609131,
      "learning_rate": 2.9924058592143473e-05,
      "loss": 1.7488,
      "step": 16455
    },
    {
      "epoch": 0.0320728776790687,
      "grad_norm": 3.4814274311065674,
      "learning_rate": 2.9923920193363176e-05,
      "loss": 1.9659,
      "step": 16470
    },
    {
      "epoch": 0.032102087950178965,
      "grad_norm": 2.5652506351470947,
      "learning_rate": 2.992378166890645e-05,
      "loss": 1.7986,
      "step": 16485
    },
    {
      "epoch": 0.032131298221289226,
      "grad_norm": 1.9026885032653809,
      "learning_rate": 2.9923643018774455e-05,
      "loss": 1.8294,
      "step": 16500
    },
    {
      "epoch": 0.03216050849239949,
      "grad_norm": 3.861070156097412,
      "learning_rate": 2.9923504242968365e-05,
      "loss": 1.7451,
      "step": 16515
    },
    {
      "epoch": 0.03218971876350975,
      "grad_norm": 1.9591658115386963,
      "learning_rate": 2.992336534148935e-05,
      "loss": 1.805,
      "step": 16530
    },
    {
      "epoch": 0.032218929034620016,
      "grad_norm": 4.189550399780273,
      "learning_rate": 2.992322631433857e-05,
      "loss": 1.8049,
      "step": 16545
    },
    {
      "epoch": 0.032248139305730276,
      "grad_norm": 3.1660592555999756,
      "learning_rate": 2.9923087161517205e-05,
      "loss": 1.9415,
      "step": 16560
    },
    {
      "epoch": 0.03227734957684054,
      "grad_norm": 2.801609754562378,
      "learning_rate": 2.9922947883026426e-05,
      "loss": 1.8096,
      "step": 16575
    },
    {
      "epoch": 0.032306559847950805,
      "grad_norm": 2.699336290359497,
      "learning_rate": 2.9922808478867403e-05,
      "loss": 1.723,
      "step": 16590
    },
    {
      "epoch": 0.032335770119061066,
      "grad_norm": 2.2953224182128906,
      "learning_rate": 2.992266894904131e-05,
      "loss": 2.0549,
      "step": 16605
    },
    {
      "epoch": 0.03236498039017133,
      "grad_norm": 3.8196887969970703,
      "learning_rate": 2.9922529293549327e-05,
      "loss": 1.8312,
      "step": 16620
    },
    {
      "epoch": 0.03239419066128159,
      "grad_norm": 2.974578380584717,
      "learning_rate": 2.9922389512392622e-05,
      "loss": 1.9242,
      "step": 16635
    },
    {
      "epoch": 0.032423400932391856,
      "grad_norm": 4.62038516998291,
      "learning_rate": 2.9922249605572376e-05,
      "loss": 1.9786,
      "step": 16650
    },
    {
      "epoch": 0.03245261120350212,
      "grad_norm": 1.5149112939834595,
      "learning_rate": 2.992210957308977e-05,
      "loss": 1.8984,
      "step": 16665
    },
    {
      "epoch": 0.03248182147461238,
      "grad_norm": 2.2014825344085693,
      "learning_rate": 2.992196941494598e-05,
      "loss": 1.7994,
      "step": 16680
    },
    {
      "epoch": 0.032511031745722646,
      "grad_norm": 4.065785884857178,
      "learning_rate": 2.9921829131142186e-05,
      "loss": 1.8157,
      "step": 16695
    },
    {
      "epoch": 0.03254024201683291,
      "grad_norm": 4.191883087158203,
      "learning_rate": 2.992168872167957e-05,
      "loss": 1.8684,
      "step": 16710
    },
    {
      "epoch": 0.03256945228794317,
      "grad_norm": 3.4441282749176025,
      "learning_rate": 2.9921548186559314e-05,
      "loss": 1.8299,
      "step": 16725
    },
    {
      "epoch": 0.03259866255905343,
      "grad_norm": 4.0237345695495605,
      "learning_rate": 2.9921407525782604e-05,
      "loss": 1.9262,
      "step": 16740
    },
    {
      "epoch": 0.0326278728301637,
      "grad_norm": 2.171717405319214,
      "learning_rate": 2.992126673935062e-05,
      "loss": 1.7679,
      "step": 16755
    },
    {
      "epoch": 0.03265708310127396,
      "grad_norm": 4.097175121307373,
      "learning_rate": 2.992112582726455e-05,
      "loss": 1.9499,
      "step": 16770
    },
    {
      "epoch": 0.03268629337238422,
      "grad_norm": 4.390431880950928,
      "learning_rate": 2.9920984789525583e-05,
      "loss": 1.8797,
      "step": 16785
    },
    {
      "epoch": 0.03271550364349448,
      "grad_norm": 3.068178176879883,
      "learning_rate": 2.9920843626134907e-05,
      "loss": 2.0427,
      "step": 16800
    },
    {
      "epoch": 0.03274471391460475,
      "grad_norm": 3.378275156021118,
      "learning_rate": 2.9920702337093707e-05,
      "loss": 1.8853,
      "step": 16815
    },
    {
      "epoch": 0.03277392418571501,
      "grad_norm": 2.2076075077056885,
      "learning_rate": 2.992056092240317e-05,
      "loss": 1.9406,
      "step": 16830
    },
    {
      "epoch": 0.03280313445682527,
      "grad_norm": 3.0358242988586426,
      "learning_rate": 2.992041938206449e-05,
      "loss": 2.0143,
      "step": 16845
    },
    {
      "epoch": 0.03283234472793554,
      "grad_norm": 4.094939231872559,
      "learning_rate": 2.9920277716078868e-05,
      "loss": 1.9757,
      "step": 16860
    },
    {
      "epoch": 0.0328615549990458,
      "grad_norm": 1.8015364408493042,
      "learning_rate": 2.9920135924447484e-05,
      "loss": 2.0698,
      "step": 16875
    },
    {
      "epoch": 0.03289076527015606,
      "grad_norm": 3.4187324047088623,
      "learning_rate": 2.9919994007171535e-05,
      "loss": 1.9332,
      "step": 16890
    },
    {
      "epoch": 0.03291997554126632,
      "grad_norm": 3.91485595703125,
      "learning_rate": 2.991985196425222e-05,
      "loss": 1.9433,
      "step": 16905
    },
    {
      "epoch": 0.03294918581237659,
      "grad_norm": 3.0797996520996094,
      "learning_rate": 2.9919709795690732e-05,
      "loss": 1.7478,
      "step": 16920
    },
    {
      "epoch": 0.03297839608348685,
      "grad_norm": 4.396271705627441,
      "learning_rate": 2.9919567501488273e-05,
      "loss": 1.9191,
      "step": 16935
    },
    {
      "epoch": 0.03300760635459711,
      "grad_norm": 2.7897489070892334,
      "learning_rate": 2.9919425081646036e-05,
      "loss": 1.9495,
      "step": 16950
    },
    {
      "epoch": 0.03303681662570738,
      "grad_norm": 3.063068151473999,
      "learning_rate": 2.991928253616522e-05,
      "loss": 2.1288,
      "step": 16965
    },
    {
      "epoch": 0.03306602689681764,
      "grad_norm": 5.236184120178223,
      "learning_rate": 2.991913986504703e-05,
      "loss": 1.8698,
      "step": 16980
    },
    {
      "epoch": 0.0330952371679279,
      "grad_norm": 2.758821725845337,
      "learning_rate": 2.9918997068292666e-05,
      "loss": 1.8664,
      "step": 16995
    },
    {
      "epoch": 0.03312444743903816,
      "grad_norm": 3.3068835735321045,
      "learning_rate": 2.9918854145903326e-05,
      "loss": 2.0247,
      "step": 17010
    },
    {
      "epoch": 0.03315365771014843,
      "grad_norm": 5.055883884429932,
      "learning_rate": 2.991871109788022e-05,
      "loss": 1.7333,
      "step": 17025
    },
    {
      "epoch": 0.03318286798125869,
      "grad_norm": 2.9909791946411133,
      "learning_rate": 2.9918567924224545e-05,
      "loss": 1.7868,
      "step": 17040
    },
    {
      "epoch": 0.03321207825236895,
      "grad_norm": 4.3141913414001465,
      "learning_rate": 2.9918424624937514e-05,
      "loss": 1.8976,
      "step": 17055
    },
    {
      "epoch": 0.03324128852347922,
      "grad_norm": 3.3365187644958496,
      "learning_rate": 2.991828120002033e-05,
      "loss": 1.8246,
      "step": 17070
    },
    {
      "epoch": 0.03327049879458948,
      "grad_norm": 1.7029449939727783,
      "learning_rate": 2.991813764947421e-05,
      "loss": 1.7464,
      "step": 17085
    },
    {
      "epoch": 0.03329970906569974,
      "grad_norm": 2.34187650680542,
      "learning_rate": 2.9917993973300343e-05,
      "loss": 2.0614,
      "step": 17100
    },
    {
      "epoch": 0.03332891933681,
      "grad_norm": 2.3299100399017334,
      "learning_rate": 2.9917850171499957e-05,
      "loss": 1.9139,
      "step": 17115
    },
    {
      "epoch": 0.03335812960792027,
      "grad_norm": 2.58627986907959,
      "learning_rate": 2.9917706244074254e-05,
      "loss": 2.0092,
      "step": 17130
    },
    {
      "epoch": 0.03338733987903053,
      "grad_norm": 3.580124855041504,
      "learning_rate": 2.991756219102445e-05,
      "loss": 1.8548,
      "step": 17145
    },
    {
      "epoch": 0.03341655015014079,
      "grad_norm": 1.6242703199386597,
      "learning_rate": 2.9917418012351755e-05,
      "loss": 1.8455,
      "step": 17160
    },
    {
      "epoch": 0.03344576042125106,
      "grad_norm": 3.0739760398864746,
      "learning_rate": 2.991727370805739e-05,
      "loss": 1.846,
      "step": 17175
    },
    {
      "epoch": 0.03347497069236132,
      "grad_norm": 2.6895573139190674,
      "learning_rate": 2.991712927814256e-05,
      "loss": 1.8201,
      "step": 17190
    },
    {
      "epoch": 0.03350418096347158,
      "grad_norm": 3.353482961654663,
      "learning_rate": 2.9916984722608488e-05,
      "loss": 1.9598,
      "step": 17205
    },
    {
      "epoch": 0.03353339123458184,
      "grad_norm": 3.334764242172241,
      "learning_rate": 2.991684004145639e-05,
      "loss": 1.8388,
      "step": 17220
    },
    {
      "epoch": 0.03356260150569211,
      "grad_norm": 2.8904106616973877,
      "learning_rate": 2.9916695234687484e-05,
      "loss": 1.9402,
      "step": 17235
    },
    {
      "epoch": 0.03359181177680237,
      "grad_norm": 2.01975154876709,
      "learning_rate": 2.991655030230299e-05,
      "loss": 1.9462,
      "step": 17250
    },
    {
      "epoch": 0.03362102204791263,
      "grad_norm": 3.490748882293701,
      "learning_rate": 2.9916405244304123e-05,
      "loss": 1.9171,
      "step": 17265
    },
    {
      "epoch": 0.033650232319022894,
      "grad_norm": 3.2388625144958496,
      "learning_rate": 2.9916260060692114e-05,
      "loss": 1.8305,
      "step": 17280
    },
    {
      "epoch": 0.03367944259013316,
      "grad_norm": 3.952791690826416,
      "learning_rate": 2.991611475146818e-05,
      "loss": 1.9577,
      "step": 17295
    },
    {
      "epoch": 0.03370865286124342,
      "grad_norm": 2.0647149085998535,
      "learning_rate": 2.9915969316633548e-05,
      "loss": 1.7789,
      "step": 17310
    },
    {
      "epoch": 0.03373786313235368,
      "grad_norm": 2.9091007709503174,
      "learning_rate": 2.9915823756189438e-05,
      "loss": 2.0568,
      "step": 17325
    },
    {
      "epoch": 0.03376707340346395,
      "grad_norm": 3.5968873500823975,
      "learning_rate": 2.9915678070137078e-05,
      "loss": 1.9264,
      "step": 17340
    },
    {
      "epoch": 0.03379628367457421,
      "grad_norm": 2.7411301136016846,
      "learning_rate": 2.9915532258477697e-05,
      "loss": 1.832,
      "step": 17355
    },
    {
      "epoch": 0.03382549394568447,
      "grad_norm": 3.8037030696868896,
      "learning_rate": 2.991538632121252e-05,
      "loss": 1.9751,
      "step": 17370
    },
    {
      "epoch": 0.033854704216794734,
      "grad_norm": 2.2467167377471924,
      "learning_rate": 2.9915240258342776e-05,
      "loss": 1.8501,
      "step": 17385
    },
    {
      "epoch": 0.033883914487905,
      "grad_norm": 2.01926589012146,
      "learning_rate": 2.9915094069869696e-05,
      "loss": 1.8398,
      "step": 17400
    },
    {
      "epoch": 0.03391312475901526,
      "grad_norm": 3.427839994430542,
      "learning_rate": 2.9914947755794515e-05,
      "loss": 2.0322,
      "step": 17415
    },
    {
      "epoch": 0.033942335030125524,
      "grad_norm": 2.9915285110473633,
      "learning_rate": 2.991480131611846e-05,
      "loss": 1.7704,
      "step": 17430
    },
    {
      "epoch": 0.03397154530123579,
      "grad_norm": 2.0821945667266846,
      "learning_rate": 2.9914654750842765e-05,
      "loss": 1.9759,
      "step": 17445
    },
    {
      "epoch": 0.03400075557234605,
      "grad_norm": 3.167320966720581,
      "learning_rate": 2.9914508059968664e-05,
      "loss": 1.8321,
      "step": 17460
    },
    {
      "epoch": 0.034029965843456314,
      "grad_norm": 3.4027440547943115,
      "learning_rate": 2.991436124349739e-05,
      "loss": 1.8643,
      "step": 17475
    },
    {
      "epoch": 0.034059176114566575,
      "grad_norm": 1.9895273447036743,
      "learning_rate": 2.9914214301430183e-05,
      "loss": 2.018,
      "step": 17490
    },
    {
      "epoch": 0.03408838638567684,
      "grad_norm": 3.4797520637512207,
      "learning_rate": 2.9914067233768285e-05,
      "loss": 1.8666,
      "step": 17505
    },
    {
      "epoch": 0.034117596656787104,
      "grad_norm": 2.578434944152832,
      "learning_rate": 2.9913920040512925e-05,
      "loss": 1.8808,
      "step": 17520
    },
    {
      "epoch": 0.034146806927897365,
      "grad_norm": 3.499577522277832,
      "learning_rate": 2.991377272166535e-05,
      "loss": 2.112,
      "step": 17535
    },
    {
      "epoch": 0.03417601719900763,
      "grad_norm": 2.6072142124176025,
      "learning_rate": 2.9913625277226795e-05,
      "loss": 1.7954,
      "step": 17550
    },
    {
      "epoch": 0.034205227470117894,
      "grad_norm": 4.382633686065674,
      "learning_rate": 2.9913477707198505e-05,
      "loss": 1.8943,
      "step": 17565
    },
    {
      "epoch": 0.034234437741228155,
      "grad_norm": 2.883723497390747,
      "learning_rate": 2.9913330011581718e-05,
      "loss": 1.8615,
      "step": 17580
    },
    {
      "epoch": 0.034263648012338416,
      "grad_norm": 2.309401273727417,
      "learning_rate": 2.991318219037769e-05,
      "loss": 1.9813,
      "step": 17595
    },
    {
      "epoch": 0.034292858283448684,
      "grad_norm": 3.3916573524475098,
      "learning_rate": 2.991303424358765e-05,
      "loss": 2.0303,
      "step": 17610
    },
    {
      "epoch": 0.034322068554558945,
      "grad_norm": 2.0413196086883545,
      "learning_rate": 2.9912886171212855e-05,
      "loss": 1.7976,
      "step": 17625
    },
    {
      "epoch": 0.034351278825669206,
      "grad_norm": 2.3423221111297607,
      "learning_rate": 2.991273797325454e-05,
      "loss": 2.0449,
      "step": 17640
    },
    {
      "epoch": 0.034380489096779474,
      "grad_norm": 2.9180712699890137,
      "learning_rate": 2.991258964971397e-05,
      "loss": 1.937,
      "step": 17655
    },
    {
      "epoch": 0.034409699367889734,
      "grad_norm": 2.8222568035125732,
      "learning_rate": 2.9912441200592385e-05,
      "loss": 1.8245,
      "step": 17670
    },
    {
      "epoch": 0.034438909638999995,
      "grad_norm": 2.49934720993042,
      "learning_rate": 2.991229262589103e-05,
      "loss": 1.7475,
      "step": 17685
    },
    {
      "epoch": 0.034468119910110256,
      "grad_norm": 2.517383098602295,
      "learning_rate": 2.9912143925611166e-05,
      "loss": 2.0156,
      "step": 17700
    },
    {
      "epoch": 0.034497330181220524,
      "grad_norm": 1.9787654876708984,
      "learning_rate": 2.9911995099754037e-05,
      "loss": 1.7771,
      "step": 17715
    },
    {
      "epoch": 0.034526540452330785,
      "grad_norm": 3.3665497303009033,
      "learning_rate": 2.9911846148320903e-05,
      "loss": 1.7394,
      "step": 17730
    },
    {
      "epoch": 0.034555750723441046,
      "grad_norm": 2.7270398139953613,
      "learning_rate": 2.9911697071313017e-05,
      "loss": 1.8666,
      "step": 17745
    },
    {
      "epoch": 0.034584960994551314,
      "grad_norm": 2.7560067176818848,
      "learning_rate": 2.9911547868731626e-05,
      "loss": 1.7811,
      "step": 17760
    },
    {
      "epoch": 0.034614171265661575,
      "grad_norm": 3.9651834964752197,
      "learning_rate": 2.9911398540577996e-05,
      "loss": 1.8758,
      "step": 17775
    },
    {
      "epoch": 0.034643381536771836,
      "grad_norm": 2.7909586429595947,
      "learning_rate": 2.9911249086853386e-05,
      "loss": 2.0029,
      "step": 17790
    },
    {
      "epoch": 0.0346725918078821,
      "grad_norm": 3.109741449356079,
      "learning_rate": 2.9911099507559045e-05,
      "loss": 1.9692,
      "step": 17805
    },
    {
      "epoch": 0.034701802078992365,
      "grad_norm": 1.7844855785369873,
      "learning_rate": 2.9910949802696244e-05,
      "loss": 1.7244,
      "step": 17820
    },
    {
      "epoch": 0.034731012350102626,
      "grad_norm": 1.8406388759613037,
      "learning_rate": 2.9910799972266232e-05,
      "loss": 1.8689,
      "step": 17835
    },
    {
      "epoch": 0.03476022262121289,
      "grad_norm": 2.9145405292510986,
      "learning_rate": 2.9910650016270278e-05,
      "loss": 1.8982,
      "step": 17850
    },
    {
      "epoch": 0.03478943289232315,
      "grad_norm": 1.9398448467254639,
      "learning_rate": 2.991049993470964e-05,
      "loss": 1.5994,
      "step": 17865
    },
    {
      "epoch": 0.034818643163433416,
      "grad_norm": 2.3878304958343506,
      "learning_rate": 2.991034972758559e-05,
      "loss": 1.8854,
      "step": 17880
    },
    {
      "epoch": 0.03484785343454368,
      "grad_norm": 1.8284587860107422,
      "learning_rate": 2.9910199394899385e-05,
      "loss": 1.8261,
      "step": 17895
    },
    {
      "epoch": 0.03487706370565394,
      "grad_norm": 3.306748151779175,
      "learning_rate": 2.9910048936652294e-05,
      "loss": 1.8727,
      "step": 17910
    },
    {
      "epoch": 0.034906273976764206,
      "grad_norm": 3.8391776084899902,
      "learning_rate": 2.9909898352845585e-05,
      "loss": 1.8347,
      "step": 17925
    },
    {
      "epoch": 0.03493548424787447,
      "grad_norm": 2.275381565093994,
      "learning_rate": 2.9909747643480526e-05,
      "loss": 1.8268,
      "step": 17940
    },
    {
      "epoch": 0.03496469451898473,
      "grad_norm": 2.871506929397583,
      "learning_rate": 2.9909596808558385e-05,
      "loss": 1.9375,
      "step": 17955
    },
    {
      "epoch": 0.03499390479009499,
      "grad_norm": 3.8660874366760254,
      "learning_rate": 2.990944584808043e-05,
      "loss": 1.678,
      "step": 17970
    },
    {
      "epoch": 0.03502311506120526,
      "grad_norm": 4.991106033325195,
      "learning_rate": 2.9909294762047935e-05,
      "loss": 1.9136,
      "step": 17985
    },
    {
      "epoch": 0.03505232533231552,
      "grad_norm": 3.8985953330993652,
      "learning_rate": 2.990914355046217e-05,
      "loss": 1.7425,
      "step": 18000
    },
    {
      "epoch": 0.03508153560342578,
      "grad_norm": 2.2547903060913086,
      "learning_rate": 2.9908992213324413e-05,
      "loss": 1.9613,
      "step": 18015
    },
    {
      "epoch": 0.035110745874536047,
      "grad_norm": 4.33575963973999,
      "learning_rate": 2.9908840750635936e-05,
      "loss": 2.0328,
      "step": 18030
    },
    {
      "epoch": 0.03513995614564631,
      "grad_norm": 3.8632030487060547,
      "learning_rate": 2.9908689162398012e-05,
      "loss": 1.7863,
      "step": 18045
    },
    {
      "epoch": 0.03516916641675657,
      "grad_norm": 4.932112216949463,
      "learning_rate": 2.9908537448611927e-05,
      "loss": 1.7159,
      "step": 18060
    },
    {
      "epoch": 0.03519837668786683,
      "grad_norm": 2.3018031120300293,
      "learning_rate": 2.9908385609278943e-05,
      "loss": 1.9104,
      "step": 18075
    },
    {
      "epoch": 0.0352275869589771,
      "grad_norm": 2.7413060665130615,
      "learning_rate": 2.990823364440035e-05,
      "loss": 1.6733,
      "step": 18090
    },
    {
      "epoch": 0.03525679723008736,
      "grad_norm": 3.928980588912964,
      "learning_rate": 2.9908081553977424e-05,
      "loss": 1.9619,
      "step": 18105
    },
    {
      "epoch": 0.03528600750119762,
      "grad_norm": 1.6751407384872437,
      "learning_rate": 2.9907929338011447e-05,
      "loss": 1.8283,
      "step": 18120
    },
    {
      "epoch": 0.03531521777230789,
      "grad_norm": 4.515170574188232,
      "learning_rate": 2.99077769965037e-05,
      "loss": 1.7808,
      "step": 18135
    },
    {
      "epoch": 0.03534442804341815,
      "grad_norm": 3.367410182952881,
      "learning_rate": 2.9907624529455468e-05,
      "loss": 1.9427,
      "step": 18150
    },
    {
      "epoch": 0.03537363831452841,
      "grad_norm": 2.9150290489196777,
      "learning_rate": 2.990747193686803e-05,
      "loss": 2.021,
      "step": 18165
    },
    {
      "epoch": 0.03540284858563867,
      "grad_norm": 3.456367015838623,
      "learning_rate": 2.9907319218742677e-05,
      "loss": 1.8711,
      "step": 18180
    },
    {
      "epoch": 0.03543205885674894,
      "grad_norm": 2.504117727279663,
      "learning_rate": 2.990716637508069e-05,
      "loss": 1.9605,
      "step": 18195
    },
    {
      "epoch": 0.0354612691278592,
      "grad_norm": 3.5525166988372803,
      "learning_rate": 2.990701340588336e-05,
      "loss": 1.6192,
      "step": 18210
    },
    {
      "epoch": 0.03549047939896946,
      "grad_norm": 4.048015594482422,
      "learning_rate": 2.9906860311151973e-05,
      "loss": 1.9809,
      "step": 18225
    },
    {
      "epoch": 0.03551968967007973,
      "grad_norm": 3.0211918354034424,
      "learning_rate": 2.990670709088782e-05,
      "loss": 1.8898,
      "step": 18240
    },
    {
      "epoch": 0.03554889994118999,
      "grad_norm": 2.4371862411499023,
      "learning_rate": 2.9906553745092184e-05,
      "loss": 1.8836,
      "step": 18255
    },
    {
      "epoch": 0.03557811021230025,
      "grad_norm": 3.461329460144043,
      "learning_rate": 2.990640027376637e-05,
      "loss": 1.7893,
      "step": 18270
    },
    {
      "epoch": 0.03560732048341051,
      "grad_norm": 2.6082050800323486,
      "learning_rate": 2.990624667691166e-05,
      "loss": 1.94,
      "step": 18285
    },
    {
      "epoch": 0.03563653075452078,
      "grad_norm": 2.8006784915924072,
      "learning_rate": 2.990609295452935e-05,
      "loss": 1.7607,
      "step": 18300
    },
    {
      "epoch": 0.03566574102563104,
      "grad_norm": 4.304664611816406,
      "learning_rate": 2.990593910662073e-05,
      "loss": 1.8973,
      "step": 18315
    },
    {
      "epoch": 0.0356949512967413,
      "grad_norm": 2.063380479812622,
      "learning_rate": 2.9905785133187108e-05,
      "loss": 1.7456,
      "step": 18330
    },
    {
      "epoch": 0.03572416156785156,
      "grad_norm": 4.626766681671143,
      "learning_rate": 2.9905631034229772e-05,
      "loss": 1.8402,
      "step": 18345
    },
    {
      "epoch": 0.03575337183896183,
      "grad_norm": 4.0876288414001465,
      "learning_rate": 2.9905476809750017e-05,
      "loss": 1.8239,
      "step": 18360
    },
    {
      "epoch": 0.03578258211007209,
      "grad_norm": 2.5533339977264404,
      "learning_rate": 2.9905322459749148e-05,
      "loss": 1.9771,
      "step": 18375
    },
    {
      "epoch": 0.03581179238118235,
      "grad_norm": 2.268286943435669,
      "learning_rate": 2.990516798422846e-05,
      "loss": 1.8886,
      "step": 18390
    },
    {
      "epoch": 0.03584100265229262,
      "grad_norm": 2.343240261077881,
      "learning_rate": 2.9905013383189257e-05,
      "loss": 1.8676,
      "step": 18405
    },
    {
      "epoch": 0.03587021292340288,
      "grad_norm": 3.544220209121704,
      "learning_rate": 2.990485865663284e-05,
      "loss": 1.7922,
      "step": 18420
    },
    {
      "epoch": 0.03589942319451314,
      "grad_norm": 3.608947277069092,
      "learning_rate": 2.9904703804560516e-05,
      "loss": 1.7506,
      "step": 18435
    },
    {
      "epoch": 0.0359286334656234,
      "grad_norm": 2.520986557006836,
      "learning_rate": 2.990454882697358e-05,
      "loss": 1.9715,
      "step": 18450
    },
    {
      "epoch": 0.03595784373673367,
      "grad_norm": 2.598273277282715,
      "learning_rate": 2.9904393723873342e-05,
      "loss": 2.0018,
      "step": 18465
    },
    {
      "epoch": 0.03598705400784393,
      "grad_norm": 3.4648282527923584,
      "learning_rate": 2.990423849526111e-05,
      "loss": 1.8609,
      "step": 18480
    },
    {
      "epoch": 0.03601626427895419,
      "grad_norm": 4.387876510620117,
      "learning_rate": 2.9904083141138194e-05,
      "loss": 1.9949,
      "step": 18495
    },
    {
      "epoch": 0.03604547455006446,
      "grad_norm": 2.121781349182129,
      "learning_rate": 2.9903927661505888e-05,
      "loss": 1.91,
      "step": 18510
    },
    {
      "epoch": 0.03607468482117472,
      "grad_norm": 2.5395514965057373,
      "learning_rate": 2.9903772056365516e-05,
      "loss": 1.7663,
      "step": 18525
    },
    {
      "epoch": 0.03610389509228498,
      "grad_norm": 4.20078706741333,
      "learning_rate": 2.9903616325718385e-05,
      "loss": 1.8051,
      "step": 18540
    },
    {
      "epoch": 0.03613310536339524,
      "grad_norm": 3.0787668228149414,
      "learning_rate": 2.9903460469565802e-05,
      "loss": 1.7457,
      "step": 18555
    },
    {
      "epoch": 0.03616231563450551,
      "grad_norm": 2.493914842605591,
      "learning_rate": 2.9903304487909084e-05,
      "loss": 1.841,
      "step": 18570
    },
    {
      "epoch": 0.03619152590561577,
      "grad_norm": 4.367134094238281,
      "learning_rate": 2.9903148380749543e-05,
      "loss": 1.8202,
      "step": 18585
    },
    {
      "epoch": 0.03622073617672603,
      "grad_norm": 4.713270664215088,
      "learning_rate": 2.990299214808849e-05,
      "loss": 1.7427,
      "step": 18600
    },
    {
      "epoch": 0.0362499464478363,
      "grad_norm": 2.3630452156066895,
      "learning_rate": 2.9902835789927246e-05,
      "loss": 1.8379,
      "step": 18615
    },
    {
      "epoch": 0.03627915671894656,
      "grad_norm": 4.240983009338379,
      "learning_rate": 2.9902679306267127e-05,
      "loss": 1.8876,
      "step": 18630
    },
    {
      "epoch": 0.03630836699005682,
      "grad_norm": 2.1189780235290527,
      "learning_rate": 2.990252269710945e-05,
      "loss": 1.7273,
      "step": 18645
    },
    {
      "epoch": 0.036337577261167084,
      "grad_norm": 4.207765579223633,
      "learning_rate": 2.9902365962455533e-05,
      "loss": 1.7961,
      "step": 18660
    },
    {
      "epoch": 0.03636678753227735,
      "grad_norm": 4.526651859283447,
      "learning_rate": 2.9902209102306694e-05,
      "loss": 1.8292,
      "step": 18675
    },
    {
      "epoch": 0.03639599780338761,
      "grad_norm": 4.371520042419434,
      "learning_rate": 2.990205211666426e-05,
      "loss": 1.9646,
      "step": 18690
    },
    {
      "epoch": 0.036425208074497874,
      "grad_norm": 3.766552209854126,
      "learning_rate": 2.990189500552955e-05,
      "loss": 1.6872,
      "step": 18705
    },
    {
      "epoch": 0.03645441834560814,
      "grad_norm": 3.610987901687622,
      "learning_rate": 2.9901737768903882e-05,
      "loss": 1.8095,
      "step": 18720
    },
    {
      "epoch": 0.0364836286167184,
      "grad_norm": 2.77374529838562,
      "learning_rate": 2.9901580406788583e-05,
      "loss": 1.7853,
      "step": 18735
    },
    {
      "epoch": 0.036512838887828664,
      "grad_norm": 1.9032909870147705,
      "learning_rate": 2.9901422919184984e-05,
      "loss": 2.066,
      "step": 18750
    },
    {
      "epoch": 0.036542049158938925,
      "grad_norm": 6.839906215667725,
      "learning_rate": 2.9901265306094406e-05,
      "loss": 2.0159,
      "step": 18765
    },
    {
      "epoch": 0.03657125943004919,
      "grad_norm": 3.8178064823150635,
      "learning_rate": 2.9901107567518177e-05,
      "loss": 1.9257,
      "step": 18780
    },
    {
      "epoch": 0.036600469701159453,
      "grad_norm": 2.0359580516815186,
      "learning_rate": 2.990094970345762e-05,
      "loss": 1.8261,
      "step": 18795
    },
    {
      "epoch": 0.036629679972269714,
      "grad_norm": 3.5759871006011963,
      "learning_rate": 2.9900791713914078e-05,
      "loss": 1.9477,
      "step": 18810
    },
    {
      "epoch": 0.036658890243379975,
      "grad_norm": 2.471161127090454,
      "learning_rate": 2.990063359888887e-05,
      "loss": 1.9519,
      "step": 18825
    },
    {
      "epoch": 0.03668810051449024,
      "grad_norm": 4.123219966888428,
      "learning_rate": 2.9900475358383327e-05,
      "loss": 1.9215,
      "step": 18840
    },
    {
      "epoch": 0.036717310785600504,
      "grad_norm": 2.517751693725586,
      "learning_rate": 2.9900316992398793e-05,
      "loss": 1.9601,
      "step": 18855
    },
    {
      "epoch": 0.036746521056710765,
      "grad_norm": 3.0052502155303955,
      "learning_rate": 2.9900158500936587e-05,
      "loss": 1.8937,
      "step": 18870
    },
    {
      "epoch": 0.03677573132782103,
      "grad_norm": 3.820679187774658,
      "learning_rate": 2.9899999883998057e-05,
      "loss": 1.7623,
      "step": 18885
    },
    {
      "epoch": 0.036804941598931294,
      "grad_norm": 4.087255954742432,
      "learning_rate": 2.989984114158453e-05,
      "loss": 1.7651,
      "step": 18900
    },
    {
      "epoch": 0.036834151870041555,
      "grad_norm": 2.3399763107299805,
      "learning_rate": 2.989968227369734e-05,
      "loss": 1.8389,
      "step": 18915
    },
    {
      "epoch": 0.036863362141151816,
      "grad_norm": 4.5868072509765625,
      "learning_rate": 2.9899523280337836e-05,
      "loss": 1.9238,
      "step": 18930
    },
    {
      "epoch": 0.036892572412262084,
      "grad_norm": 2.5856025218963623,
      "learning_rate": 2.989936416150735e-05,
      "loss": 1.8044,
      "step": 18945
    },
    {
      "epoch": 0.036921782683372345,
      "grad_norm": 2.635301351547241,
      "learning_rate": 2.9899204917207222e-05,
      "loss": 1.9433,
      "step": 18960
    },
    {
      "epoch": 0.036950992954482606,
      "grad_norm": 2.7841100692749023,
      "learning_rate": 2.9899045547438792e-05,
      "loss": 1.8401,
      "step": 18975
    },
    {
      "epoch": 0.036980203225592874,
      "grad_norm": 3.0509424209594727,
      "learning_rate": 2.9898886052203407e-05,
      "loss": 1.8828,
      "step": 18990
    },
    {
      "epoch": 0.037009413496703135,
      "grad_norm": 2.99225115776062,
      "learning_rate": 2.9898726431502402e-05,
      "loss": 1.8158,
      "step": 19005
    },
    {
      "epoch": 0.037038623767813396,
      "grad_norm": 2.5677740573883057,
      "learning_rate": 2.9898566685337135e-05,
      "loss": 1.8706,
      "step": 19020
    },
    {
      "epoch": 0.03706783403892366,
      "grad_norm": 3.502387762069702,
      "learning_rate": 2.9898406813708934e-05,
      "loss": 1.9128,
      "step": 19035
    },
    {
      "epoch": 0.037097044310033925,
      "grad_norm": 3.5594568252563477,
      "learning_rate": 2.9898246816619162e-05,
      "loss": 1.732,
      "step": 19050
    },
    {
      "epoch": 0.037126254581144186,
      "grad_norm": 2.2263100147247314,
      "learning_rate": 2.989808669406915e-05,
      "loss": 1.9151,
      "step": 19065
    },
    {
      "epoch": 0.03715546485225445,
      "grad_norm": 3.859412670135498,
      "learning_rate": 2.989792644606026e-05,
      "loss": 1.9925,
      "step": 19080
    },
    {
      "epoch": 0.037184675123364715,
      "grad_norm": 2.2860965728759766,
      "learning_rate": 2.9897766072593834e-05,
      "loss": 1.9347,
      "step": 19095
    },
    {
      "epoch": 0.037213885394474976,
      "grad_norm": 3.5833935737609863,
      "learning_rate": 2.9897605573671224e-05,
      "loss": 1.7977,
      "step": 19110
    },
    {
      "epoch": 0.03724309566558524,
      "grad_norm": 3.612175226211548,
      "learning_rate": 2.989744494929378e-05,
      "loss": 1.9048,
      "step": 19125
    },
    {
      "epoch": 0.0372723059366955,
      "grad_norm": 2.6882388591766357,
      "learning_rate": 2.989728419946286e-05,
      "loss": 1.814,
      "step": 19140
    },
    {
      "epoch": 0.037301516207805765,
      "grad_norm": 2.753767728805542,
      "learning_rate": 2.989712332417982e-05,
      "loss": 1.8261,
      "step": 19155
    },
    {
      "epoch": 0.037330726478916026,
      "grad_norm": 3.5201592445373535,
      "learning_rate": 2.9896962323446004e-05,
      "loss": 1.9855,
      "step": 19170
    },
    {
      "epoch": 0.03735993675002629,
      "grad_norm": 2.892357587814331,
      "learning_rate": 2.9896801197262773e-05,
      "loss": 1.8325,
      "step": 19185
    },
    {
      "epoch": 0.037389147021136555,
      "grad_norm": 3.37709903717041,
      "learning_rate": 2.9896639945631483e-05,
      "loss": 1.9563,
      "step": 19200
    },
    {
      "epoch": 0.037418357292246816,
      "grad_norm": 1.9762299060821533,
      "learning_rate": 2.9896478568553492e-05,
      "loss": 1.8228,
      "step": 19215
    },
    {
      "epoch": 0.03744756756335708,
      "grad_norm": 5.543561935424805,
      "learning_rate": 2.9896317066030162e-05,
      "loss": 1.7656,
      "step": 19230
    },
    {
      "epoch": 0.03747677783446734,
      "grad_norm": 1.8155004978179932,
      "learning_rate": 2.9896155438062852e-05,
      "loss": 1.8033,
      "step": 19245
    },
    {
      "epoch": 0.037505988105577606,
      "grad_norm": 3.613931655883789,
      "learning_rate": 2.989599368465292e-05,
      "loss": 1.9275,
      "step": 19260
    },
    {
      "epoch": 0.03753519837668787,
      "grad_norm": 2.2457938194274902,
      "learning_rate": 2.989583180580173e-05,
      "loss": 1.9459,
      "step": 19275
    },
    {
      "epoch": 0.03756440864779813,
      "grad_norm": 2.4014956951141357,
      "learning_rate": 2.9895669801510646e-05,
      "loss": 1.7973,
      "step": 19290
    },
    {
      "epoch": 0.037593618918908396,
      "grad_norm": 2.1578032970428467,
      "learning_rate": 2.9895507671781032e-05,
      "loss": 1.8551,
      "step": 19305
    },
    {
      "epoch": 0.03762282919001866,
      "grad_norm": 2.952676296234131,
      "learning_rate": 2.9895345416614254e-05,
      "loss": 1.7808,
      "step": 19320
    },
    {
      "epoch": 0.03765203946112892,
      "grad_norm": 3.352534770965576,
      "learning_rate": 2.989518303601167e-05,
      "loss": 1.8889,
      "step": 19335
    },
    {
      "epoch": 0.03768124973223918,
      "grad_norm": 5.104668617248535,
      "learning_rate": 2.9895020529974667e-05,
      "loss": 1.9522,
      "step": 19350
    },
    {
      "epoch": 0.03771046000334945,
      "grad_norm": 3.0823004245758057,
      "learning_rate": 2.9894857898504595e-05,
      "loss": 1.916,
      "step": 19365
    },
    {
      "epoch": 0.03773967027445971,
      "grad_norm": 2.544391393661499,
      "learning_rate": 2.9894695141602824e-05,
      "loss": 1.9019,
      "step": 19380
    },
    {
      "epoch": 0.03776888054556997,
      "grad_norm": 2.4823343753814697,
      "learning_rate": 2.989453225927074e-05,
      "loss": 1.7927,
      "step": 19395
    },
    {
      "epoch": 0.03779809081668023,
      "grad_norm": 1.8794801235198975,
      "learning_rate": 2.98943692515097e-05,
      "loss": 2.0112,
      "step": 19410
    },
    {
      "epoch": 0.0378273010877905,
      "grad_norm": 2.5527946949005127,
      "learning_rate": 2.9894206118321083e-05,
      "loss": 1.8095,
      "step": 19425
    },
    {
      "epoch": 0.03785651135890076,
      "grad_norm": 3.4643588066101074,
      "learning_rate": 2.9894042859706265e-05,
      "loss": 1.7435,
      "step": 19440
    },
    {
      "epoch": 0.03788572163001102,
      "grad_norm": 3.913656234741211,
      "learning_rate": 2.9893879475666613e-05,
      "loss": 2.0963,
      "step": 19455
    },
    {
      "epoch": 0.03791493190112129,
      "grad_norm": 5.583499431610107,
      "learning_rate": 2.9893715966203502e-05,
      "loss": 1.8735,
      "step": 19470
    },
    {
      "epoch": 0.03794414217223155,
      "grad_norm": 1.9320791959762573,
      "learning_rate": 2.989355233131832e-05,
      "loss": 1.9904,
      "step": 19485
    },
    {
      "epoch": 0.03797335244334181,
      "grad_norm": 3.6746532917022705,
      "learning_rate": 2.9893388571012443e-05,
      "loss": 1.8231,
      "step": 19500
    },
    {
      "epoch": 0.03800256271445207,
      "grad_norm": 2.002924680709839,
      "learning_rate": 2.989322468528724e-05,
      "loss": 1.989,
      "step": 19515
    },
    {
      "epoch": 0.03803177298556234,
      "grad_norm": 3.4807581901550293,
      "learning_rate": 2.98930606741441e-05,
      "loss": 1.849,
      "step": 19530
    },
    {
      "epoch": 0.0380609832566726,
      "grad_norm": 4.430756568908691,
      "learning_rate": 2.98928965375844e-05,
      "loss": 1.9161,
      "step": 19545
    },
    {
      "epoch": 0.03809019352778286,
      "grad_norm": 3.5762476921081543,
      "learning_rate": 2.9892732275609525e-05,
      "loss": 1.8861,
      "step": 19560
    },
    {
      "epoch": 0.03811940379889313,
      "grad_norm": 2.6605224609375,
      "learning_rate": 2.9892567888220855e-05,
      "loss": 1.9023,
      "step": 19575
    },
    {
      "epoch": 0.03814861407000339,
      "grad_norm": 2.929426431655884,
      "learning_rate": 2.9892403375419778e-05,
      "loss": 1.7741,
      "step": 19590
    },
    {
      "epoch": 0.03817782434111365,
      "grad_norm": 3.075709342956543,
      "learning_rate": 2.9892238737207677e-05,
      "loss": 1.9575,
      "step": 19605
    },
    {
      "epoch": 0.03820703461222391,
      "grad_norm": 3.5462255477905273,
      "learning_rate": 2.9892073973585942e-05,
      "loss": 1.8336,
      "step": 19620
    },
    {
      "epoch": 0.03823624488333418,
      "grad_norm": 2.098604202270508,
      "learning_rate": 2.9891909084555954e-05,
      "loss": 1.8161,
      "step": 19635
    },
    {
      "epoch": 0.03826545515444444,
      "grad_norm": 3.255275249481201,
      "learning_rate": 2.9891744070119106e-05,
      "loss": 1.9291,
      "step": 19650
    },
    {
      "epoch": 0.0382946654255547,
      "grad_norm": 3.4731807708740234,
      "learning_rate": 2.9891578930276787e-05,
      "loss": 1.9403,
      "step": 19665
    },
    {
      "epoch": 0.03832387569666497,
      "grad_norm": 2.0823373794555664,
      "learning_rate": 2.9891413665030387e-05,
      "loss": 1.8713,
      "step": 19680
    },
    {
      "epoch": 0.03835308596777523,
      "grad_norm": 2.6273398399353027,
      "learning_rate": 2.98912482743813e-05,
      "loss": 1.8764,
      "step": 19695
    },
    {
      "epoch": 0.03838229623888549,
      "grad_norm": 3.4412078857421875,
      "learning_rate": 2.9891082758330915e-05,
      "loss": 1.8471,
      "step": 19710
    },
    {
      "epoch": 0.03841150650999575,
      "grad_norm": 2.652684211730957,
      "learning_rate": 2.9890917116880625e-05,
      "loss": 2.0607,
      "step": 19725
    },
    {
      "epoch": 0.03844071678110602,
      "grad_norm": 2.8258442878723145,
      "learning_rate": 2.989075135003183e-05,
      "loss": 1.8497,
      "step": 19740
    },
    {
      "epoch": 0.03846992705221628,
      "grad_norm": 3.239922285079956,
      "learning_rate": 2.9890585457785923e-05,
      "loss": 1.7763,
      "step": 19755
    },
    {
      "epoch": 0.03849913732332654,
      "grad_norm": 2.4102368354797363,
      "learning_rate": 2.9890419440144303e-05,
      "loss": 1.9937,
      "step": 19770
    },
    {
      "epoch": 0.03852834759443681,
      "grad_norm": 3.4451348781585693,
      "learning_rate": 2.989025329710837e-05,
      "loss": 1.8446,
      "step": 19785
    },
    {
      "epoch": 0.03855755786554707,
      "grad_norm": 2.9538424015045166,
      "learning_rate": 2.9890087028679517e-05,
      "loss": 1.8478,
      "step": 19800
    },
    {
      "epoch": 0.03858676813665733,
      "grad_norm": 2.6566977500915527,
      "learning_rate": 2.9889920634859144e-05,
      "loss": 1.7005,
      "step": 19815
    },
    {
      "epoch": 0.03861597840776759,
      "grad_norm": 1.9971015453338623,
      "learning_rate": 2.988975411564866e-05,
      "loss": 1.8428,
      "step": 19830
    },
    {
      "epoch": 0.03864518867887786,
      "grad_norm": 2.648861885070801,
      "learning_rate": 2.9889587471049456e-05,
      "loss": 2.0656,
      "step": 19845
    },
    {
      "epoch": 0.03867439894998812,
      "grad_norm": 4.70402193069458,
      "learning_rate": 2.9889420701062947e-05,
      "loss": 1.8098,
      "step": 19860
    },
    {
      "epoch": 0.03870360922109838,
      "grad_norm": 4.031128883361816,
      "learning_rate": 2.988925380569053e-05,
      "loss": 1.9847,
      "step": 19875
    },
    {
      "epoch": 0.038732819492208644,
      "grad_norm": 3.7528605461120605,
      "learning_rate": 2.988908678493361e-05,
      "loss": 1.8671,
      "step": 19890
    },
    {
      "epoch": 0.03876202976331891,
      "grad_norm": 2.24607253074646,
      "learning_rate": 2.9888919638793604e-05,
      "loss": 1.8053,
      "step": 19905
    },
    {
      "epoch": 0.03879124003442917,
      "grad_norm": 2.6132078170776367,
      "learning_rate": 2.9888752367271903e-05,
      "loss": 2.0429,
      "step": 19920
    },
    {
      "epoch": 0.03882045030553943,
      "grad_norm": 2.334711790084839,
      "learning_rate": 2.988858497036993e-05,
      "loss": 1.7526,
      "step": 19935
    },
    {
      "epoch": 0.0388496605766497,
      "grad_norm": 3.065218210220337,
      "learning_rate": 2.988841744808909e-05,
      "loss": 1.9287,
      "step": 19950
    },
    {
      "epoch": 0.03887887084775996,
      "grad_norm": 4.216036796569824,
      "learning_rate": 2.9888249800430787e-05,
      "loss": 1.7889,
      "step": 19965
    },
    {
      "epoch": 0.03890808111887022,
      "grad_norm": 4.185488224029541,
      "learning_rate": 2.988808202739644e-05,
      "loss": 1.865,
      "step": 19980
    },
    {
      "epoch": 0.038937291389980484,
      "grad_norm": 2.3708608150482178,
      "learning_rate": 2.9887914128987465e-05,
      "loss": 1.852,
      "step": 19995
    },
    {
      "epoch": 0.03896650166109075,
      "grad_norm": 4.136464595794678,
      "learning_rate": 2.9887746105205264e-05,
      "loss": 1.8378,
      "step": 20010
    },
    {
      "epoch": 0.03899571193220101,
      "grad_norm": 3.3061976432800293,
      "learning_rate": 2.9887577956051263e-05,
      "loss": 1.8702,
      "step": 20025
    },
    {
      "epoch": 0.039024922203311274,
      "grad_norm": 3.5822839736938477,
      "learning_rate": 2.9887409681526876e-05,
      "loss": 1.87,
      "step": 20040
    },
    {
      "epoch": 0.03905413247442154,
      "grad_norm": 4.132532119750977,
      "learning_rate": 2.9887241281633518e-05,
      "loss": 2.051,
      "step": 20055
    },
    {
      "epoch": 0.0390833427455318,
      "grad_norm": 3.9095962047576904,
      "learning_rate": 2.9887072756372606e-05,
      "loss": 1.8271,
      "step": 20070
    },
    {
      "epoch": 0.039112553016642064,
      "grad_norm": 3.028393507003784,
      "learning_rate": 2.988690410574556e-05,
      "loss": 1.825,
      "step": 20085
    },
    {
      "epoch": 0.039141763287752325,
      "grad_norm": 1.7903603315353394,
      "learning_rate": 2.98867353297538e-05,
      "loss": 1.9565,
      "step": 20100
    },
    {
      "epoch": 0.03917097355886259,
      "grad_norm": 2.168895959854126,
      "learning_rate": 2.988656642839875e-05,
      "loss": 1.8483,
      "step": 20115
    },
    {
      "epoch": 0.039200183829972854,
      "grad_norm": 7.789179801940918,
      "learning_rate": 2.988639740168183e-05,
      "loss": 2.0627,
      "step": 20130
    },
    {
      "epoch": 0.039229394101083115,
      "grad_norm": 2.033658981323242,
      "learning_rate": 2.9886228249604464e-05,
      "loss": 1.8905,
      "step": 20145
    },
    {
      "epoch": 0.03925860437219338,
      "grad_norm": 3.6176931858062744,
      "learning_rate": 2.9886058972168076e-05,
      "loss": 1.8981,
      "step": 20160
    },
    {
      "epoch": 0.039287814643303644,
      "grad_norm": 3.776540517807007,
      "learning_rate": 2.9885889569374088e-05,
      "loss": 1.7417,
      "step": 20175
    },
    {
      "epoch": 0.039317024914413905,
      "grad_norm": 2.138796329498291,
      "learning_rate": 2.9885720041223934e-05,
      "loss": 1.8845,
      "step": 20190
    },
    {
      "epoch": 0.039346235185524166,
      "grad_norm": 5.0401482582092285,
      "learning_rate": 2.988555038771904e-05,
      "loss": 1.8966,
      "step": 20205
    },
    {
      "epoch": 0.039375445456634434,
      "grad_norm": 2.2886552810668945,
      "learning_rate": 2.9885380608860827e-05,
      "loss": 1.8898,
      "step": 20220
    },
    {
      "epoch": 0.039404655727744695,
      "grad_norm": 2.286959648132324,
      "learning_rate": 2.9885210704650734e-05,
      "loss": 1.9597,
      "step": 20235
    },
    {
      "epoch": 0.039433865998854956,
      "grad_norm": 2.538752794265747,
      "learning_rate": 2.988504067509019e-05,
      "loss": 1.8451,
      "step": 20250
    },
    {
      "epoch": 0.039463076269965223,
      "grad_norm": 3.4922502040863037,
      "learning_rate": 2.988487052018062e-05,
      "loss": 1.8717,
      "step": 20265
    },
    {
      "epoch": 0.039492286541075484,
      "grad_norm": 3.5919086933135986,
      "learning_rate": 2.9884700239923467e-05,
      "loss": 1.8182,
      "step": 20280
    },
    {
      "epoch": 0.039521496812185745,
      "grad_norm": 1.9253063201904297,
      "learning_rate": 2.988452983432016e-05,
      "loss": 1.8067,
      "step": 20295
    },
    {
      "epoch": 0.039550707083296006,
      "grad_norm": 3.168278217315674,
      "learning_rate": 2.9884359303372127e-05,
      "loss": 1.9619,
      "step": 20310
    },
    {
      "epoch": 0.039579917354406274,
      "grad_norm": 2.574296236038208,
      "learning_rate": 2.9884188647080816e-05,
      "loss": 1.9037,
      "step": 20325
    },
    {
      "epoch": 0.039609127625516535,
      "grad_norm": 3.87908673286438,
      "learning_rate": 2.9884017865447657e-05,
      "loss": 1.8106,
      "step": 20340
    },
    {
      "epoch": 0.039638337896626796,
      "grad_norm": 3.060088872909546,
      "learning_rate": 2.9883846958474093e-05,
      "loss": 1.8878,
      "step": 20355
    },
    {
      "epoch": 0.039667548167737064,
      "grad_norm": 2.6092071533203125,
      "learning_rate": 2.988367592616156e-05,
      "loss": 2.0189,
      "step": 20370
    },
    {
      "epoch": 0.039696758438847325,
      "grad_norm": 2.688831329345703,
      "learning_rate": 2.9883504768511496e-05,
      "loss": 1.9439,
      "step": 20385
    },
    {
      "epoch": 0.039725968709957586,
      "grad_norm": 4.126415252685547,
      "learning_rate": 2.988333348552535e-05,
      "loss": 1.7334,
      "step": 20400
    },
    {
      "epoch": 0.03975517898106785,
      "grad_norm": 2.5536105632781982,
      "learning_rate": 2.988316207720455e-05,
      "loss": 1.8741,
      "step": 20415
    },
    {
      "epoch": 0.039784389252178115,
      "grad_norm": 4.146584510803223,
      "learning_rate": 2.9882990543550557e-05,
      "loss": 1.839,
      "step": 20430
    },
    {
      "epoch": 0.039813599523288376,
      "grad_norm": 4.531203746795654,
      "learning_rate": 2.9882818884564805e-05,
      "loss": 1.801,
      "step": 20445
    },
    {
      "epoch": 0.03984280979439864,
      "grad_norm": 4.3206562995910645,
      "learning_rate": 2.988264710024874e-05,
      "loss": 1.9127,
      "step": 20460
    },
    {
      "epoch": 0.0398720200655089,
      "grad_norm": 2.258627414703369,
      "learning_rate": 2.9882475190603815e-05,
      "loss": 1.9329,
      "step": 20475
    },
    {
      "epoch": 0.039901230336619166,
      "grad_norm": 2.874171018600464,
      "learning_rate": 2.988230315563147e-05,
      "loss": 1.8632,
      "step": 20490
    },
    {
      "epoch": 0.03993044060772943,
      "grad_norm": 2.2248058319091797,
      "learning_rate": 2.988213099533316e-05,
      "loss": 1.7653,
      "step": 20505
    },
    {
      "epoch": 0.03995965087883969,
      "grad_norm": 2.2211756706237793,
      "learning_rate": 2.988195870971033e-05,
      "loss": 2.0817,
      "step": 20520
    },
    {
      "epoch": 0.039988861149949956,
      "grad_norm": 2.266099214553833,
      "learning_rate": 2.9881786298764432e-05,
      "loss": 2.0993,
      "step": 20535
    },
    {
      "epoch": 0.04001807142106022,
      "grad_norm": 3.2420692443847656,
      "learning_rate": 2.988161376249692e-05,
      "loss": 1.8448,
      "step": 20550
    },
    {
      "epoch": 0.04004728169217048,
      "grad_norm": 2.1219029426574707,
      "learning_rate": 2.9881441100909244e-05,
      "loss": 1.9547,
      "step": 20565
    },
    {
      "epoch": 0.04007649196328074,
      "grad_norm": 4.585850238800049,
      "learning_rate": 2.988126831400286e-05,
      "loss": 1.9619,
      "step": 20580
    },
    {
      "epoch": 0.04010570223439101,
      "grad_norm": 3.4670188426971436,
      "learning_rate": 2.9881095401779224e-05,
      "loss": 2.0,
      "step": 20595
    },
    {
      "epoch": 0.04013491250550127,
      "grad_norm": 2.0285186767578125,
      "learning_rate": 2.9880922364239787e-05,
      "loss": 1.9434,
      "step": 20610
    },
    {
      "epoch": 0.04016412277661153,
      "grad_norm": 1.873279094696045,
      "learning_rate": 2.9880749201386014e-05,
      "loss": 1.8546,
      "step": 20625
    },
    {
      "epoch": 0.040193333047721796,
      "grad_norm": 2.1878137588500977,
      "learning_rate": 2.9880575913219354e-05,
      "loss": 1.8624,
      "step": 20640
    },
    {
      "epoch": 0.04022254331883206,
      "grad_norm": 3.9366180896759033,
      "learning_rate": 2.988040249974128e-05,
      "loss": 1.9143,
      "step": 20655
    },
    {
      "epoch": 0.04025175358994232,
      "grad_norm": 2.3610141277313232,
      "learning_rate": 2.9880228960953236e-05,
      "loss": 1.9616,
      "step": 20670
    },
    {
      "epoch": 0.04028096386105258,
      "grad_norm": 2.8658604621887207,
      "learning_rate": 2.9880055296856695e-05,
      "loss": 1.7601,
      "step": 20685
    },
    {
      "epoch": 0.04031017413216285,
      "grad_norm": 4.454057216644287,
      "learning_rate": 2.9879881507453112e-05,
      "loss": 1.7578,
      "step": 20700
    },
    {
      "epoch": 0.04033938440327311,
      "grad_norm": 2.244455575942993,
      "learning_rate": 2.9879707592743957e-05,
      "loss": 1.8334,
      "step": 20715
    },
    {
      "epoch": 0.04036859467438337,
      "grad_norm": 1.61302649974823,
      "learning_rate": 2.987953355273069e-05,
      "loss": 1.9746,
      "step": 20730
    },
    {
      "epoch": 0.04039780494549364,
      "grad_norm": 2.8323678970336914,
      "learning_rate": 2.987935938741478e-05,
      "loss": 2.1251,
      "step": 20745
    },
    {
      "epoch": 0.0404270152166039,
      "grad_norm": 2.418241262435913,
      "learning_rate": 2.987918509679769e-05,
      "loss": 2.0798,
      "step": 20760
    },
    {
      "epoch": 0.04045622548771416,
      "grad_norm": 3.465550661087036,
      "learning_rate": 2.987901068088089e-05,
      "loss": 1.8706,
      "step": 20775
    },
    {
      "epoch": 0.04048543575882442,
      "grad_norm": 4.148263931274414,
      "learning_rate": 2.987883613966585e-05,
      "loss": 1.7925,
      "step": 20790
    },
    {
      "epoch": 0.04051464602993469,
      "grad_norm": 2.4039666652679443,
      "learning_rate": 2.9878661473154037e-05,
      "loss": 1.8672,
      "step": 20805
    },
    {
      "epoch": 0.04054385630104495,
      "grad_norm": 2.12880277633667,
      "learning_rate": 2.9878486681346923e-05,
      "loss": 1.8978,
      "step": 20820
    },
    {
      "epoch": 0.04057306657215521,
      "grad_norm": 1.9612979888916016,
      "learning_rate": 2.987831176424598e-05,
      "loss": 1.9175,
      "step": 20835
    },
    {
      "epoch": 0.04060227684326548,
      "grad_norm": 2.161982297897339,
      "learning_rate": 2.9878136721852682e-05,
      "loss": 1.7752,
      "step": 20850
    },
    {
      "epoch": 0.04063148711437574,
      "grad_norm": 4.4879961013793945,
      "learning_rate": 2.9877961554168498e-05,
      "loss": 2.0857,
      "step": 20865
    },
    {
      "epoch": 0.040660697385486,
      "grad_norm": 4.1571364402771,
      "learning_rate": 2.9877786261194914e-05,
      "loss": 1.949,
      "step": 20880
    },
    {
      "epoch": 0.04068990765659626,
      "grad_norm": 3.3120033740997314,
      "learning_rate": 2.9877610842933397e-05,
      "loss": 1.8585,
      "step": 20895
    },
    {
      "epoch": 0.04071911792770653,
      "grad_norm": 3.193117618560791,
      "learning_rate": 2.9877435299385424e-05,
      "loss": 1.731,
      "step": 20910
    },
    {
      "epoch": 0.04074832819881679,
      "grad_norm": 2.375343084335327,
      "learning_rate": 2.987725963055248e-05,
      "loss": 1.8269,
      "step": 20925
    },
    {
      "epoch": 0.04077753846992705,
      "grad_norm": 2.3607242107391357,
      "learning_rate": 2.9877083836436036e-05,
      "loss": 1.8305,
      "step": 20940
    },
    {
      "epoch": 0.04080674874103731,
      "grad_norm": 3.0205342769622803,
      "learning_rate": 2.987690791703758e-05,
      "loss": 1.8631,
      "step": 20955
    },
    {
      "epoch": 0.04083595901214758,
      "grad_norm": 3.530947685241699,
      "learning_rate": 2.9876731872358585e-05,
      "loss": 1.8431,
      "step": 20970
    },
    {
      "epoch": 0.04086516928325784,
      "grad_norm": 2.8419220447540283,
      "learning_rate": 2.987655570240054e-05,
      "loss": 1.8519,
      "step": 20985
    },
    {
      "epoch": 0.0408943795543681,
      "grad_norm": 2.253532886505127,
      "learning_rate": 2.9876379407164933e-05,
      "loss": 1.8688,
      "step": 21000
    },
    {
      "epoch": 0.04092358982547837,
      "grad_norm": 1.9279251098632812,
      "learning_rate": 2.987620298665324e-05,
      "loss": 1.9111,
      "step": 21015
    },
    {
      "epoch": 0.04095280009658863,
      "grad_norm": 3.0645790100097656,
      "learning_rate": 2.987602644086695e-05,
      "loss": 1.8359,
      "step": 21030
    },
    {
      "epoch": 0.04098201036769889,
      "grad_norm": 2.9384896755218506,
      "learning_rate": 2.9875849769807544e-05,
      "loss": 1.747,
      "step": 21045
    },
    {
      "epoch": 0.04101122063880915,
      "grad_norm": 3.7138500213623047,
      "learning_rate": 2.987567297347652e-05,
      "loss": 1.8528,
      "step": 21060
    },
    {
      "epoch": 0.04104043090991942,
      "grad_norm": 4.440821170806885,
      "learning_rate": 2.987549605187536e-05,
      "loss": 1.9611,
      "step": 21075
    },
    {
      "epoch": 0.04106964118102968,
      "grad_norm": 3.0378854274749756,
      "learning_rate": 2.9875319005005552e-05,
      "loss": 2.012,
      "step": 21090
    },
    {
      "epoch": 0.04109885145213994,
      "grad_norm": 4.378201961517334,
      "learning_rate": 2.9875141832868598e-05,
      "loss": 1.7366,
      "step": 21105
    },
    {
      "epoch": 0.04112806172325021,
      "grad_norm": 4.794327735900879,
      "learning_rate": 2.9874964535465978e-05,
      "loss": 1.8728,
      "step": 21120
    },
    {
      "epoch": 0.04115727199436047,
      "grad_norm": 2.5137903690338135,
      "learning_rate": 2.987478711279919e-05,
      "loss": 1.9813,
      "step": 21135
    },
    {
      "epoch": 0.04118648226547073,
      "grad_norm": 2.2024412155151367,
      "learning_rate": 2.987460956486973e-05,
      "loss": 1.7721,
      "step": 21150
    },
    {
      "epoch": 0.04121569253658099,
      "grad_norm": 3.0043609142303467,
      "learning_rate": 2.987443189167909e-05,
      "loss": 1.9015,
      "step": 21165
    },
    {
      "epoch": 0.04124490280769126,
      "grad_norm": 4.627270698547363,
      "learning_rate": 2.9874254093228763e-05,
      "loss": 1.9487,
      "step": 21180
    },
    {
      "epoch": 0.04127411307880152,
      "grad_norm": 3.208395481109619,
      "learning_rate": 2.987407616952025e-05,
      "loss": 1.6989,
      "step": 21195
    },
    {
      "epoch": 0.04130332334991178,
      "grad_norm": 4.102930545806885,
      "learning_rate": 2.9873898120555055e-05,
      "loss": 2.0639,
      "step": 21210
    },
    {
      "epoch": 0.04133253362102205,
      "grad_norm": 3.846593141555786,
      "learning_rate": 2.987371994633467e-05,
      "loss": 1.7067,
      "step": 21225
    },
    {
      "epoch": 0.04136174389213231,
      "grad_norm": 3.6651105880737305,
      "learning_rate": 2.9873541646860597e-05,
      "loss": 1.8983,
      "step": 21240
    },
    {
      "epoch": 0.04139095416324257,
      "grad_norm": 3.715604543685913,
      "learning_rate": 2.987336322213434e-05,
      "loss": 1.6676,
      "step": 21255
    },
    {
      "epoch": 0.041420164434352834,
      "grad_norm": 3.0780601501464844,
      "learning_rate": 2.9873184672157395e-05,
      "loss": 1.9342,
      "step": 21270
    },
    {
      "epoch": 0.0414493747054631,
      "grad_norm": 3.385103225708008,
      "learning_rate": 2.9873005996931274e-05,
      "loss": 1.9494,
      "step": 21285
    },
    {
      "epoch": 0.04147858497657336,
      "grad_norm": 2.9652836322784424,
      "learning_rate": 2.9872827196457475e-05,
      "loss": 1.8491,
      "step": 21300
    },
    {
      "epoch": 0.041507795247683624,
      "grad_norm": 2.346210479736328,
      "learning_rate": 2.9872648270737507e-05,
      "loss": 1.7948,
      "step": 21315
    },
    {
      "epoch": 0.04153700551879389,
      "grad_norm": 2.0421793460845947,
      "learning_rate": 2.9872469219772877e-05,
      "loss": 1.7642,
      "step": 21330
    },
    {
      "epoch": 0.04156621578990415,
      "grad_norm": 3.2347426414489746,
      "learning_rate": 2.9872290043565094e-05,
      "loss": 1.9741,
      "step": 21345
    },
    {
      "epoch": 0.041595426061014414,
      "grad_norm": 2.8749160766601562,
      "learning_rate": 2.987211074211566e-05,
      "loss": 1.8364,
      "step": 21360
    },
    {
      "epoch": 0.041624636332124675,
      "grad_norm": 3.484539031982422,
      "learning_rate": 2.9871931315426094e-05,
      "loss": 1.7097,
      "step": 21375
    },
    {
      "epoch": 0.04165384660323494,
      "grad_norm": 3.105286121368408,
      "learning_rate": 2.98717517634979e-05,
      "loss": 1.8239,
      "step": 21390
    },
    {
      "epoch": 0.0416830568743452,
      "grad_norm": 3.804901361465454,
      "learning_rate": 2.9871572086332594e-05,
      "loss": 1.7356,
      "step": 21405
    },
    {
      "epoch": 0.041712267145455464,
      "grad_norm": 2.4536283016204834,
      "learning_rate": 2.9871392283931686e-05,
      "loss": 1.8601,
      "step": 21420
    },
    {
      "epoch": 0.04174147741656573,
      "grad_norm": 2.8864688873291016,
      "learning_rate": 2.9871212356296697e-05,
      "loss": 1.802,
      "step": 21435
    },
    {
      "epoch": 0.04177068768767599,
      "grad_norm": 3.4735238552093506,
      "learning_rate": 2.9871032303429133e-05,
      "loss": 1.9557,
      "step": 21450
    },
    {
      "epoch": 0.041799897958786254,
      "grad_norm": 4.21823263168335,
      "learning_rate": 2.9870852125330513e-05,
      "loss": 1.8513,
      "step": 21465
    },
    {
      "epoch": 0.041829108229896515,
      "grad_norm": 1.6568303108215332,
      "learning_rate": 2.9870671822002357e-05,
      "loss": 1.8443,
      "step": 21480
    },
    {
      "epoch": 0.04185831850100678,
      "grad_norm": 3.8471972942352295,
      "learning_rate": 2.9870491393446184e-05,
      "loss": 1.8831,
      "step": 21495
    },
    {
      "epoch": 0.041887528772117044,
      "grad_norm": 4.5181803703308105,
      "learning_rate": 2.987031083966351e-05,
      "loss": 2.0743,
      "step": 21510
    },
    {
      "epoch": 0.041916739043227305,
      "grad_norm": 2.7604262828826904,
      "learning_rate": 2.987013016065586e-05,
      "loss": 1.7647,
      "step": 21525
    },
    {
      "epoch": 0.041945949314337566,
      "grad_norm": 1.6995900869369507,
      "learning_rate": 2.986994935642475e-05,
      "loss": 1.9221,
      "step": 21540
    },
    {
      "epoch": 0.041975159585447834,
      "grad_norm": 3.400073766708374,
      "learning_rate": 2.9869768426971706e-05,
      "loss": 1.7217,
      "step": 21555
    },
    {
      "epoch": 0.042004369856558095,
      "grad_norm": 2.7783281803131104,
      "learning_rate": 2.986958737229825e-05,
      "loss": 1.8149,
      "step": 21570
    },
    {
      "epoch": 0.042033580127668356,
      "grad_norm": 1.9334073066711426,
      "learning_rate": 2.9869406192405904e-05,
      "loss": 1.7657,
      "step": 21585
    },
    {
      "epoch": 0.042062790398778624,
      "grad_norm": 3.244271993637085,
      "learning_rate": 2.9869224887296205e-05,
      "loss": 1.8122,
      "step": 21600
    },
    {
      "epoch": 0.042092000669888885,
      "grad_norm": 3.9582252502441406,
      "learning_rate": 2.9869043456970662e-05,
      "loss": 1.8296,
      "step": 21615
    },
    {
      "epoch": 0.042121210940999146,
      "grad_norm": 3.9118635654449463,
      "learning_rate": 2.986886190143082e-05,
      "loss": 1.9162,
      "step": 21630
    },
    {
      "epoch": 0.04215042121210941,
      "grad_norm": 3.3247733116149902,
      "learning_rate": 2.9868680220678198e-05,
      "loss": 1.9345,
      "step": 21645
    },
    {
      "epoch": 0.042179631483219675,
      "grad_norm": 4.189952373504639,
      "learning_rate": 2.9868498414714332e-05,
      "loss": 1.7394,
      "step": 21660
    },
    {
      "epoch": 0.042208841754329936,
      "grad_norm": 4.380448818206787,
      "learning_rate": 2.986831648354075e-05,
      "loss": 1.8411,
      "step": 21675
    },
    {
      "epoch": 0.0422380520254402,
      "grad_norm": 3.3280911445617676,
      "learning_rate": 2.986813442715898e-05,
      "loss": 1.7338,
      "step": 21690
    },
    {
      "epoch": 0.042267262296550465,
      "grad_norm": 2.147976875305176,
      "learning_rate": 2.9867952245570557e-05,
      "loss": 1.7418,
      "step": 21705
    },
    {
      "epoch": 0.042296472567660726,
      "grad_norm": 4.927828788757324,
      "learning_rate": 2.9867769938777025e-05,
      "loss": 1.9729,
      "step": 21720
    },
    {
      "epoch": 0.04232568283877099,
      "grad_norm": 2.1555793285369873,
      "learning_rate": 2.9867587506779903e-05,
      "loss": 1.9733,
      "step": 21735
    },
    {
      "epoch": 0.04235489310988125,
      "grad_norm": 4.826751232147217,
      "learning_rate": 2.986740494958074e-05,
      "loss": 1.9141,
      "step": 21750
    },
    {
      "epoch": 0.042384103380991515,
      "grad_norm": 2.62713360786438,
      "learning_rate": 2.986722226718107e-05,
      "loss": 1.7623,
      "step": 21765
    },
    {
      "epoch": 0.042413313652101776,
      "grad_norm": 2.9087536334991455,
      "learning_rate": 2.9867039459582422e-05,
      "loss": 2.0298,
      "step": 21780
    },
    {
      "epoch": 0.04244252392321204,
      "grad_norm": 3.2233824729919434,
      "learning_rate": 2.986685652678635e-05,
      "loss": 1.8063,
      "step": 21795
    },
    {
      "epoch": 0.042471734194322305,
      "grad_norm": 3.8226282596588135,
      "learning_rate": 2.9866673468794392e-05,
      "loss": 1.8173,
      "step": 21810
    },
    {
      "epoch": 0.042500944465432566,
      "grad_norm": 2.362210273742676,
      "learning_rate": 2.986649028560808e-05,
      "loss": 2.0107,
      "step": 21825
    },
    {
      "epoch": 0.04253015473654283,
      "grad_norm": 1.4686235189437866,
      "learning_rate": 2.9866306977228964e-05,
      "loss": 1.8339,
      "step": 21840
    },
    {
      "epoch": 0.04255936500765309,
      "grad_norm": 3.2174501419067383,
      "learning_rate": 2.9866123543658585e-05,
      "loss": 1.9372,
      "step": 21855
    },
    {
      "epoch": 0.042588575278763356,
      "grad_norm": 4.215010643005371,
      "learning_rate": 2.9865939984898494e-05,
      "loss": 1.7492,
      "step": 21870
    },
    {
      "epoch": 0.04261778554987362,
      "grad_norm": 6.015155792236328,
      "learning_rate": 2.9865756300950224e-05,
      "loss": 1.8146,
      "step": 21885
    },
    {
      "epoch": 0.04264699582098388,
      "grad_norm": 3.102923631668091,
      "learning_rate": 2.9865572491815336e-05,
      "loss": 1.9206,
      "step": 21900
    },
    {
      "epoch": 0.042676206092094146,
      "grad_norm": 3.5606796741485596,
      "learning_rate": 2.986538855749537e-05,
      "loss": 1.7442,
      "step": 21915
    },
    {
      "epoch": 0.04270541636320441,
      "grad_norm": 3.877696990966797,
      "learning_rate": 2.9865204497991874e-05,
      "loss": 1.8356,
      "step": 21930
    },
    {
      "epoch": 0.04273462663431467,
      "grad_norm": 2.7707462310791016,
      "learning_rate": 2.98650203133064e-05,
      "loss": 2.1114,
      "step": 21945
    },
    {
      "epoch": 0.04276383690542493,
      "grad_norm": 3.8471269607543945,
      "learning_rate": 2.9864836003440496e-05,
      "loss": 1.974,
      "step": 21960
    },
    {
      "epoch": 0.0427930471765352,
      "grad_norm": 4.265331268310547,
      "learning_rate": 2.9864651568395728e-05,
      "loss": 1.7851,
      "step": 21975
    },
    {
      "epoch": 0.04282225744764546,
      "grad_norm": 3.578641414642334,
      "learning_rate": 2.986446700817363e-05,
      "loss": 1.7996,
      "step": 21990
    },
    {
      "epoch": 0.04285146771875572,
      "grad_norm": 3.289726495742798,
      "learning_rate": 2.9864282322775768e-05,
      "loss": 1.7165,
      "step": 22005
    },
    {
      "epoch": 0.04288067798986598,
      "grad_norm": 1.883353590965271,
      "learning_rate": 2.986409751220369e-05,
      "loss": 1.8633,
      "step": 22020
    },
    {
      "epoch": 0.04290988826097625,
      "grad_norm": 1.7199418544769287,
      "learning_rate": 2.9863912576458955e-05,
      "loss": 1.7696,
      "step": 22035
    },
    {
      "epoch": 0.04293909853208651,
      "grad_norm": 3.445349931716919,
      "learning_rate": 2.986372751554313e-05,
      "loss": 1.8349,
      "step": 22050
    },
    {
      "epoch": 0.04296830880319677,
      "grad_norm": 2.827427387237549,
      "learning_rate": 2.986354232945776e-05,
      "loss": 1.8647,
      "step": 22065
    },
    {
      "epoch": 0.04299751907430704,
      "grad_norm": 2.4225332736968994,
      "learning_rate": 2.986335701820441e-05,
      "loss": 2.0951,
      "step": 22080
    },
    {
      "epoch": 0.0430267293454173,
      "grad_norm": 2.9156410694122314,
      "learning_rate": 2.986317158178464e-05,
      "loss": 1.7989,
      "step": 22095
    },
    {
      "epoch": 0.04305593961652756,
      "grad_norm": 2.3038766384124756,
      "learning_rate": 2.986298602020001e-05,
      "loss": 1.8867,
      "step": 22110
    },
    {
      "epoch": 0.04308514988763782,
      "grad_norm": 3.085129976272583,
      "learning_rate": 2.986280033345209e-05,
      "loss": 1.6969,
      "step": 22125
    },
    {
      "epoch": 0.04311436015874809,
      "grad_norm": 4.071490287780762,
      "learning_rate": 2.986261452154243e-05,
      "loss": 1.7865,
      "step": 22140
    },
    {
      "epoch": 0.04314357042985835,
      "grad_norm": 2.2542295455932617,
      "learning_rate": 2.986242858447261e-05,
      "loss": 1.8687,
      "step": 22155
    },
    {
      "epoch": 0.04317278070096861,
      "grad_norm": 3.3869330883026123,
      "learning_rate": 2.9862242522244183e-05,
      "loss": 1.8037,
      "step": 22170
    },
    {
      "epoch": 0.04320199097207888,
      "grad_norm": 3.7993576526641846,
      "learning_rate": 2.9862056334858727e-05,
      "loss": 1.8819,
      "step": 22185
    },
    {
      "epoch": 0.04323120124318914,
      "grad_norm": 3.019289255142212,
      "learning_rate": 2.9861870022317798e-05,
      "loss": 1.8932,
      "step": 22200
    },
    {
      "epoch": 0.0432604115142994,
      "grad_norm": 2.831664562225342,
      "learning_rate": 2.9861683584622976e-05,
      "loss": 1.8813,
      "step": 22215
    },
    {
      "epoch": 0.04328962178540966,
      "grad_norm": 4.9506754875183105,
      "learning_rate": 2.9861497021775825e-05,
      "loss": 1.7917,
      "step": 22230
    },
    {
      "epoch": 0.04331883205651993,
      "grad_norm": 2.384033203125,
      "learning_rate": 2.986131033377792e-05,
      "loss": 2.0226,
      "step": 22245
    },
    {
      "epoch": 0.04334804232763019,
      "grad_norm": 1.8661621809005737,
      "learning_rate": 2.9861123520630828e-05,
      "loss": 1.817,
      "step": 22260
    },
    {
      "epoch": 0.04337725259874045,
      "grad_norm": 3.5283803939819336,
      "learning_rate": 2.9860936582336123e-05,
      "loss": 1.806,
      "step": 22275
    },
    {
      "epoch": 0.04340646286985072,
      "grad_norm": 4.240048408508301,
      "learning_rate": 2.9860749518895386e-05,
      "loss": 1.7773,
      "step": 22290
    },
    {
      "epoch": 0.04343567314096098,
      "grad_norm": 3.9326212406158447,
      "learning_rate": 2.986056233031018e-05,
      "loss": 1.8494,
      "step": 22305
    },
    {
      "epoch": 0.04346488341207124,
      "grad_norm": 4.390437602996826,
      "learning_rate": 2.986037501658209e-05,
      "loss": 1.7608,
      "step": 22320
    },
    {
      "epoch": 0.0434940936831815,
      "grad_norm": 2.5305089950561523,
      "learning_rate": 2.98601875777127e-05,
      "loss": 1.7295,
      "step": 22335
    },
    {
      "epoch": 0.04352330395429177,
      "grad_norm": 2.4276509284973145,
      "learning_rate": 2.9860000013703576e-05,
      "loss": 2.0082,
      "step": 22350
    },
    {
      "epoch": 0.04355251422540203,
      "grad_norm": 3.4239320755004883,
      "learning_rate": 2.9859812324556298e-05,
      "loss": 1.6979,
      "step": 22365
    },
    {
      "epoch": 0.04358172449651229,
      "grad_norm": 2.544837474822998,
      "learning_rate": 2.985962451027245e-05,
      "loss": 1.7844,
      "step": 22380
    },
    {
      "epoch": 0.04361093476762256,
      "grad_norm": 5.094317436218262,
      "learning_rate": 2.985943657085362e-05,
      "loss": 1.8792,
      "step": 22395
    },
    {
      "epoch": 0.04364014503873282,
      "grad_norm": 3.235121250152588,
      "learning_rate": 2.985924850630138e-05,
      "loss": 1.7795,
      "step": 22410
    },
    {
      "epoch": 0.04366935530984308,
      "grad_norm": 4.88664436340332,
      "learning_rate": 2.9859060316617325e-05,
      "loss": 1.9035,
      "step": 22425
    },
    {
      "epoch": 0.04369856558095334,
      "grad_norm": 2.401301622390747,
      "learning_rate": 2.9858872001803025e-05,
      "loss": 1.9182,
      "step": 22440
    },
    {
      "epoch": 0.04372777585206361,
      "grad_norm": 2.24088978767395,
      "learning_rate": 2.9858683561860077e-05,
      "loss": 1.688,
      "step": 22455
    },
    {
      "epoch": 0.04375698612317387,
      "grad_norm": 2.521176338195801,
      "learning_rate": 2.9858494996790065e-05,
      "loss": 1.962,
      "step": 22470
    },
    {
      "epoch": 0.04378619639428413,
      "grad_norm": 2.897905111312866,
      "learning_rate": 2.9858306306594578e-05,
      "loss": 1.7388,
      "step": 22485
    },
    {
      "epoch": 0.0438154066653944,
      "grad_norm": 2.3301055431365967,
      "learning_rate": 2.9858117491275204e-05,
      "loss": 1.8834,
      "step": 22500
    },
    {
      "epoch": 0.04384461693650466,
      "grad_norm": 2.078205108642578,
      "learning_rate": 2.9857928550833533e-05,
      "loss": 1.8959,
      "step": 22515
    },
    {
      "epoch": 0.04387382720761492,
      "grad_norm": 2.8938355445861816,
      "learning_rate": 2.9857739485271153e-05,
      "loss": 1.8529,
      "step": 22530
    },
    {
      "epoch": 0.04390303747872518,
      "grad_norm": 2.0363752841949463,
      "learning_rate": 2.9857550294589663e-05,
      "loss": 1.9974,
      "step": 22545
    },
    {
      "epoch": 0.04393224774983545,
      "grad_norm": 3.58372163772583,
      "learning_rate": 2.9857360978790647e-05,
      "loss": 1.7921,
      "step": 22560
    },
    {
      "epoch": 0.04396145802094571,
      "grad_norm": 3.4086432456970215,
      "learning_rate": 2.985717153787571e-05,
      "loss": 1.8401,
      "step": 22575
    },
    {
      "epoch": 0.04399066829205597,
      "grad_norm": 2.3264150619506836,
      "learning_rate": 2.985698197184644e-05,
      "loss": 1.8789,
      "step": 22590
    },
    {
      "epoch": 0.044019878563166234,
      "grad_norm": 2.727571964263916,
      "learning_rate": 2.9856792280704435e-05,
      "loss": 1.7859,
      "step": 22605
    },
    {
      "epoch": 0.0440490888342765,
      "grad_norm": 3.833146095275879,
      "learning_rate": 2.9856602464451293e-05,
      "loss": 1.9976,
      "step": 22620
    },
    {
      "epoch": 0.04407829910538676,
      "grad_norm": 1.8543720245361328,
      "learning_rate": 2.9856412523088612e-05,
      "loss": 1.8875,
      "step": 22635
    },
    {
      "epoch": 0.044107509376497024,
      "grad_norm": 2.6003103256225586,
      "learning_rate": 2.9856222456617993e-05,
      "loss": 2.0006,
      "step": 22650
    },
    {
      "epoch": 0.04413671964760729,
      "grad_norm": 3.0136611461639404,
      "learning_rate": 2.9856032265041035e-05,
      "loss": 1.8309,
      "step": 22665
    },
    {
      "epoch": 0.04416592991871755,
      "grad_norm": 1.948042631149292,
      "learning_rate": 2.9855841948359337e-05,
      "loss": 1.8721,
      "step": 22680
    },
    {
      "epoch": 0.044195140189827814,
      "grad_norm": 3.1908979415893555,
      "learning_rate": 2.9855651506574507e-05,
      "loss": 1.8373,
      "step": 22695
    },
    {
      "epoch": 0.044224350460938075,
      "grad_norm": 3.519826889038086,
      "learning_rate": 2.985546093968815e-05,
      "loss": 1.8992,
      "step": 22710
    },
    {
      "epoch": 0.04425356073204834,
      "grad_norm": 4.259973526000977,
      "learning_rate": 2.985527024770186e-05,
      "loss": 1.9466,
      "step": 22725
    },
    {
      "epoch": 0.044282771003158604,
      "grad_norm": 2.69942569732666,
      "learning_rate": 2.9855079430617253e-05,
      "loss": 1.6805,
      "step": 22740
    },
    {
      "epoch": 0.044311981274268865,
      "grad_norm": 3.0210318565368652,
      "learning_rate": 2.9854888488435933e-05,
      "loss": 1.8744,
      "step": 22755
    },
    {
      "epoch": 0.04434119154537913,
      "grad_norm": 2.448091745376587,
      "learning_rate": 2.9854697421159505e-05,
      "loss": 1.8449,
      "step": 22770
    },
    {
      "epoch": 0.044370401816489394,
      "grad_norm": 3.8297231197357178,
      "learning_rate": 2.9854506228789586e-05,
      "loss": 1.8071,
      "step": 22785
    },
    {
      "epoch": 0.044399612087599655,
      "grad_norm": 4.065975189208984,
      "learning_rate": 2.9854314911327777e-05,
      "loss": 1.9335,
      "step": 22800
    },
    {
      "epoch": 0.044428822358709916,
      "grad_norm": 2.0296108722686768,
      "learning_rate": 2.9854123468775693e-05,
      "loss": 1.748,
      "step": 22815
    },
    {
      "epoch": 0.044458032629820184,
      "grad_norm": 1.7504347562789917,
      "learning_rate": 2.985393190113495e-05,
      "loss": 1.9084,
      "step": 22830
    },
    {
      "epoch": 0.044487242900930445,
      "grad_norm": 2.8802502155303955,
      "learning_rate": 2.9853740208407152e-05,
      "loss": 1.8649,
      "step": 22845
    },
    {
      "epoch": 0.044516453172040706,
      "grad_norm": 3.5376250743865967,
      "learning_rate": 2.985354839059392e-05,
      "loss": 1.7617,
      "step": 22860
    },
    {
      "epoch": 0.044545663443150973,
      "grad_norm": 4.846216201782227,
      "learning_rate": 2.985335644769687e-05,
      "loss": 1.7531,
      "step": 22875
    },
    {
      "epoch": 0.044574873714261234,
      "grad_norm": 1.9456549882888794,
      "learning_rate": 2.9853164379717615e-05,
      "loss": 1.9782,
      "step": 22890
    },
    {
      "epoch": 0.044604083985371495,
      "grad_norm": 3.7463254928588867,
      "learning_rate": 2.9852972186657774e-05,
      "loss": 1.9347,
      "step": 22905
    },
    {
      "epoch": 0.044633294256481756,
      "grad_norm": 1.9641201496124268,
      "learning_rate": 2.9852779868518967e-05,
      "loss": 1.8366,
      "step": 22920
    },
    {
      "epoch": 0.044662504527592024,
      "grad_norm": 3.9136605262756348,
      "learning_rate": 2.9852587425302812e-05,
      "loss": 2.0271,
      "step": 22935
    },
    {
      "epoch": 0.044691714798702285,
      "grad_norm": 2.161766767501831,
      "learning_rate": 2.9852394857010923e-05,
      "loss": 1.8681,
      "step": 22950
    },
    {
      "epoch": 0.044720925069812546,
      "grad_norm": 3.1569862365722656,
      "learning_rate": 2.9852202163644937e-05,
      "loss": 1.7996,
      "step": 22965
    },
    {
      "epoch": 0.044750135340922814,
      "grad_norm": 2.0586421489715576,
      "learning_rate": 2.9852009345206458e-05,
      "loss": 1.7727,
      "step": 22980
    },
    {
      "epoch": 0.044779345612033075,
      "grad_norm": 5.425686359405518,
      "learning_rate": 2.9851816401697127e-05,
      "loss": 1.9209,
      "step": 22995
    },
    {
      "epoch": 0.044808555883143336,
      "grad_norm": 5.385043621063232,
      "learning_rate": 2.985162333311856e-05,
      "loss": 1.8473,
      "step": 23010
    },
    {
      "epoch": 0.0448377661542536,
      "grad_norm": 3.4577736854553223,
      "learning_rate": 2.985143013947238e-05,
      "loss": 1.848,
      "step": 23025
    },
    {
      "epoch": 0.044866976425363865,
      "grad_norm": 3.0630152225494385,
      "learning_rate": 2.985123682076022e-05,
      "loss": 1.7272,
      "step": 23040
    },
    {
      "epoch": 0.044896186696474126,
      "grad_norm": 3.8338427543640137,
      "learning_rate": 2.985104337698371e-05,
      "loss": 1.9042,
      "step": 23055
    },
    {
      "epoch": 0.04492539696758439,
      "grad_norm": 3.1702969074249268,
      "learning_rate": 2.985084980814447e-05,
      "loss": 1.9389,
      "step": 23070
    },
    {
      "epoch": 0.04495460723869465,
      "grad_norm": 2.758162021636963,
      "learning_rate": 2.985065611424414e-05,
      "loss": 1.8825,
      "step": 23085
    },
    {
      "epoch": 0.044983817509804916,
      "grad_norm": 2.1373486518859863,
      "learning_rate": 2.985046229528434e-05,
      "loss": 1.9451,
      "step": 23100
    },
    {
      "epoch": 0.04501302778091518,
      "grad_norm": 2.1302478313446045,
      "learning_rate": 2.985026835126671e-05,
      "loss": 2.0208,
      "step": 23115
    },
    {
      "epoch": 0.04504223805202544,
      "grad_norm": 3.8777012825012207,
      "learning_rate": 2.985007428219289e-05,
      "loss": 2.0138,
      "step": 23130
    },
    {
      "epoch": 0.045071448323135706,
      "grad_norm": 3.428769111633301,
      "learning_rate": 2.9849880088064497e-05,
      "loss": 1.9868,
      "step": 23145
    },
    {
      "epoch": 0.04510065859424597,
      "grad_norm": 5.036014556884766,
      "learning_rate": 2.9849685768883172e-05,
      "loss": 1.7348,
      "step": 23160
    },
    {
      "epoch": 0.04512986886535623,
      "grad_norm": 3.206932306289673,
      "learning_rate": 2.9849491324650563e-05,
      "loss": 1.9991,
      "step": 23175
    },
    {
      "epoch": 0.04515907913646649,
      "grad_norm": 2.5311954021453857,
      "learning_rate": 2.9849296755368297e-05,
      "loss": 1.7501,
      "step": 23190
    },
    {
      "epoch": 0.04518828940757676,
      "grad_norm": 2.1587252616882324,
      "learning_rate": 2.984910206103801e-05,
      "loss": 1.7981,
      "step": 23205
    },
    {
      "epoch": 0.04521749967868702,
      "grad_norm": 2.198859453201294,
      "learning_rate": 2.984890724166135e-05,
      "loss": 1.9349,
      "step": 23220
    },
    {
      "epoch": 0.04524670994979728,
      "grad_norm": 3.513099193572998,
      "learning_rate": 2.9848712297239955e-05,
      "loss": 1.7639,
      "step": 23235
    },
    {
      "epoch": 0.045275920220907546,
      "grad_norm": 4.66519021987915,
      "learning_rate": 2.984851722777546e-05,
      "loss": 1.6865,
      "step": 23250
    },
    {
      "epoch": 0.04530513049201781,
      "grad_norm": 2.0970771312713623,
      "learning_rate": 2.9848322033269523e-05,
      "loss": 1.8861,
      "step": 23265
    },
    {
      "epoch": 0.04533434076312807,
      "grad_norm": 4.110560417175293,
      "learning_rate": 2.984812671372377e-05,
      "loss": 1.8482,
      "step": 23280
    },
    {
      "epoch": 0.04536355103423833,
      "grad_norm": 3.2896182537078857,
      "learning_rate": 2.9847931269139854e-05,
      "loss": 1.9021,
      "step": 23295
    },
    {
      "epoch": 0.0453927613053486,
      "grad_norm": 1.929777979850769,
      "learning_rate": 2.9847735699519423e-05,
      "loss": 1.9236,
      "step": 23310
    },
    {
      "epoch": 0.04542197157645886,
      "grad_norm": 2.989177703857422,
      "learning_rate": 2.9847540004864115e-05,
      "loss": 1.8256,
      "step": 23325
    },
    {
      "epoch": 0.04545118184756912,
      "grad_norm": 2.735318899154663,
      "learning_rate": 2.984734418517559e-05,
      "loss": 1.8279,
      "step": 23340
    },
    {
      "epoch": 0.04548039211867939,
      "grad_norm": 2.597491502761841,
      "learning_rate": 2.9847148240455495e-05,
      "loss": 1.8952,
      "step": 23355
    },
    {
      "epoch": 0.04550960238978965,
      "grad_norm": 3.5635156631469727,
      "learning_rate": 2.9846952170705473e-05,
      "loss": 1.7302,
      "step": 23370
    },
    {
      "epoch": 0.04553881266089991,
      "grad_norm": 6.853522777557373,
      "learning_rate": 2.9846755975927174e-05,
      "loss": 2.0572,
      "step": 23385
    },
    {
      "epoch": 0.04556802293201017,
      "grad_norm": 2.006582498550415,
      "learning_rate": 2.984655965612226e-05,
      "loss": 1.939,
      "step": 23400
    },
    {
      "epoch": 0.04559723320312044,
      "grad_norm": 3.6737654209136963,
      "learning_rate": 2.984636321129238e-05,
      "loss": 1.8714,
      "step": 23415
    },
    {
      "epoch": 0.0456264434742307,
      "grad_norm": 2.5802881717681885,
      "learning_rate": 2.9846166641439185e-05,
      "loss": 1.8687,
      "step": 23430
    },
    {
      "epoch": 0.04565565374534096,
      "grad_norm": 3.748178005218506,
      "learning_rate": 2.984596994656433e-05,
      "loss": 1.9146,
      "step": 23445
    },
    {
      "epoch": 0.04568486401645123,
      "grad_norm": 2.5193021297454834,
      "learning_rate": 2.9845773126669475e-05,
      "loss": 1.8845,
      "step": 23460
    },
    {
      "epoch": 0.04571407428756149,
      "grad_norm": 2.3217403888702393,
      "learning_rate": 2.984557618175628e-05,
      "loss": 1.9149,
      "step": 23475
    },
    {
      "epoch": 0.04574328455867175,
      "grad_norm": 3.9094045162200928,
      "learning_rate": 2.98453791118264e-05,
      "loss": 1.8615,
      "step": 23490
    },
    {
      "epoch": 0.04577249482978201,
      "grad_norm": 2.190436840057373,
      "learning_rate": 2.9845181916881495e-05,
      "loss": 1.8181,
      "step": 23505
    },
    {
      "epoch": 0.04580170510089228,
      "grad_norm": 3.33785343170166,
      "learning_rate": 2.984498459692322e-05,
      "loss": 1.9345,
      "step": 23520
    },
    {
      "epoch": 0.04583091537200254,
      "grad_norm": 4.065570831298828,
      "learning_rate": 2.9844787151953242e-05,
      "loss": 1.7801,
      "step": 23535
    },
    {
      "epoch": 0.0458601256431128,
      "grad_norm": 3.4585118293762207,
      "learning_rate": 2.984458958197323e-05,
      "loss": 1.8181,
      "step": 23550
    },
    {
      "epoch": 0.04588933591422307,
      "grad_norm": 2.1615355014801025,
      "learning_rate": 2.984439188698484e-05,
      "loss": 1.8265,
      "step": 23565
    },
    {
      "epoch": 0.04591854618533333,
      "grad_norm": 3.9168038368225098,
      "learning_rate": 2.9844194066989737e-05,
      "loss": 1.8508,
      "step": 23580
    },
    {
      "epoch": 0.04594775645644359,
      "grad_norm": 2.436638593673706,
      "learning_rate": 2.9843996121989587e-05,
      "loss": 1.9247,
      "step": 23595
    },
    {
      "epoch": 0.04597696672755385,
      "grad_norm": 3.637157678604126,
      "learning_rate": 2.9843798051986053e-05,
      "loss": 1.9791,
      "step": 23610
    },
    {
      "epoch": 0.04600617699866412,
      "grad_norm": 2.9777755737304688,
      "learning_rate": 2.9843599856980815e-05,
      "loss": 1.8227,
      "step": 23625
    },
    {
      "epoch": 0.04603538726977438,
      "grad_norm": 5.862057209014893,
      "learning_rate": 2.9843401536975533e-05,
      "loss": 1.9745,
      "step": 23640
    },
    {
      "epoch": 0.04606459754088464,
      "grad_norm": 2.937347173690796,
      "learning_rate": 2.9843203091971878e-05,
      "loss": 1.7575,
      "step": 23655
    },
    {
      "epoch": 0.0460938078119949,
      "grad_norm": 3.7695438861846924,
      "learning_rate": 2.984300452197152e-05,
      "loss": 1.9655,
      "step": 23670
    },
    {
      "epoch": 0.04612301808310517,
      "grad_norm": 3.1122677326202393,
      "learning_rate": 2.9842805826976137e-05,
      "loss": 2.0459,
      "step": 23685
    },
    {
      "epoch": 0.04615222835421543,
      "grad_norm": 4.503921031951904,
      "learning_rate": 2.98426070069874e-05,
      "loss": 1.8319,
      "step": 23700
    },
    {
      "epoch": 0.04618143862532569,
      "grad_norm": 2.247570514678955,
      "learning_rate": 2.9842408062006982e-05,
      "loss": 1.765,
      "step": 23715
    },
    {
      "epoch": 0.04621064889643596,
      "grad_norm": 3.512427806854248,
      "learning_rate": 2.9842208992036554e-05,
      "loss": 1.8118,
      "step": 23730
    },
    {
      "epoch": 0.04623985916754622,
      "grad_norm": 3.174893379211426,
      "learning_rate": 2.98420097970778e-05,
      "loss": 1.8791,
      "step": 23745
    },
    {
      "epoch": 0.04626906943865648,
      "grad_norm": 5.151320934295654,
      "learning_rate": 2.9841810477132392e-05,
      "loss": 1.817,
      "step": 23760
    },
    {
      "epoch": 0.04629827970976674,
      "grad_norm": 2.6584534645080566,
      "learning_rate": 2.984161103220201e-05,
      "loss": 1.8441,
      "step": 23775
    },
    {
      "epoch": 0.04632748998087701,
      "grad_norm": 1.967596173286438,
      "learning_rate": 2.9841411462288335e-05,
      "loss": 2.0349,
      "step": 23790
    },
    {
      "epoch": 0.04635670025198727,
      "grad_norm": 3.7256250381469727,
      "learning_rate": 2.9841211767393048e-05,
      "loss": 1.8131,
      "step": 23805
    },
    {
      "epoch": 0.04638591052309753,
      "grad_norm": 4.081748962402344,
      "learning_rate": 2.9841011947517826e-05,
      "loss": 1.8664,
      "step": 23820
    },
    {
      "epoch": 0.0464151207942078,
      "grad_norm": 2.527451276779175,
      "learning_rate": 2.984081200266436e-05,
      "loss": 1.8916,
      "step": 23835
    },
    {
      "epoch": 0.04644433106531806,
      "grad_norm": 2.159346103668213,
      "learning_rate": 2.9840611932834326e-05,
      "loss": 1.9254,
      "step": 23850
    },
    {
      "epoch": 0.04647354133642832,
      "grad_norm": 2.636519432067871,
      "learning_rate": 2.9840411738029412e-05,
      "loss": 1.8062,
      "step": 23865
    },
    {
      "epoch": 0.046502751607538584,
      "grad_norm": 2.3750388622283936,
      "learning_rate": 2.9840211418251303e-05,
      "loss": 1.7347,
      "step": 23880
    },
    {
      "epoch": 0.04653196187864885,
      "grad_norm": 2.181845188140869,
      "learning_rate": 2.9840010973501685e-05,
      "loss": 1.9207,
      "step": 23895
    },
    {
      "epoch": 0.04656117214975911,
      "grad_norm": 3.098555564880371,
      "learning_rate": 2.9839810403782252e-05,
      "loss": 1.9566,
      "step": 23910
    },
    {
      "epoch": 0.046590382420869374,
      "grad_norm": 5.142632484436035,
      "learning_rate": 2.9839609709094685e-05,
      "loss": 1.8871,
      "step": 23925
    },
    {
      "epoch": 0.04661959269197964,
      "grad_norm": 1.7162665128707886,
      "learning_rate": 2.9839408889440678e-05,
      "loss": 1.8784,
      "step": 23940
    },
    {
      "epoch": 0.0466488029630899,
      "grad_norm": 3.743401050567627,
      "learning_rate": 2.9839207944821925e-05,
      "loss": 1.9179,
      "step": 23955
    },
    {
      "epoch": 0.046678013234200164,
      "grad_norm": 2.952817440032959,
      "learning_rate": 2.983900687524011e-05,
      "loss": 1.9182,
      "step": 23970
    },
    {
      "epoch": 0.046707223505310425,
      "grad_norm": 2.3484249114990234,
      "learning_rate": 2.983880568069693e-05,
      "loss": 1.8336,
      "step": 23985
    },
    {
      "epoch": 0.04673643377642069,
      "grad_norm": 1.931921362876892,
      "learning_rate": 2.9838604361194087e-05,
      "loss": 1.771,
      "step": 24000
    },
    {
      "epoch": 0.04676564404753095,
      "grad_norm": 5.197054386138916,
      "learning_rate": 2.9838402916733263e-05,
      "loss": 1.8558,
      "step": 24015
    },
    {
      "epoch": 0.046794854318641214,
      "grad_norm": 2.862427234649658,
      "learning_rate": 2.9838201347316164e-05,
      "loss": 1.6956,
      "step": 24030
    },
    {
      "epoch": 0.04682406458975148,
      "grad_norm": 2.6003293991088867,
      "learning_rate": 2.9837999652944487e-05,
      "loss": 2.0473,
      "step": 24045
    },
    {
      "epoch": 0.04685327486086174,
      "grad_norm": 2.068455934524536,
      "learning_rate": 2.9837797833619926e-05,
      "loss": 1.9233,
      "step": 24060
    },
    {
      "epoch": 0.046882485131972004,
      "grad_norm": 3.4722557067871094,
      "learning_rate": 2.983759588934418e-05,
      "loss": 2.0052,
      "step": 24075
    },
    {
      "epoch": 0.046911695403082265,
      "grad_norm": 3.198732852935791,
      "learning_rate": 2.9837393820118954e-05,
      "loss": 1.906,
      "step": 24090
    },
    {
      "epoch": 0.04694090567419253,
      "grad_norm": 2.5050504207611084,
      "learning_rate": 2.983719162594595e-05,
      "loss": 1.833,
      "step": 24105
    },
    {
      "epoch": 0.046970115945302794,
      "grad_norm": 2.1438305377960205,
      "learning_rate": 2.9836989306826866e-05,
      "loss": 1.7439,
      "step": 24120
    },
    {
      "epoch": 0.046999326216413055,
      "grad_norm": 1.5980552434921265,
      "learning_rate": 2.983678686276341e-05,
      "loss": 1.8018,
      "step": 24135
    },
    {
      "epoch": 0.047028536487523316,
      "grad_norm": 2.566882610321045,
      "learning_rate": 2.9836584293757282e-05,
      "loss": 1.8807,
      "step": 24150
    },
    {
      "epoch": 0.047057746758633584,
      "grad_norm": 4.125818252563477,
      "learning_rate": 2.9836381599810196e-05,
      "loss": 1.8198,
      "step": 24165
    },
    {
      "epoch": 0.047086957029743845,
      "grad_norm": 3.1217780113220215,
      "learning_rate": 2.983617878092385e-05,
      "loss": 2.0202,
      "step": 24180
    },
    {
      "epoch": 0.047116167300854106,
      "grad_norm": 3.001835823059082,
      "learning_rate": 2.9835975837099956e-05,
      "loss": 1.851,
      "step": 24195
    },
    {
      "epoch": 0.047145377571964374,
      "grad_norm": 4.060051918029785,
      "learning_rate": 2.9835772768340225e-05,
      "loss": 1.8918,
      "step": 24210
    },
    {
      "epoch": 0.047174587843074635,
      "grad_norm": 4.199372291564941,
      "learning_rate": 2.9835569574646363e-05,
      "loss": 1.848,
      "step": 24225
    },
    {
      "epoch": 0.047203798114184896,
      "grad_norm": 3.1068427562713623,
      "learning_rate": 2.9835366256020085e-05,
      "loss": 1.793,
      "step": 24240
    },
    {
      "epoch": 0.04723300838529516,
      "grad_norm": 3.30556321144104,
      "learning_rate": 2.9835162812463098e-05,
      "loss": 1.8914,
      "step": 24255
    },
    {
      "epoch": 0.047262218656405425,
      "grad_norm": 3.898911476135254,
      "learning_rate": 2.9834959243977123e-05,
      "loss": 1.929,
      "step": 24270
    },
    {
      "epoch": 0.047291428927515686,
      "grad_norm": 2.1169817447662354,
      "learning_rate": 2.9834755550563865e-05,
      "loss": 1.9513,
      "step": 24285
    },
    {
      "epoch": 0.04732063919862595,
      "grad_norm": 3.0540499687194824,
      "learning_rate": 2.9834551732225044e-05,
      "loss": 1.8283,
      "step": 24300
    },
    {
      "epoch": 0.047349849469736215,
      "grad_norm": 2.495234727859497,
      "learning_rate": 2.9834347788962383e-05,
      "loss": 1.8268,
      "step": 24315
    },
    {
      "epoch": 0.047379059740846476,
      "grad_norm": 4.254831790924072,
      "learning_rate": 2.9834143720777588e-05,
      "loss": 1.8933,
      "step": 24330
    },
    {
      "epoch": 0.04740827001195674,
      "grad_norm": 5.430483818054199,
      "learning_rate": 2.9833939527672384e-05,
      "loss": 1.8128,
      "step": 24345
    },
    {
      "epoch": 0.047437480283067,
      "grad_norm": 3.332515239715576,
      "learning_rate": 2.983373520964849e-05,
      "loss": 1.9089,
      "step": 24360
    },
    {
      "epoch": 0.047466690554177265,
      "grad_norm": 2.1588125228881836,
      "learning_rate": 2.983353076670762e-05,
      "loss": 1.7578,
      "step": 24375
    },
    {
      "epoch": 0.047495900825287526,
      "grad_norm": 2.6773619651794434,
      "learning_rate": 2.9833326198851503e-05,
      "loss": 1.9486,
      "step": 24390
    },
    {
      "epoch": 0.04752511109639779,
      "grad_norm": 2.855078935623169,
      "learning_rate": 2.9833121506081862e-05,
      "loss": 1.9339,
      "step": 24405
    },
    {
      "epoch": 0.047554321367508055,
      "grad_norm": 2.539210557937622,
      "learning_rate": 2.983291668840042e-05,
      "loss": 1.8479,
      "step": 24420
    },
    {
      "epoch": 0.047583531638618316,
      "grad_norm": 3.0605132579803467,
      "learning_rate": 2.9832711745808895e-05,
      "loss": 1.8924,
      "step": 24435
    },
    {
      "epoch": 0.04761274190972858,
      "grad_norm": 2.142998218536377,
      "learning_rate": 2.9832506678309025e-05,
      "loss": 2.0331,
      "step": 24450
    },
    {
      "epoch": 0.04764195218083884,
      "grad_norm": 3.082610845565796,
      "learning_rate": 2.983230148590253e-05,
      "loss": 1.8299,
      "step": 24465
    },
    {
      "epoch": 0.047671162451949106,
      "grad_norm": 4.011190414428711,
      "learning_rate": 2.9832096168591128e-05,
      "loss": 1.8591,
      "step": 24480
    },
    {
      "epoch": 0.04770037272305937,
      "grad_norm": 2.326756238937378,
      "learning_rate": 2.983189072637657e-05,
      "loss": 1.8687,
      "step": 24495
    },
    {
      "epoch": 0.04772958299416963,
      "grad_norm": 2.1641993522644043,
      "learning_rate": 2.9831685159260568e-05,
      "loss": 1.9402,
      "step": 24510
    },
    {
      "epoch": 0.047758793265279896,
      "grad_norm": 2.614429473876953,
      "learning_rate": 2.983147946724486e-05,
      "loss": 1.8911,
      "step": 24525
    },
    {
      "epoch": 0.04778800353639016,
      "grad_norm": 2.163760185241699,
      "learning_rate": 2.983127365033118e-05,
      "loss": 1.7623,
      "step": 24540
    },
    {
      "epoch": 0.04781721380750042,
      "grad_norm": 1.837815761566162,
      "learning_rate": 2.9831067708521257e-05,
      "loss": 2.0103,
      "step": 24555
    },
    {
      "epoch": 0.04784642407861068,
      "grad_norm": 4.629458904266357,
      "learning_rate": 2.9830861641816826e-05,
      "loss": 1.8473,
      "step": 24570
    },
    {
      "epoch": 0.04787563434972095,
      "grad_norm": 2.741942882537842,
      "learning_rate": 2.9830655450219623e-05,
      "loss": 1.8219,
      "step": 24585
    },
    {
      "epoch": 0.04790484462083121,
      "grad_norm": 2.6273906230926514,
      "learning_rate": 2.9830449133731387e-05,
      "loss": 1.7688,
      "step": 24600
    },
    {
      "epoch": 0.04793405489194147,
      "grad_norm": 2.2749216556549072,
      "learning_rate": 2.983024269235385e-05,
      "loss": 1.9809,
      "step": 24615
    },
    {
      "epoch": 0.04796326516305174,
      "grad_norm": 4.1970367431640625,
      "learning_rate": 2.9830036126088754e-05,
      "loss": 1.7998,
      "step": 24630
    },
    {
      "epoch": 0.047992475434162,
      "grad_norm": 2.87906813621521,
      "learning_rate": 2.982982943493784e-05,
      "loss": 1.8496,
      "step": 24645
    },
    {
      "epoch": 0.04802168570527226,
      "grad_norm": 1.9954204559326172,
      "learning_rate": 2.9829622618902848e-05,
      "loss": 1.8782,
      "step": 24660
    },
    {
      "epoch": 0.04805089597638252,
      "grad_norm": 3.3936777114868164,
      "learning_rate": 2.982941567798551e-05,
      "loss": 1.7,
      "step": 24675
    },
    {
      "epoch": 0.04808010624749279,
      "grad_norm": 5.30580997467041,
      "learning_rate": 2.9829208612187585e-05,
      "loss": 1.8712,
      "step": 24690
    },
    {
      "epoch": 0.04810931651860305,
      "grad_norm": 1.797080636024475,
      "learning_rate": 2.982900142151081e-05,
      "loss": 1.8711,
      "step": 24705
    },
    {
      "epoch": 0.04813852678971331,
      "grad_norm": 1.7338413000106812,
      "learning_rate": 2.9828794105956922e-05,
      "loss": 1.9125,
      "step": 24720
    },
    {
      "epoch": 0.04816773706082357,
      "grad_norm": 6.833633899688721,
      "learning_rate": 2.9828586665527677e-05,
      "loss": 1.8622,
      "step": 24735
    },
    {
      "epoch": 0.04819694733193384,
      "grad_norm": 2.8855645656585693,
      "learning_rate": 2.9828379100224814e-05,
      "loss": 1.7596,
      "step": 24750
    },
    {
      "epoch": 0.0482261576030441,
      "grad_norm": 3.7425286769866943,
      "learning_rate": 2.982817141005009e-05,
      "loss": 1.8452,
      "step": 24765
    },
    {
      "epoch": 0.04825536787415436,
      "grad_norm": 4.187405109405518,
      "learning_rate": 2.9827963595005248e-05,
      "loss": 1.8669,
      "step": 24780
    },
    {
      "epoch": 0.04828457814526463,
      "grad_norm": 4.476016998291016,
      "learning_rate": 2.982775565509204e-05,
      "loss": 1.9221,
      "step": 24795
    },
    {
      "epoch": 0.04831378841637489,
      "grad_norm": 3.6991500854492188,
      "learning_rate": 2.9827547590312213e-05,
      "loss": 1.5434,
      "step": 24810
    },
    {
      "epoch": 0.04834299868748515,
      "grad_norm": 3.0707576274871826,
      "learning_rate": 2.9827339400667524e-05,
      "loss": 1.88,
      "step": 24825
    },
    {
      "epoch": 0.04837220895859541,
      "grad_norm": 2.604163646697998,
      "learning_rate": 2.9827131086159723e-05,
      "loss": 1.9946,
      "step": 24840
    },
    {
      "epoch": 0.04840141922970568,
      "grad_norm": 4.7181549072265625,
      "learning_rate": 2.9826922646790568e-05,
      "loss": 1.8807,
      "step": 24855
    },
    {
      "epoch": 0.04843062950081594,
      "grad_norm": 3.983146905899048,
      "learning_rate": 2.9826714082561808e-05,
      "loss": 1.8593,
      "step": 24870
    },
    {
      "epoch": 0.0484598397719262,
      "grad_norm": 4.0342912673950195,
      "learning_rate": 2.982650539347521e-05,
      "loss": 1.9208,
      "step": 24885
    },
    {
      "epoch": 0.04848905004303647,
      "grad_norm": 3.092644691467285,
      "learning_rate": 2.982629657953252e-05,
      "loss": 1.9611,
      "step": 24900
    },
    {
      "epoch": 0.04851826031414673,
      "grad_norm": 2.244311809539795,
      "learning_rate": 2.98260876407355e-05,
      "loss": 1.7965,
      "step": 24915
    },
    {
      "epoch": 0.04854747058525699,
      "grad_norm": 4.740140914916992,
      "learning_rate": 2.9825878577085917e-05,
      "loss": 2.0418,
      "step": 24930
    },
    {
      "epoch": 0.04857668085636725,
      "grad_norm": 1.6131473779678345,
      "learning_rate": 2.9825669388585523e-05,
      "loss": 1.7781,
      "step": 24945
    },
    {
      "epoch": 0.04860589112747752,
      "grad_norm": 3.3024797439575195,
      "learning_rate": 2.9825460075236077e-05,
      "loss": 1.9176,
      "step": 24960
    },
    {
      "epoch": 0.04863510139858778,
      "grad_norm": 3.6958138942718506,
      "learning_rate": 2.9825250637039348e-05,
      "loss": 1.8613,
      "step": 24975
    },
    {
      "epoch": 0.04866431166969804,
      "grad_norm": 2.277597427368164,
      "learning_rate": 2.9825041073997102e-05,
      "loss": 1.9942,
      "step": 24990
    },
    {
      "epoch": 0.04869352194080831,
      "grad_norm": 3.3070085048675537,
      "learning_rate": 2.9824831386111103e-05,
      "loss": 1.5052,
      "step": 25005
    },
    {
      "epoch": 0.04872273221191857,
      "grad_norm": 2.769148826599121,
      "learning_rate": 2.9824621573383107e-05,
      "loss": 1.8361,
      "step": 25020
    },
    {
      "epoch": 0.04875194248302883,
      "grad_norm": 2.3901126384735107,
      "learning_rate": 2.982441163581489e-05,
      "loss": 1.8346,
      "step": 25035
    },
    {
      "epoch": 0.04878115275413909,
      "grad_norm": 3.8897323608398438,
      "learning_rate": 2.9824201573408218e-05,
      "loss": 1.8186,
      "step": 25050
    },
    {
      "epoch": 0.04881036302524936,
      "grad_norm": 1.947713017463684,
      "learning_rate": 2.982399138616486e-05,
      "loss": 1.8812,
      "step": 25065
    },
    {
      "epoch": 0.04883957329635962,
      "grad_norm": 1.6136304140090942,
      "learning_rate": 2.9823781074086582e-05,
      "loss": 2.0161,
      "step": 25080
    },
    {
      "epoch": 0.04886878356746988,
      "grad_norm": 2.6880128383636475,
      "learning_rate": 2.9823570637175166e-05,
      "loss": 1.8861,
      "step": 25095
    },
    {
      "epoch": 0.04889799383858015,
      "grad_norm": 4.511691570281982,
      "learning_rate": 2.982336007543237e-05,
      "loss": 1.8334,
      "step": 25110
    },
    {
      "epoch": 0.04892720410969041,
      "grad_norm": 2.540619134902954,
      "learning_rate": 2.9823149388859975e-05,
      "loss": 1.8554,
      "step": 25125
    },
    {
      "epoch": 0.04895641438080067,
      "grad_norm": 2.650416135787964,
      "learning_rate": 2.982293857745976e-05,
      "loss": 2.0556,
      "step": 25140
    },
    {
      "epoch": 0.04898562465191093,
      "grad_norm": 4.075965881347656,
      "learning_rate": 2.9822727641233488e-05,
      "loss": 1.7369,
      "step": 25155
    },
    {
      "epoch": 0.0490148349230212,
      "grad_norm": 4.21481466293335,
      "learning_rate": 2.9822516580182944e-05,
      "loss": 1.9624,
      "step": 25170
    },
    {
      "epoch": 0.04904404519413146,
      "grad_norm": 4.617081642150879,
      "learning_rate": 2.98223053943099e-05,
      "loss": 1.7276,
      "step": 25185
    },
    {
      "epoch": 0.04907325546524172,
      "grad_norm": 3.5099408626556396,
      "learning_rate": 2.9822094083616145e-05,
      "loss": 1.8783,
      "step": 25200
    },
    {
      "epoch": 0.049102465736351984,
      "grad_norm": 4.118253231048584,
      "learning_rate": 2.9821882648103445e-05,
      "loss": 1.7848,
      "step": 25215
    },
    {
      "epoch": 0.04913167600746225,
      "grad_norm": 3.617659091949463,
      "learning_rate": 2.982167108777359e-05,
      "loss": 1.8307,
      "step": 25230
    },
    {
      "epoch": 0.04916088627857251,
      "grad_norm": 3.3717000484466553,
      "learning_rate": 2.9821459402628357e-05,
      "loss": 1.6414,
      "step": 25245
    },
    {
      "epoch": 0.049190096549682774,
      "grad_norm": 3.3341469764709473,
      "learning_rate": 2.9821247592669526e-05,
      "loss": 1.9641,
      "step": 25260
    },
    {
      "epoch": 0.04921930682079304,
      "grad_norm": 3.8818039894104004,
      "learning_rate": 2.9821035657898886e-05,
      "loss": 1.8227,
      "step": 25275
    },
    {
      "epoch": 0.0492485170919033,
      "grad_norm": 2.4647955894470215,
      "learning_rate": 2.9820823598318226e-05,
      "loss": 1.9441,
      "step": 25290
    },
    {
      "epoch": 0.049277727363013564,
      "grad_norm": 5.145657062530518,
      "learning_rate": 2.9820611413929318e-05,
      "loss": 1.7515,
      "step": 25305
    },
    {
      "epoch": 0.049306937634123825,
      "grad_norm": 2.595554828643799,
      "learning_rate": 2.9820399104733964e-05,
      "loss": 1.8185,
      "step": 25320
    },
    {
      "epoch": 0.04933614790523409,
      "grad_norm": 3.0854387283325195,
      "learning_rate": 2.9820186670733944e-05,
      "loss": 2.0462,
      "step": 25335
    },
    {
      "epoch": 0.049365358176344354,
      "grad_norm": 4.086148262023926,
      "learning_rate": 2.9819974111931045e-05,
      "loss": 1.923,
      "step": 25350
    },
    {
      "epoch": 0.049394568447454615,
      "grad_norm": 2.3999664783477783,
      "learning_rate": 2.9819761428327057e-05,
      "loss": 1.9243,
      "step": 25365
    },
    {
      "epoch": 0.04942377871856488,
      "grad_norm": 1.8367825746536255,
      "learning_rate": 2.981954861992378e-05,
      "loss": 1.9104,
      "step": 25380
    },
    {
      "epoch": 0.049452988989675144,
      "grad_norm": 2.3233165740966797,
      "learning_rate": 2.9819335686722997e-05,
      "loss": 1.7446,
      "step": 25395
    },
    {
      "epoch": 0.049482199260785405,
      "grad_norm": 3.7498881816864014,
      "learning_rate": 2.98191226287265e-05,
      "loss": 2.0172,
      "step": 25410
    },
    {
      "epoch": 0.049511409531895666,
      "grad_norm": 2.636087656021118,
      "learning_rate": 2.9818909445936092e-05,
      "loss": 1.9958,
      "step": 25425
    },
    {
      "epoch": 0.049540619803005934,
      "grad_norm": 2.6167049407958984,
      "learning_rate": 2.9818696138353564e-05,
      "loss": 1.9746,
      "step": 25440
    },
    {
      "epoch": 0.049569830074116195,
      "grad_norm": 2.699380874633789,
      "learning_rate": 2.9818482705980708e-05,
      "loss": 2.0158,
      "step": 25455
    },
    {
      "epoch": 0.049599040345226456,
      "grad_norm": 2.8639230728149414,
      "learning_rate": 2.9818269148819326e-05,
      "loss": 1.8795,
      "step": 25470
    },
    {
      "epoch": 0.04962825061633672,
      "grad_norm": 3.6716597080230713,
      "learning_rate": 2.9818055466871217e-05,
      "loss": 2.0066,
      "step": 25485
    },
    {
      "epoch": 0.049657460887446984,
      "grad_norm": 3.0852763652801514,
      "learning_rate": 2.981784166013818e-05,
      "loss": 1.8392,
      "step": 25500
    },
    {
      "epoch": 0.049686671158557245,
      "grad_norm": 2.0497000217437744,
      "learning_rate": 2.981762772862201e-05,
      "loss": 2.0587,
      "step": 25515
    },
    {
      "epoch": 0.049715881429667506,
      "grad_norm": 4.302377700805664,
      "learning_rate": 2.9817413672324517e-05,
      "loss": 1.9415,
      "step": 25530
    },
    {
      "epoch": 0.049745091700777774,
      "grad_norm": 2.478428840637207,
      "learning_rate": 2.9817199491247495e-05,
      "loss": 2.0082,
      "step": 25545
    },
    {
      "epoch": 0.049774301971888035,
      "grad_norm": 3.375516891479492,
      "learning_rate": 2.9816985185392752e-05,
      "loss": 2.061,
      "step": 25560
    },
    {
      "epoch": 0.049803512242998296,
      "grad_norm": 2.3733456134796143,
      "learning_rate": 2.9816770754762094e-05,
      "loss": 1.8752,
      "step": 25575
    },
    {
      "epoch": 0.049832722514108564,
      "grad_norm": 2.8646862506866455,
      "learning_rate": 2.9816556199357334e-05,
      "loss": 1.9048,
      "step": 25590
    },
    {
      "epoch": 0.049861932785218825,
      "grad_norm": 3.711494207382202,
      "learning_rate": 2.981634151918026e-05,
      "loss": 1.8555,
      "step": 25605
    },
    {
      "epoch": 0.049891143056329086,
      "grad_norm": 4.652657985687256,
      "learning_rate": 2.9816126714232694e-05,
      "loss": 1.9591,
      "step": 25620
    },
    {
      "epoch": 0.04992035332743935,
      "grad_norm": 3.5854547023773193,
      "learning_rate": 2.981591178451644e-05,
      "loss": 1.8043,
      "step": 25635
    },
    {
      "epoch": 0.049949563598549615,
      "grad_norm": 2.532128095626831,
      "learning_rate": 2.981569673003331e-05,
      "loss": 1.7456,
      "step": 25650
    },
    {
      "epoch": 0.049978773869659876,
      "grad_norm": 1.8968348503112793,
      "learning_rate": 2.9815481550785116e-05,
      "loss": 1.8971,
      "step": 25665
    },
    {
      "epoch": 0.05000798414077014,
      "grad_norm": 4.662414073944092,
      "learning_rate": 2.9815266246773663e-05,
      "loss": 2.1576,
      "step": 25680
    },
    {
      "epoch": 0.050037194411880405,
      "grad_norm": 2.1178083419799805,
      "learning_rate": 2.9815050818000773e-05,
      "loss": 1.9055,
      "step": 25695
    },
    {
      "epoch": 0.050066404682990666,
      "grad_norm": 3.8748810291290283,
      "learning_rate": 2.9814835264468254e-05,
      "loss": 1.8177,
      "step": 25710
    },
    {
      "epoch": 0.05009561495410093,
      "grad_norm": 3.424405097961426,
      "learning_rate": 2.9814619586177926e-05,
      "loss": 1.8804,
      "step": 25725
    },
    {
      "epoch": 0.05012482522521119,
      "grad_norm": 4.104612827301025,
      "learning_rate": 2.98144037831316e-05,
      "loss": 1.973,
      "step": 25740
    },
    {
      "epoch": 0.050154035496321456,
      "grad_norm": 2.1848320960998535,
      "learning_rate": 2.98141878553311e-05,
      "loss": 1.8529,
      "step": 25755
    },
    {
      "epoch": 0.05018324576743172,
      "grad_norm": 4.191700458526611,
      "learning_rate": 2.981397180277824e-05,
      "loss": 1.8094,
      "step": 25770
    },
    {
      "epoch": 0.05021245603854198,
      "grad_norm": 3.3454208374023438,
      "learning_rate": 2.981375562547484e-05,
      "loss": 1.9105,
      "step": 25785
    },
    {
      "epoch": 0.05024166630965224,
      "grad_norm": 2.848618507385254,
      "learning_rate": 2.9813539323422717e-05,
      "loss": 1.9054,
      "step": 25800
    },
    {
      "epoch": 0.05027087658076251,
      "grad_norm": 1.7352811098098755,
      "learning_rate": 2.98133228966237e-05,
      "loss": 1.91,
      "step": 25815
    },
    {
      "epoch": 0.05030008685187277,
      "grad_norm": 2.177297592163086,
      "learning_rate": 2.9813106345079604e-05,
      "loss": 1.8276,
      "step": 25830
    },
    {
      "epoch": 0.05032929712298303,
      "grad_norm": 2.1822316646575928,
      "learning_rate": 2.981288966879226e-05,
      "loss": 1.7961,
      "step": 25845
    },
    {
      "epoch": 0.050358507394093296,
      "grad_norm": 2.70729398727417,
      "learning_rate": 2.9812672867763482e-05,
      "loss": 1.7497,
      "step": 25860
    },
    {
      "epoch": 0.05038771766520356,
      "grad_norm": 2.869450807571411,
      "learning_rate": 2.981245594199511e-05,
      "loss": 1.8309,
      "step": 25875
    },
    {
      "epoch": 0.05041692793631382,
      "grad_norm": 2.233219623565674,
      "learning_rate": 2.981223889148896e-05,
      "loss": 1.8577,
      "step": 25890
    },
    {
      "epoch": 0.05044613820742408,
      "grad_norm": 5.402493000030518,
      "learning_rate": 2.981202171624686e-05,
      "loss": 1.9151,
      "step": 25905
    },
    {
      "epoch": 0.05047534847853435,
      "grad_norm": 4.127412796020508,
      "learning_rate": 2.9811804416270648e-05,
      "loss": 1.8644,
      "step": 25920
    },
    {
      "epoch": 0.05050455874964461,
      "grad_norm": 2.9654555320739746,
      "learning_rate": 2.9811586991562145e-05,
      "loss": 1.99,
      "step": 25935
    },
    {
      "epoch": 0.05053376902075487,
      "grad_norm": 2.6316134929656982,
      "learning_rate": 2.981136944212318e-05,
      "loss": 1.7434,
      "step": 25950
    },
    {
      "epoch": 0.05056297929186514,
      "grad_norm": 2.6688010692596436,
      "learning_rate": 2.9811151767955597e-05,
      "loss": 1.8075,
      "step": 25965
    },
    {
      "epoch": 0.0505921895629754,
      "grad_norm": 1.7729136943817139,
      "learning_rate": 2.981093396906122e-05,
      "loss": 1.7828,
      "step": 25980
    },
    {
      "epoch": 0.05062139983408566,
      "grad_norm": 3.017512321472168,
      "learning_rate": 2.9810716045441884e-05,
      "loss": 1.9602,
      "step": 25995
    },
    {
      "epoch": 0.05065061010519592,
      "grad_norm": 3.954099416732788,
      "learning_rate": 2.9810497997099427e-05,
      "loss": 1.9158,
      "step": 26010
    },
    {
      "epoch": 0.05067982037630619,
      "grad_norm": 4.384187698364258,
      "learning_rate": 2.981027982403568e-05,
      "loss": 2.0172,
      "step": 26025
    },
    {
      "epoch": 0.05070903064741645,
      "grad_norm": 2.7591800689697266,
      "learning_rate": 2.9810061526252488e-05,
      "loss": 1.8372,
      "step": 26040
    },
    {
      "epoch": 0.05073824091852671,
      "grad_norm": 2.6615748405456543,
      "learning_rate": 2.980984310375168e-05,
      "loss": 1.8127,
      "step": 26055
    },
    {
      "epoch": 0.05076745118963698,
      "grad_norm": 3.1193888187408447,
      "learning_rate": 2.9809624556535106e-05,
      "loss": 1.8826,
      "step": 26070
    },
    {
      "epoch": 0.05079666146074724,
      "grad_norm": 2.338146448135376,
      "learning_rate": 2.9809405884604594e-05,
      "loss": 1.9003,
      "step": 26085
    },
    {
      "epoch": 0.0508258717318575,
      "grad_norm": 2.339162588119507,
      "learning_rate": 2.9809187087961993e-05,
      "loss": 1.8444,
      "step": 26100
    },
    {
      "epoch": 0.05085508200296776,
      "grad_norm": 6.189070224761963,
      "learning_rate": 2.980896816660915e-05,
      "loss": 1.8682,
      "step": 26115
    },
    {
      "epoch": 0.05088429227407803,
      "grad_norm": 3.7556257247924805,
      "learning_rate": 2.9808749120547898e-05,
      "loss": 1.8692,
      "step": 26130
    },
    {
      "epoch": 0.05091350254518829,
      "grad_norm": 2.4181950092315674,
      "learning_rate": 2.980852994978009e-05,
      "loss": 1.8497,
      "step": 26145
    },
    {
      "epoch": 0.05094271281629855,
      "grad_norm": 2.1751251220703125,
      "learning_rate": 2.9808310654307566e-05,
      "loss": 1.7236,
      "step": 26160
    },
    {
      "epoch": 0.05097192308740882,
      "grad_norm": 2.304203748703003,
      "learning_rate": 2.9808091234132177e-05,
      "loss": 1.8394,
      "step": 26175
    },
    {
      "epoch": 0.05100113335851908,
      "grad_norm": 2.03320050239563,
      "learning_rate": 2.980787168925577e-05,
      "loss": 2.0082,
      "step": 26190
    },
    {
      "epoch": 0.05103034362962934,
      "grad_norm": 3.1527299880981445,
      "learning_rate": 2.9807652019680195e-05,
      "loss": 1.8783,
      "step": 26205
    },
    {
      "epoch": 0.0510595539007396,
      "grad_norm": 3.638120412826538,
      "learning_rate": 2.9807432225407295e-05,
      "loss": 1.8912,
      "step": 26220
    },
    {
      "epoch": 0.05108876417184987,
      "grad_norm": 1.9177019596099854,
      "learning_rate": 2.9807212306438927e-05,
      "loss": 1.7949,
      "step": 26235
    },
    {
      "epoch": 0.05111797444296013,
      "grad_norm": 4.763120174407959,
      "learning_rate": 2.9806992262776945e-05,
      "loss": 1.9464,
      "step": 26250
    },
    {
      "epoch": 0.05114718471407039,
      "grad_norm": 4.4592132568359375,
      "learning_rate": 2.980677209442319e-05,
      "loss": 1.834,
      "step": 26265
    },
    {
      "epoch": 0.05117639498518065,
      "grad_norm": 2.6289279460906982,
      "learning_rate": 2.980655180137953e-05,
      "loss": 1.7644,
      "step": 26280
    },
    {
      "epoch": 0.05120560525629092,
      "grad_norm": 2.2794735431671143,
      "learning_rate": 2.9806331383647816e-05,
      "loss": 1.9125,
      "step": 26295
    },
    {
      "epoch": 0.05123481552740118,
      "grad_norm": 3.5884125232696533,
      "learning_rate": 2.9806110841229904e-05,
      "loss": 1.925,
      "step": 26310
    },
    {
      "epoch": 0.05126402579851144,
      "grad_norm": 2.8750176429748535,
      "learning_rate": 2.9805890174127648e-05,
      "loss": 1.6463,
      "step": 26325
    },
    {
      "epoch": 0.05129323606962171,
      "grad_norm": 3.1167027950286865,
      "learning_rate": 2.980566938234291e-05,
      "loss": 1.8897,
      "step": 26340
    },
    {
      "epoch": 0.05132244634073197,
      "grad_norm": 2.8106181621551514,
      "learning_rate": 2.9805448465877546e-05,
      "loss": 1.855,
      "step": 26355
    },
    {
      "epoch": 0.05135165661184223,
      "grad_norm": 3.8905303478240967,
      "learning_rate": 2.980522742473342e-05,
      "loss": 1.9127,
      "step": 26370
    },
    {
      "epoch": 0.05138086688295249,
      "grad_norm": 3.031163215637207,
      "learning_rate": 2.980500625891239e-05,
      "loss": 1.9577,
      "step": 26385
    },
    {
      "epoch": 0.05141007715406276,
      "grad_norm": 1.991543173789978,
      "learning_rate": 2.980478496841632e-05,
      "loss": 1.6416,
      "step": 26400
    },
    {
      "epoch": 0.05143928742517302,
      "grad_norm": 3.90432071685791,
      "learning_rate": 2.9804563553247076e-05,
      "loss": 1.7318,
      "step": 26415
    },
    {
      "epoch": 0.05146849769628328,
      "grad_norm": 3.582280158996582,
      "learning_rate": 2.980434201340652e-05,
      "loss": 1.7349,
      "step": 26430
    },
    {
      "epoch": 0.05149770796739355,
      "grad_norm": 3.1633496284484863,
      "learning_rate": 2.980412034889651e-05,
      "loss": 1.8023,
      "step": 26445
    },
    {
      "epoch": 0.05152691823850381,
      "grad_norm": 3.7939155101776123,
      "learning_rate": 2.9803898559718927e-05,
      "loss": 1.6626,
      "step": 26460
    },
    {
      "epoch": 0.05155612850961407,
      "grad_norm": 3.093492031097412,
      "learning_rate": 2.9803676645875634e-05,
      "loss": 1.7914,
      "step": 26475
    },
    {
      "epoch": 0.051585338780724334,
      "grad_norm": 2.4469592571258545,
      "learning_rate": 2.9803454607368493e-05,
      "loss": 1.7484,
      "step": 26490
    },
    {
      "epoch": 0.0516145490518346,
      "grad_norm": 2.613276243209839,
      "learning_rate": 2.9803232444199382e-05,
      "loss": 1.7984,
      "step": 26505
    },
    {
      "epoch": 0.05164375932294486,
      "grad_norm": 3.1793646812438965,
      "learning_rate": 2.9803010156370166e-05,
      "loss": 1.7114,
      "step": 26520
    },
    {
      "epoch": 0.051672969594055124,
      "grad_norm": 1.9021662473678589,
      "learning_rate": 2.980278774388272e-05,
      "loss": 1.6837,
      "step": 26535
    },
    {
      "epoch": 0.05170217986516539,
      "grad_norm": 2.415710687637329,
      "learning_rate": 2.9802565206738922e-05,
      "loss": 1.9818,
      "step": 26550
    },
    {
      "epoch": 0.05173139013627565,
      "grad_norm": 3.8227622509002686,
      "learning_rate": 2.9802342544940635e-05,
      "loss": 1.9691,
      "step": 26565
    },
    {
      "epoch": 0.051760600407385914,
      "grad_norm": 1.6945210695266724,
      "learning_rate": 2.980211975848974e-05,
      "loss": 1.9788,
      "step": 26580
    },
    {
      "epoch": 0.051789810678496175,
      "grad_norm": 2.1171348094940186,
      "learning_rate": 2.980189684738811e-05,
      "loss": 1.8498,
      "step": 26595
    },
    {
      "epoch": 0.05181902094960644,
      "grad_norm": 3.8129332065582275,
      "learning_rate": 2.9801673811637628e-05,
      "loss": 1.8109,
      "step": 26610
    },
    {
      "epoch": 0.0518482312207167,
      "grad_norm": 3.1358275413513184,
      "learning_rate": 2.9801450651240173e-05,
      "loss": 1.7809,
      "step": 26625
    },
    {
      "epoch": 0.051877441491826964,
      "grad_norm": 3.9287755489349365,
      "learning_rate": 2.9801227366197614e-05,
      "loss": 1.7791,
      "step": 26640
    },
    {
      "epoch": 0.05190665176293723,
      "grad_norm": 3.269742488861084,
      "learning_rate": 2.980100395651184e-05,
      "loss": 2.0053,
      "step": 26655
    },
    {
      "epoch": 0.05193586203404749,
      "grad_norm": 3.23358154296875,
      "learning_rate": 2.980078042218473e-05,
      "loss": 1.7863,
      "step": 26670
    },
    {
      "epoch": 0.051965072305157754,
      "grad_norm": 3.6313607692718506,
      "learning_rate": 2.980055676321817e-05,
      "loss": 1.8125,
      "step": 26685
    },
    {
      "epoch": 0.051994282576268015,
      "grad_norm": 2.303229331970215,
      "learning_rate": 2.9800332979614035e-05,
      "loss": 1.8336,
      "step": 26700
    },
    {
      "epoch": 0.05202349284737828,
      "grad_norm": 4.923130035400391,
      "learning_rate": 2.9800109071374216e-05,
      "loss": 1.9392,
      "step": 26715
    },
    {
      "epoch": 0.052052703118488544,
      "grad_norm": 6.160820484161377,
      "learning_rate": 2.9799885038500597e-05,
      "loss": 1.8991,
      "step": 26730
    },
    {
      "epoch": 0.052081913389598805,
      "grad_norm": 1.9204509258270264,
      "learning_rate": 2.9799660880995065e-05,
      "loss": 1.8511,
      "step": 26745
    },
    {
      "epoch": 0.05211112366070907,
      "grad_norm": 4.1540656089782715,
      "learning_rate": 2.9799436598859507e-05,
      "loss": 1.7353,
      "step": 26760
    },
    {
      "epoch": 0.052140333931819334,
      "grad_norm": 2.5721335411071777,
      "learning_rate": 2.979921219209581e-05,
      "loss": 1.6949,
      "step": 26775
    },
    {
      "epoch": 0.052169544202929595,
      "grad_norm": 2.5524466037750244,
      "learning_rate": 2.9798987660705867e-05,
      "loss": 1.9033,
      "step": 26790
    },
    {
      "epoch": 0.052198754474039856,
      "grad_norm": 2.8870813846588135,
      "learning_rate": 2.979876300469157e-05,
      "loss": 1.6899,
      "step": 26805
    },
    {
      "epoch": 0.052227964745150124,
      "grad_norm": 2.5837631225585938,
      "learning_rate": 2.9798538224054804e-05,
      "loss": 1.7715,
      "step": 26820
    },
    {
      "epoch": 0.052257175016260385,
      "grad_norm": 2.5015158653259277,
      "learning_rate": 2.979831331879747e-05,
      "loss": 1.9121,
      "step": 26835
    },
    {
      "epoch": 0.052286385287370646,
      "grad_norm": 3.3133704662323,
      "learning_rate": 2.9798088288921457e-05,
      "loss": 1.87,
      "step": 26850
    },
    {
      "epoch": 0.05231559555848091,
      "grad_norm": 2.159090042114258,
      "learning_rate": 2.979786313442866e-05,
      "loss": 1.9665,
      "step": 26865
    },
    {
      "epoch": 0.052344805829591175,
      "grad_norm": 3.6745707988739014,
      "learning_rate": 2.9797637855320977e-05,
      "loss": 1.8956,
      "step": 26880
    },
    {
      "epoch": 0.052374016100701436,
      "grad_norm": 2.593538761138916,
      "learning_rate": 2.9797412451600305e-05,
      "loss": 1.9206,
      "step": 26895
    },
    {
      "epoch": 0.0524032263718117,
      "grad_norm": 2.1524672508239746,
      "learning_rate": 2.979718692326854e-05,
      "loss": 2.0598,
      "step": 26910
    },
    {
      "epoch": 0.052432436642921965,
      "grad_norm": 5.403210639953613,
      "learning_rate": 2.9796961270327583e-05,
      "loss": 1.9446,
      "step": 26925
    },
    {
      "epoch": 0.052461646914032226,
      "grad_norm": 2.859320878982544,
      "learning_rate": 2.9796735492779338e-05,
      "loss": 1.6969,
      "step": 26940
    },
    {
      "epoch": 0.05249085718514249,
      "grad_norm": 2.0825371742248535,
      "learning_rate": 2.9796509590625696e-05,
      "loss": 1.8951,
      "step": 26955
    },
    {
      "epoch": 0.05252006745625275,
      "grad_norm": 2.3604981899261475,
      "learning_rate": 2.979628356386857e-05,
      "loss": 1.8861,
      "step": 26970
    },
    {
      "epoch": 0.052549277727363015,
      "grad_norm": 3.510629415512085,
      "learning_rate": 2.9796057412509856e-05,
      "loss": 1.9885,
      "step": 26985
    },
    {
      "epoch": 0.052578487998473276,
      "grad_norm": 1.72383713722229,
      "learning_rate": 2.9795831136551467e-05,
      "loss": 1.832,
      "step": 27000
    },
    {
      "epoch": 0.05260769826958354,
      "grad_norm": 2.767523765563965,
      "learning_rate": 2.9795604735995297e-05,
      "loss": 1.8956,
      "step": 27015
    },
    {
      "epoch": 0.052636908540693805,
      "grad_norm": 2.731154203414917,
      "learning_rate": 2.979537821084326e-05,
      "loss": 1.7157,
      "step": 27030
    },
    {
      "epoch": 0.052666118811804066,
      "grad_norm": 3.17053484916687,
      "learning_rate": 2.9795151561097265e-05,
      "loss": 1.9691,
      "step": 27045
    },
    {
      "epoch": 0.05269532908291433,
      "grad_norm": 3.9647130966186523,
      "learning_rate": 2.979492478675922e-05,
      "loss": 1.9736,
      "step": 27060
    },
    {
      "epoch": 0.05272453935402459,
      "grad_norm": 2.4074623584747314,
      "learning_rate": 2.9794697887831027e-05,
      "loss": 1.8325,
      "step": 27075
    },
    {
      "epoch": 0.052753749625134856,
      "grad_norm": 4.785901069641113,
      "learning_rate": 2.9794470864314603e-05,
      "loss": 1.9717,
      "step": 27090
    },
    {
      "epoch": 0.05278295989624512,
      "grad_norm": 5.4298577308654785,
      "learning_rate": 2.979424371621186e-05,
      "loss": 1.8316,
      "step": 27105
    },
    {
      "epoch": 0.05281217016735538,
      "grad_norm": 2.509413003921509,
      "learning_rate": 2.9794016443524713e-05,
      "loss": 1.8792,
      "step": 27120
    },
    {
      "epoch": 0.052841380438465646,
      "grad_norm": 1.903182029724121,
      "learning_rate": 2.979378904625507e-05,
      "loss": 1.8049,
      "step": 27135
    },
    {
      "epoch": 0.05287059070957591,
      "grad_norm": 3.3434927463531494,
      "learning_rate": 2.9793561524404846e-05,
      "loss": 1.7794,
      "step": 27150
    },
    {
      "epoch": 0.05289980098068617,
      "grad_norm": 5.064967632293701,
      "learning_rate": 2.9793333877975964e-05,
      "loss": 1.8726,
      "step": 27165
    },
    {
      "epoch": 0.05292901125179643,
      "grad_norm": 1.8450191020965576,
      "learning_rate": 2.9793106106970335e-05,
      "loss": 1.8586,
      "step": 27180
    },
    {
      "epoch": 0.0529582215229067,
      "grad_norm": 2.540570020675659,
      "learning_rate": 2.979287821138988e-05,
      "loss": 1.8988,
      "step": 27195
    },
    {
      "epoch": 0.05298743179401696,
      "grad_norm": 2.0893425941467285,
      "learning_rate": 2.9792650191236516e-05,
      "loss": 1.7794,
      "step": 27210
    },
    {
      "epoch": 0.05301664206512722,
      "grad_norm": 2.7562851905822754,
      "learning_rate": 2.979242204651216e-05,
      "loss": 1.7025,
      "step": 27225
    },
    {
      "epoch": 0.05304585233623749,
      "grad_norm": 4.634995937347412,
      "learning_rate": 2.9792193777218743e-05,
      "loss": 1.8236,
      "step": 27240
    },
    {
      "epoch": 0.05307506260734775,
      "grad_norm": 3.1855075359344482,
      "learning_rate": 2.9791965383358184e-05,
      "loss": 1.8439,
      "step": 27255
    },
    {
      "epoch": 0.05310427287845801,
      "grad_norm": 3.3286306858062744,
      "learning_rate": 2.9791736864932403e-05,
      "loss": 1.9314,
      "step": 27270
    },
    {
      "epoch": 0.05313348314956827,
      "grad_norm": 3.1028332710266113,
      "learning_rate": 2.979150822194332e-05,
      "loss": 2.0257,
      "step": 27285
    },
    {
      "epoch": 0.05316269342067854,
      "grad_norm": 4.1933393478393555,
      "learning_rate": 2.979127945439287e-05,
      "loss": 1.9908,
      "step": 27300
    },
    {
      "epoch": 0.0531919036917888,
      "grad_norm": 4.206679344177246,
      "learning_rate": 2.9791050562282974e-05,
      "loss": 1.8144,
      "step": 27315
    },
    {
      "epoch": 0.05322111396289906,
      "grad_norm": 2.231621265411377,
      "learning_rate": 2.9790821545615562e-05,
      "loss": 1.824,
      "step": 27330
    },
    {
      "epoch": 0.05325032423400932,
      "grad_norm": 3.625483512878418,
      "learning_rate": 2.9790592404392557e-05,
      "loss": 2.0087,
      "step": 27345
    },
    {
      "epoch": 0.05327953450511959,
      "grad_norm": 4.063029766082764,
      "learning_rate": 2.9790363138615902e-05,
      "loss": 1.8927,
      "step": 27360
    },
    {
      "epoch": 0.05330874477622985,
      "grad_norm": 4.166107654571533,
      "learning_rate": 2.979013374828751e-05,
      "loss": 1.9004,
      "step": 27375
    },
    {
      "epoch": 0.05333795504734011,
      "grad_norm": 2.737416982650757,
      "learning_rate": 2.9789904233409326e-05,
      "loss": 1.9678,
      "step": 27390
    },
    {
      "epoch": 0.05336716531845038,
      "grad_norm": 2.131272315979004,
      "learning_rate": 2.9789674593983277e-05,
      "loss": 1.895,
      "step": 27405
    },
    {
      "epoch": 0.05339637558956064,
      "grad_norm": 2.983872890472412,
      "learning_rate": 2.9789444830011302e-05,
      "loss": 1.7807,
      "step": 27420
    },
    {
      "epoch": 0.0534255858606709,
      "grad_norm": 5.785390377044678,
      "learning_rate": 2.978921494149533e-05,
      "loss": 1.9559,
      "step": 27435
    },
    {
      "epoch": 0.05345479613178116,
      "grad_norm": 2.7100813388824463,
      "learning_rate": 2.9788984928437298e-05,
      "loss": 1.9147,
      "step": 27450
    },
    {
      "epoch": 0.05348400640289143,
      "grad_norm": 2.0759334564208984,
      "learning_rate": 2.978875479083914e-05,
      "loss": 1.7345,
      "step": 27465
    },
    {
      "epoch": 0.05351321667400169,
      "grad_norm": 4.645893573760986,
      "learning_rate": 2.9788524528702804e-05,
      "loss": 1.7246,
      "step": 27480
    },
    {
      "epoch": 0.05354242694511195,
      "grad_norm": 3.1474320888519287,
      "learning_rate": 2.9788294142030225e-05,
      "loss": 1.8679,
      "step": 27495
    },
    {
      "epoch": 0.05357163721622222,
      "grad_norm": 2.8445558547973633,
      "learning_rate": 2.9788063630823335e-05,
      "loss": 1.748,
      "step": 27510
    },
    {
      "epoch": 0.05360084748733248,
      "grad_norm": 3.176330089569092,
      "learning_rate": 2.978783299508408e-05,
      "loss": 1.9698,
      "step": 27525
    },
    {
      "epoch": 0.05363005775844274,
      "grad_norm": 2.875300884246826,
      "learning_rate": 2.9787602234814407e-05,
      "loss": 1.7617,
      "step": 27540
    },
    {
      "epoch": 0.053659268029553,
      "grad_norm": 4.004800796508789,
      "learning_rate": 2.978737135001626e-05,
      "loss": 1.8204,
      "step": 27555
    },
    {
      "epoch": 0.05368847830066327,
      "grad_norm": 2.3588531017303467,
      "learning_rate": 2.9787140340691574e-05,
      "loss": 1.8028,
      "step": 27570
    },
    {
      "epoch": 0.05371768857177353,
      "grad_norm": 2.293210744857788,
      "learning_rate": 2.9786909206842297e-05,
      "loss": 1.9525,
      "step": 27585
    },
    {
      "epoch": 0.05374689884288379,
      "grad_norm": 3.4703333377838135,
      "learning_rate": 2.9786677948470382e-05,
      "loss": 1.8144,
      "step": 27600
    },
    {
      "epoch": 0.05377610911399406,
      "grad_norm": 2.1387853622436523,
      "learning_rate": 2.9786446565577772e-05,
      "loss": 1.7853,
      "step": 27615
    },
    {
      "epoch": 0.05380531938510432,
      "grad_norm": 1.9415435791015625,
      "learning_rate": 2.9786215058166417e-05,
      "loss": 1.8912,
      "step": 27630
    },
    {
      "epoch": 0.05383452965621458,
      "grad_norm": 3.315534830093384,
      "learning_rate": 2.978598342623826e-05,
      "loss": 1.7991,
      "step": 27645
    },
    {
      "epoch": 0.05386373992732484,
      "grad_norm": 3.2139084339141846,
      "learning_rate": 2.9785751669795265e-05,
      "loss": 1.8284,
      "step": 27660
    },
    {
      "epoch": 0.05389295019843511,
      "grad_norm": 2.0407660007476807,
      "learning_rate": 2.9785519788839368e-05,
      "loss": 1.8263,
      "step": 27675
    },
    {
      "epoch": 0.05392216046954537,
      "grad_norm": 3.6719486713409424,
      "learning_rate": 2.9785287783372538e-05,
      "loss": 1.7552,
      "step": 27690
    },
    {
      "epoch": 0.05395137074065563,
      "grad_norm": 2.3214800357818604,
      "learning_rate": 2.978505565339671e-05,
      "loss": 1.7752,
      "step": 27705
    },
    {
      "epoch": 0.0539805810117659,
      "grad_norm": 3.5290887355804443,
      "learning_rate": 2.9784823398913856e-05,
      "loss": 1.6684,
      "step": 27720
    },
    {
      "epoch": 0.05400979128287616,
      "grad_norm": 2.344564437866211,
      "learning_rate": 2.978459101992592e-05,
      "loss": 1.7649,
      "step": 27735
    },
    {
      "epoch": 0.05403900155398642,
      "grad_norm": 3.5880258083343506,
      "learning_rate": 2.9784358516434867e-05,
      "loss": 1.8194,
      "step": 27750
    },
    {
      "epoch": 0.05406821182509668,
      "grad_norm": 4.46945858001709,
      "learning_rate": 2.978412588844265e-05,
      "loss": 1.9552,
      "step": 27765
    },
    {
      "epoch": 0.05409742209620695,
      "grad_norm": 2.9063570499420166,
      "learning_rate": 2.978389313595123e-05,
      "loss": 1.9078,
      "step": 27780
    },
    {
      "epoch": 0.05412663236731721,
      "grad_norm": 4.740022659301758,
      "learning_rate": 2.9783660258962568e-05,
      "loss": 1.8773,
      "step": 27795
    },
    {
      "epoch": 0.05415584263842747,
      "grad_norm": 5.9590654373168945,
      "learning_rate": 2.9783427257478623e-05,
      "loss": 2.0891,
      "step": 27810
    },
    {
      "epoch": 0.05418505290953774,
      "grad_norm": 3.4905142784118652,
      "learning_rate": 2.978319413150136e-05,
      "loss": 1.912,
      "step": 27825
    },
    {
      "epoch": 0.054214263180648,
      "grad_norm": 4.312283039093018,
      "learning_rate": 2.978296088103273e-05,
      "loss": 1.7969,
      "step": 27840
    },
    {
      "epoch": 0.05424347345175826,
      "grad_norm": 3.419679880142212,
      "learning_rate": 2.978272750607472e-05,
      "loss": 2.0149,
      "step": 27855
    },
    {
      "epoch": 0.054272683722868524,
      "grad_norm": 1.4746593236923218,
      "learning_rate": 2.9782494006629275e-05,
      "loss": 1.7991,
      "step": 27870
    },
    {
      "epoch": 0.05430189399397879,
      "grad_norm": 2.6810593605041504,
      "learning_rate": 2.9782260382698374e-05,
      "loss": 1.8557,
      "step": 27885
    },
    {
      "epoch": 0.05433110426508905,
      "grad_norm": 4.144235134124756,
      "learning_rate": 2.9782026634283975e-05,
      "loss": 2.0213,
      "step": 27900
    },
    {
      "epoch": 0.054360314536199314,
      "grad_norm": 3.5831942558288574,
      "learning_rate": 2.9781792761388055e-05,
      "loss": 1.7606,
      "step": 27915
    },
    {
      "epoch": 0.054389524807309575,
      "grad_norm": 3.201408863067627,
      "learning_rate": 2.9781558764012573e-05,
      "loss": 1.7619,
      "step": 27930
    },
    {
      "epoch": 0.05441873507841984,
      "grad_norm": 2.8114073276519775,
      "learning_rate": 2.978132464215951e-05,
      "loss": 1.9014,
      "step": 27945
    },
    {
      "epoch": 0.054447945349530104,
      "grad_norm": 2.400465726852417,
      "learning_rate": 2.9781090395830834e-05,
      "loss": 1.9041,
      "step": 27960
    },
    {
      "epoch": 0.054477155620640365,
      "grad_norm": 2.1496269702911377,
      "learning_rate": 2.9780856025028513e-05,
      "loss": 1.7252,
      "step": 27975
    },
    {
      "epoch": 0.05450636589175063,
      "grad_norm": 3.477867603302002,
      "learning_rate": 2.978062152975453e-05,
      "loss": 1.7466,
      "step": 27990
    },
    {
      "epoch": 0.054535576162860894,
      "grad_norm": 2.1709723472595215,
      "learning_rate": 2.978038691001085e-05,
      "loss": 1.6808,
      "step": 28005
    },
    {
      "epoch": 0.054564786433971155,
      "grad_norm": 2.13655686378479,
      "learning_rate": 2.978015216579945e-05,
      "loss": 1.8089,
      "step": 28020
    },
    {
      "epoch": 0.054593996705081416,
      "grad_norm": 3.563598394393921,
      "learning_rate": 2.9779917297122318e-05,
      "loss": 1.8368,
      "step": 28035
    },
    {
      "epoch": 0.054623206976191684,
      "grad_norm": 3.1836514472961426,
      "learning_rate": 2.977968230398142e-05,
      "loss": 1.8042,
      "step": 28050
    },
    {
      "epoch": 0.054652417247301945,
      "grad_norm": 2.4931259155273438,
      "learning_rate": 2.9779447186378738e-05,
      "loss": 1.8061,
      "step": 28065
    },
    {
      "epoch": 0.054681627518412206,
      "grad_norm": 3.2399775981903076,
      "learning_rate": 2.977921194431625e-05,
      "loss": 1.8723,
      "step": 28080
    },
    {
      "epoch": 0.05471083778952247,
      "grad_norm": 3.281590461730957,
      "learning_rate": 2.977897657779594e-05,
      "loss": 1.8882,
      "step": 28095
    },
    {
      "epoch": 0.054740048060632734,
      "grad_norm": 2.607039451599121,
      "learning_rate": 2.9778741086819795e-05,
      "loss": 1.8796,
      "step": 28110
    },
    {
      "epoch": 0.054769258331742995,
      "grad_norm": 3.4543254375457764,
      "learning_rate": 2.977850547138979e-05,
      "loss": 1.8826,
      "step": 28125
    },
    {
      "epoch": 0.054798468602853256,
      "grad_norm": 3.8168885707855225,
      "learning_rate": 2.9778269731507914e-05,
      "loss": 1.9827,
      "step": 28140
    },
    {
      "epoch": 0.054827678873963524,
      "grad_norm": 2.7308406829833984,
      "learning_rate": 2.977803386717615e-05,
      "loss": 1.768,
      "step": 28155
    },
    {
      "epoch": 0.054856889145073785,
      "grad_norm": 4.276648998260498,
      "learning_rate": 2.9777797878396477e-05,
      "loss": 1.8696,
      "step": 28170
    },
    {
      "epoch": 0.054886099416184046,
      "grad_norm": 3.1874961853027344,
      "learning_rate": 2.97775617651709e-05,
      "loss": 1.881,
      "step": 28185
    },
    {
      "epoch": 0.054915309687294314,
      "grad_norm": 2.4808239936828613,
      "learning_rate": 2.977732552750139e-05,
      "loss": 1.7825,
      "step": 28200
    },
    {
      "epoch": 0.054944519958404575,
      "grad_norm": 3.0970373153686523,
      "learning_rate": 2.9777089165389942e-05,
      "loss": 1.7825,
      "step": 28215
    },
    {
      "epoch": 0.054973730229514836,
      "grad_norm": 1.757534384727478,
      "learning_rate": 2.9776852678838555e-05,
      "loss": 2.0145,
      "step": 28230
    },
    {
      "epoch": 0.0550029405006251,
      "grad_norm": 2.351555347442627,
      "learning_rate": 2.977661606784921e-05,
      "loss": 2.0268,
      "step": 28245
    },
    {
      "epoch": 0.055032150771735365,
      "grad_norm": 2.4561007022857666,
      "learning_rate": 2.9776379332423902e-05,
      "loss": 1.7745,
      "step": 28260
    },
    {
      "epoch": 0.055061361042845626,
      "grad_norm": 3.8144009113311768,
      "learning_rate": 2.9776142472564624e-05,
      "loss": 1.8408,
      "step": 28275
    },
    {
      "epoch": 0.05509057131395589,
      "grad_norm": 4.969006061553955,
      "learning_rate": 2.9775905488273373e-05,
      "loss": 1.9223,
      "step": 28290
    },
    {
      "epoch": 0.055119781585066155,
      "grad_norm": 5.71866512298584,
      "learning_rate": 2.9775668379552146e-05,
      "loss": 1.842,
      "step": 28305
    },
    {
      "epoch": 0.055148991856176416,
      "grad_norm": 1.9934380054473877,
      "learning_rate": 2.9775431146402937e-05,
      "loss": 1.976,
      "step": 28320
    },
    {
      "epoch": 0.05517820212728668,
      "grad_norm": 1.7895939350128174,
      "learning_rate": 2.9775193788827743e-05,
      "loss": 2.0921,
      "step": 28335
    },
    {
      "epoch": 0.05520741239839694,
      "grad_norm": 4.20900821685791,
      "learning_rate": 2.9774956306828566e-05,
      "loss": 1.9333,
      "step": 28350
    },
    {
      "epoch": 0.055236622669507206,
      "grad_norm": 1.7822163105010986,
      "learning_rate": 2.97747187004074e-05,
      "loss": 2.0872,
      "step": 28365
    },
    {
      "epoch": 0.05526583294061747,
      "grad_norm": 1.7469080686569214,
      "learning_rate": 2.9774480969566254e-05,
      "loss": 1.8781,
      "step": 28380
    },
    {
      "epoch": 0.05529504321172773,
      "grad_norm": 3.0533454418182373,
      "learning_rate": 2.977424311430712e-05,
      "loss": 1.7184,
      "step": 28395
    },
    {
      "epoch": 0.05532425348283799,
      "grad_norm": 2.729780912399292,
      "learning_rate": 2.977400513463201e-05,
      "loss": 1.7794,
      "step": 28410
    },
    {
      "epoch": 0.05535346375394826,
      "grad_norm": 3.2360620498657227,
      "learning_rate": 2.9773767030542926e-05,
      "loss": 1.7711,
      "step": 28425
    },
    {
      "epoch": 0.05538267402505852,
      "grad_norm": 2.3599355220794678,
      "learning_rate": 2.9773528802041873e-05,
      "loss": 1.8758,
      "step": 28440
    },
    {
      "epoch": 0.05541188429616878,
      "grad_norm": 3.3827106952667236,
      "learning_rate": 2.9773290449130856e-05,
      "loss": 1.901,
      "step": 28455
    },
    {
      "epoch": 0.055441094567279046,
      "grad_norm": 2.9506921768188477,
      "learning_rate": 2.977305197181188e-05,
      "loss": 1.9286,
      "step": 28470
    },
    {
      "epoch": 0.05547030483838931,
      "grad_norm": 3.381622791290283,
      "learning_rate": 2.9772813370086956e-05,
      "loss": 2.0169,
      "step": 28485
    },
    {
      "epoch": 0.05549951510949957,
      "grad_norm": 3.6618142127990723,
      "learning_rate": 2.9772574643958095e-05,
      "loss": 1.8318,
      "step": 28500
    },
    {
      "epoch": 0.05552872538060983,
      "grad_norm": 2.906064033508301,
      "learning_rate": 2.9772335793427304e-05,
      "loss": 1.8778,
      "step": 28515
    },
    {
      "epoch": 0.0555579356517201,
      "grad_norm": 1.9280356168746948,
      "learning_rate": 2.9772096818496592e-05,
      "loss": 1.9095,
      "step": 28530
    },
    {
      "epoch": 0.05558714592283036,
      "grad_norm": 2.451441764831543,
      "learning_rate": 2.977185771916798e-05,
      "loss": 1.7975,
      "step": 28545
    },
    {
      "epoch": 0.05561635619394062,
      "grad_norm": 3.4363293647766113,
      "learning_rate": 2.9771618495443473e-05,
      "loss": 1.9995,
      "step": 28560
    },
    {
      "epoch": 0.05564556646505089,
      "grad_norm": 3.402430772781372,
      "learning_rate": 2.9771379147325095e-05,
      "loss": 1.879,
      "step": 28575
    },
    {
      "epoch": 0.05567477673616115,
      "grad_norm": 3.7891762256622314,
      "learning_rate": 2.977113967481485e-05,
      "loss": 1.8275,
      "step": 28590
    },
    {
      "epoch": 0.05570398700727141,
      "grad_norm": 2.45809268951416,
      "learning_rate": 2.977090007791476e-05,
      "loss": 1.8131,
      "step": 28605
    },
    {
      "epoch": 0.05573319727838167,
      "grad_norm": 3.5447278022766113,
      "learning_rate": 2.9770660356626848e-05,
      "loss": 1.8373,
      "step": 28620
    },
    {
      "epoch": 0.05576240754949194,
      "grad_norm": 3.397735595703125,
      "learning_rate": 2.9770420510953124e-05,
      "loss": 1.7907,
      "step": 28635
    },
    {
      "epoch": 0.0557916178206022,
      "grad_norm": 2.185011386871338,
      "learning_rate": 2.9770180540895613e-05,
      "loss": 1.8909,
      "step": 28650
    },
    {
      "epoch": 0.05582082809171246,
      "grad_norm": 3.66780161857605,
      "learning_rate": 2.9769940446456332e-05,
      "loss": 1.8898,
      "step": 28665
    },
    {
      "epoch": 0.05585003836282273,
      "grad_norm": 3.8541507720947266,
      "learning_rate": 2.9769700227637307e-05,
      "loss": 1.8156,
      "step": 28680
    },
    {
      "epoch": 0.05587924863393299,
      "grad_norm": 2.504997491836548,
      "learning_rate": 2.9769459884440563e-05,
      "loss": 1.8584,
      "step": 28695
    },
    {
      "epoch": 0.05590845890504325,
      "grad_norm": 2.811286687850952,
      "learning_rate": 2.9769219416868114e-05,
      "loss": 1.8762,
      "step": 28710
    },
    {
      "epoch": 0.05593766917615351,
      "grad_norm": 3.2774808406829834,
      "learning_rate": 2.976897882492199e-05,
      "loss": 1.8637,
      "step": 28725
    },
    {
      "epoch": 0.05596687944726378,
      "grad_norm": 3.6415061950683594,
      "learning_rate": 2.9768738108604222e-05,
      "loss": 2.033,
      "step": 28740
    },
    {
      "epoch": 0.05599608971837404,
      "grad_norm": 2.5112550258636475,
      "learning_rate": 2.9768497267916833e-05,
      "loss": 1.9643,
      "step": 28755
    },
    {
      "epoch": 0.0560252999894843,
      "grad_norm": 3.3538918495178223,
      "learning_rate": 2.9768256302861852e-05,
      "loss": 1.7964,
      "step": 28770
    },
    {
      "epoch": 0.05605451026059457,
      "grad_norm": 1.7511709928512573,
      "learning_rate": 2.9768015213441306e-05,
      "loss": 1.9891,
      "step": 28785
    },
    {
      "epoch": 0.05608372053170483,
      "grad_norm": 4.007176399230957,
      "learning_rate": 2.9767773999657225e-05,
      "loss": 2.0263,
      "step": 28800
    },
    {
      "epoch": 0.05611293080281509,
      "grad_norm": 2.968200206756592,
      "learning_rate": 2.9767532661511644e-05,
      "loss": 1.7171,
      "step": 28815
    },
    {
      "epoch": 0.05614214107392535,
      "grad_norm": 3.2194066047668457,
      "learning_rate": 2.9767291199006594e-05,
      "loss": 1.8825,
      "step": 28830
    },
    {
      "epoch": 0.05617135134503562,
      "grad_norm": 3.994147539138794,
      "learning_rate": 2.976704961214411e-05,
      "loss": 1.6859,
      "step": 28845
    },
    {
      "epoch": 0.05620056161614588,
      "grad_norm": 2.1359071731567383,
      "learning_rate": 2.976680790092622e-05,
      "loss": 1.7796,
      "step": 28860
    },
    {
      "epoch": 0.05622977188725614,
      "grad_norm": 3.074885606765747,
      "learning_rate": 2.976656606535497e-05,
      "loss": 1.9507,
      "step": 28875
    },
    {
      "epoch": 0.05625898215836641,
      "grad_norm": 3.184913396835327,
      "learning_rate": 2.9766324105432385e-05,
      "loss": 1.9536,
      "step": 28890
    },
    {
      "epoch": 0.05628819242947667,
      "grad_norm": 3.911243438720703,
      "learning_rate": 2.976608202116051e-05,
      "loss": 1.8609,
      "step": 28905
    },
    {
      "epoch": 0.05631740270058693,
      "grad_norm": 4.364305019378662,
      "learning_rate": 2.9765839812541378e-05,
      "loss": 1.8573,
      "step": 28920
    },
    {
      "epoch": 0.05634661297169719,
      "grad_norm": 3.0088484287261963,
      "learning_rate": 2.9765597479577034e-05,
      "loss": 2.0879,
      "step": 28935
    },
    {
      "epoch": 0.05637582324280746,
      "grad_norm": 1.9867300987243652,
      "learning_rate": 2.9765355022269518e-05,
      "loss": 1.7799,
      "step": 28950
    },
    {
      "epoch": 0.05640503351391772,
      "grad_norm": 2.1443519592285156,
      "learning_rate": 2.9765112440620874e-05,
      "loss": 1.7201,
      "step": 28965
    },
    {
      "epoch": 0.05643424378502798,
      "grad_norm": 2.226407051086426,
      "learning_rate": 2.9764869734633134e-05,
      "loss": 1.7974,
      "step": 28980
    },
    {
      "epoch": 0.05646345405613824,
      "grad_norm": 4.08579158782959,
      "learning_rate": 2.9764626904308354e-05,
      "loss": 1.8633,
      "step": 28995
    },
    {
      "epoch": 0.05649266432724851,
      "grad_norm": 4.314965724945068,
      "learning_rate": 2.9764383949648576e-05,
      "loss": 1.9242,
      "step": 29010
    },
    {
      "epoch": 0.05652187459835877,
      "grad_norm": 4.151242733001709,
      "learning_rate": 2.976414087065584e-05,
      "loss": 1.942,
      "step": 29025
    },
    {
      "epoch": 0.05655108486946903,
      "grad_norm": 1.9677321910858154,
      "learning_rate": 2.97638976673322e-05,
      "loss": 1.8371,
      "step": 29040
    },
    {
      "epoch": 0.0565802951405793,
      "grad_norm": 2.814545154571533,
      "learning_rate": 2.97636543396797e-05,
      "loss": 1.9929,
      "step": 29055
    },
    {
      "epoch": 0.05660950541168956,
      "grad_norm": 3.4638845920562744,
      "learning_rate": 2.976341088770039e-05,
      "loss": 1.8635,
      "step": 29070
    },
    {
      "epoch": 0.05663871568279982,
      "grad_norm": 2.0748291015625,
      "learning_rate": 2.976316731139632e-05,
      "loss": 1.6827,
      "step": 29085
    },
    {
      "epoch": 0.056667925953910084,
      "grad_norm": 2.4861667156219482,
      "learning_rate": 2.9762923610769545e-05,
      "loss": 1.8022,
      "step": 29100
    },
    {
      "epoch": 0.05669713622502035,
      "grad_norm": 2.12156343460083,
      "learning_rate": 2.9762679785822113e-05,
      "loss": 1.8912,
      "step": 29115
    },
    {
      "epoch": 0.05672634649613061,
      "grad_norm": 2.3051788806915283,
      "learning_rate": 2.9762435836556075e-05,
      "loss": 1.786,
      "step": 29130
    },
    {
      "epoch": 0.056755556767240874,
      "grad_norm": 3.0045323371887207,
      "learning_rate": 2.9762191762973492e-05,
      "loss": 1.9105,
      "step": 29145
    },
    {
      "epoch": 0.05678476703835114,
      "grad_norm": 2.9411308765411377,
      "learning_rate": 2.9761947565076413e-05,
      "loss": 1.7949,
      "step": 29160
    },
    {
      "epoch": 0.0568139773094614,
      "grad_norm": 3.0440449714660645,
      "learning_rate": 2.97617032428669e-05,
      "loss": 2.0024,
      "step": 29175
    },
    {
      "epoch": 0.056843187580571664,
      "grad_norm": 3.836582660675049,
      "learning_rate": 2.976145879634701e-05,
      "loss": 1.9192,
      "step": 29190
    },
    {
      "epoch": 0.056872397851681925,
      "grad_norm": 3.63226580619812,
      "learning_rate": 2.9761214225518792e-05,
      "loss": 1.6728,
      "step": 29205
    },
    {
      "epoch": 0.05690160812279219,
      "grad_norm": 2.588270902633667,
      "learning_rate": 2.9760969530384317e-05,
      "loss": 1.8828,
      "step": 29220
    },
    {
      "epoch": 0.05693081839390245,
      "grad_norm": 3.4929933547973633,
      "learning_rate": 2.9760724710945642e-05,
      "loss": 2.0755,
      "step": 29235
    },
    {
      "epoch": 0.056960028665012714,
      "grad_norm": 2.916142463684082,
      "learning_rate": 2.976047976720483e-05,
      "loss": 1.8955,
      "step": 29250
    },
    {
      "epoch": 0.05698923893612298,
      "grad_norm": 3.606926918029785,
      "learning_rate": 2.976023469916394e-05,
      "loss": 1.9196,
      "step": 29265
    },
    {
      "epoch": 0.05701844920723324,
      "grad_norm": 4.033385276794434,
      "learning_rate": 2.9759989506825033e-05,
      "loss": 1.9555,
      "step": 29280
    },
    {
      "epoch": 0.057047659478343504,
      "grad_norm": 2.1779022216796875,
      "learning_rate": 2.9759744190190185e-05,
      "loss": 1.7735,
      "step": 29295
    },
    {
      "epoch": 0.057076869749453765,
      "grad_norm": 4.592677116394043,
      "learning_rate": 2.9759498749261452e-05,
      "loss": 1.8971,
      "step": 29310
    },
    {
      "epoch": 0.05710608002056403,
      "grad_norm": 4.797886371612549,
      "learning_rate": 2.9759253184040906e-05,
      "loss": 1.8494,
      "step": 29325
    },
    {
      "epoch": 0.057135290291674294,
      "grad_norm": 2.316049337387085,
      "learning_rate": 2.9759007494530615e-05,
      "loss": 1.7288,
      "step": 29340
    },
    {
      "epoch": 0.057164500562784555,
      "grad_norm": 2.9396653175354004,
      "learning_rate": 2.975876168073264e-05,
      "loss": 1.8751,
      "step": 29355
    },
    {
      "epoch": 0.05719371083389482,
      "grad_norm": 4.224997043609619,
      "learning_rate": 2.9758515742649063e-05,
      "loss": 1.78,
      "step": 29370
    },
    {
      "epoch": 0.057222921105005084,
      "grad_norm": 2.5549795627593994,
      "learning_rate": 2.9758269680281946e-05,
      "loss": 1.7375,
      "step": 29385
    },
    {
      "epoch": 0.057252131376115345,
      "grad_norm": 3.9945852756500244,
      "learning_rate": 2.9758023493633365e-05,
      "loss": 1.7078,
      "step": 29400
    },
    {
      "epoch": 0.057281341647225606,
      "grad_norm": 3.5278730392456055,
      "learning_rate": 2.975777718270539e-05,
      "loss": 1.7572,
      "step": 29415
    },
    {
      "epoch": 0.057310551918335874,
      "grad_norm": 3.9674766063690186,
      "learning_rate": 2.9757530747500104e-05,
      "loss": 1.9125,
      "step": 29430
    },
    {
      "epoch": 0.057339762189446135,
      "grad_norm": 3.498309373855591,
      "learning_rate": 2.9757284188019573e-05,
      "loss": 1.8242,
      "step": 29445
    },
    {
      "epoch": 0.057368972460556396,
      "grad_norm": 3.2913856506347656,
      "learning_rate": 2.9757037504265874e-05,
      "loss": 1.8559,
      "step": 29460
    },
    {
      "epoch": 0.05739818273166666,
      "grad_norm": 2.72259521484375,
      "learning_rate": 2.9756790696241088e-05,
      "loss": 1.7855,
      "step": 29475
    },
    {
      "epoch": 0.057427393002776925,
      "grad_norm": 7.3963704109191895,
      "learning_rate": 2.9756543763947292e-05,
      "loss": 1.9272,
      "step": 29490
    },
    {
      "epoch": 0.057456603273887186,
      "grad_norm": 2.901502847671509,
      "learning_rate": 2.9756296707386566e-05,
      "loss": 1.7294,
      "step": 29505
    },
    {
      "epoch": 0.05748581354499745,
      "grad_norm": 1.9630255699157715,
      "learning_rate": 2.9756049526560995e-05,
      "loss": 1.8477,
      "step": 29520
    },
    {
      "epoch": 0.057515023816107715,
      "grad_norm": 2.449836492538452,
      "learning_rate": 2.975580222147265e-05,
      "loss": 1.9282,
      "step": 29535
    },
    {
      "epoch": 0.057544234087217976,
      "grad_norm": 2.7461347579956055,
      "learning_rate": 2.9755554792123617e-05,
      "loss": 1.8883,
      "step": 29550
    },
    {
      "epoch": 0.05757344435832824,
      "grad_norm": 3.3567450046539307,
      "learning_rate": 2.9755307238515986e-05,
      "loss": 1.9717,
      "step": 29565
    },
    {
      "epoch": 0.0576026546294385,
      "grad_norm": 3.6341872215270996,
      "learning_rate": 2.975505956065184e-05,
      "loss": 1.9378,
      "step": 29580
    },
    {
      "epoch": 0.057631864900548765,
      "grad_norm": 4.170873165130615,
      "learning_rate": 2.9754811758533253e-05,
      "loss": 1.9077,
      "step": 29595
    },
    {
      "epoch": 0.057661075171659026,
      "grad_norm": 1.9185329675674438,
      "learning_rate": 2.975456383216233e-05,
      "loss": 1.7987,
      "step": 29610
    },
    {
      "epoch": 0.05769028544276929,
      "grad_norm": 2.3139781951904297,
      "learning_rate": 2.9754315781541144e-05,
      "loss": 1.9518,
      "step": 29625
    },
    {
      "epoch": 0.057719495713879555,
      "grad_norm": 3.581907033920288,
      "learning_rate": 2.9754067606671794e-05,
      "loss": 1.8729,
      "step": 29640
    },
    {
      "epoch": 0.057748705984989816,
      "grad_norm": 2.490278720855713,
      "learning_rate": 2.975381930755636e-05,
      "loss": 1.7576,
      "step": 29655
    },
    {
      "epoch": 0.05777791625610008,
      "grad_norm": 3.5740153789520264,
      "learning_rate": 2.9753570884196942e-05,
      "loss": 1.9749,
      "step": 29670
    },
    {
      "epoch": 0.05780712652721034,
      "grad_norm": 3.012131929397583,
      "learning_rate": 2.975332233659563e-05,
      "loss": 1.8704,
      "step": 29685
    },
    {
      "epoch": 0.057836336798320606,
      "grad_norm": 4.225703239440918,
      "learning_rate": 2.9753073664754514e-05,
      "loss": 1.6134,
      "step": 29700
    },
    {
      "epoch": 0.05786554706943087,
      "grad_norm": 3.840487480163574,
      "learning_rate": 2.9752824868675693e-05,
      "loss": 1.822,
      "step": 29715
    },
    {
      "epoch": 0.05789475734054113,
      "grad_norm": 2.8694260120391846,
      "learning_rate": 2.975257594836125e-05,
      "loss": 1.939,
      "step": 29730
    },
    {
      "epoch": 0.057923967611651396,
      "grad_norm": 3.4660465717315674,
      "learning_rate": 2.97523269038133e-05,
      "loss": 1.9517,
      "step": 29745
    },
    {
      "epoch": 0.05795317788276166,
      "grad_norm": 3.125666379928589,
      "learning_rate": 2.9752077735033924e-05,
      "loss": 1.7034,
      "step": 29760
    },
    {
      "epoch": 0.05798238815387192,
      "grad_norm": 4.895527362823486,
      "learning_rate": 2.975182844202523e-05,
      "loss": 1.8021,
      "step": 29775
    },
    {
      "epoch": 0.05801159842498218,
      "grad_norm": 3.4364778995513916,
      "learning_rate": 2.9751579024789314e-05,
      "loss": 1.9357,
      "step": 29790
    },
    {
      "epoch": 0.05804080869609245,
      "grad_norm": 3.209791421890259,
      "learning_rate": 2.9751329483328276e-05,
      "loss": 1.8969,
      "step": 29805
    },
    {
      "epoch": 0.05807001896720271,
      "grad_norm": 2.851810932159424,
      "learning_rate": 2.9751079817644217e-05,
      "loss": 1.8274,
      "step": 29820
    },
    {
      "epoch": 0.05809922923831297,
      "grad_norm": 1.9028266668319702,
      "learning_rate": 2.975083002773924e-05,
      "loss": 1.7498,
      "step": 29835
    },
    {
      "epoch": 0.05812843950942324,
      "grad_norm": 4.442659378051758,
      "learning_rate": 2.9750580113615448e-05,
      "loss": 1.906,
      "step": 29850
    },
    {
      "epoch": 0.0581576497805335,
      "grad_norm": 4.513755798339844,
      "learning_rate": 2.9750330075274948e-05,
      "loss": 1.7855,
      "step": 29865
    },
    {
      "epoch": 0.05818686005164376,
      "grad_norm": 2.4421234130859375,
      "learning_rate": 2.975007991271984e-05,
      "loss": 1.8355,
      "step": 29880
    },
    {
      "epoch": 0.05821607032275402,
      "grad_norm": 2.0383033752441406,
      "learning_rate": 2.974982962595224e-05,
      "loss": 1.8456,
      "step": 29895
    },
    {
      "epoch": 0.05824528059386429,
      "grad_norm": 2.9908933639526367,
      "learning_rate": 2.9749579214974245e-05,
      "loss": 1.7223,
      "step": 29910
    },
    {
      "epoch": 0.05827449086497455,
      "grad_norm": 3.1746137142181396,
      "learning_rate": 2.9749328679787976e-05,
      "loss": 1.8949,
      "step": 29925
    },
    {
      "epoch": 0.05830370113608481,
      "grad_norm": 3.351712942123413,
      "learning_rate": 2.9749078020395526e-05,
      "loss": 1.8423,
      "step": 29940
    },
    {
      "epoch": 0.05833291140719508,
      "grad_norm": 4.363383769989014,
      "learning_rate": 2.9748827236799024e-05,
      "loss": 2.0718,
      "step": 29955
    },
    {
      "epoch": 0.05836212167830534,
      "grad_norm": 3.5311388969421387,
      "learning_rate": 2.974857632900057e-05,
      "loss": 1.8393,
      "step": 29970
    },
    {
      "epoch": 0.0583913319494156,
      "grad_norm": 2.739178419113159,
      "learning_rate": 2.974832529700228e-05,
      "loss": 1.901,
      "step": 29985
    },
    {
      "epoch": 0.05842054222052586,
      "grad_norm": 5.376321315765381,
      "learning_rate": 2.974807414080627e-05,
      "loss": 1.9742,
      "step": 30000
    },
    {
      "epoch": 0.05844975249163613,
      "grad_norm": 4.261213302612305,
      "learning_rate": 2.9747822860414652e-05,
      "loss": 1.747,
      "step": 30015
    },
    {
      "epoch": 0.05847896276274639,
      "grad_norm": 2.775156259536743,
      "learning_rate": 2.9747571455829544e-05,
      "loss": 1.8235,
      "step": 30030
    },
    {
      "epoch": 0.05850817303385665,
      "grad_norm": 4.271034240722656,
      "learning_rate": 2.9747319927053066e-05,
      "loss": 1.834,
      "step": 30045
    },
    {
      "epoch": 0.05853738330496691,
      "grad_norm": 3.628793239593506,
      "learning_rate": 2.9747068274087327e-05,
      "loss": 2.0451,
      "step": 30060
    },
    {
      "epoch": 0.05856659357607718,
      "grad_norm": 2.388171434402466,
      "learning_rate": 2.9746816496934452e-05,
      "loss": 1.9012,
      "step": 30075
    },
    {
      "epoch": 0.05859580384718744,
      "grad_norm": 5.260224342346191,
      "learning_rate": 2.9746564595596566e-05,
      "loss": 1.8394,
      "step": 30090
    },
    {
      "epoch": 0.0586250141182977,
      "grad_norm": 3.47929310798645,
      "learning_rate": 2.9746312570075786e-05,
      "loss": 1.7858,
      "step": 30105
    },
    {
      "epoch": 0.05865422438940797,
      "grad_norm": 2.082864761352539,
      "learning_rate": 2.974606042037423e-05,
      "loss": 1.8976,
      "step": 30120
    },
    {
      "epoch": 0.05868343466051823,
      "grad_norm": 2.003638982772827,
      "learning_rate": 2.9745808146494027e-05,
      "loss": 1.7789,
      "step": 30135
    },
    {
      "epoch": 0.05871264493162849,
      "grad_norm": 3.8583061695098877,
      "learning_rate": 2.9745555748437298e-05,
      "loss": 1.9207,
      "step": 30150
    },
    {
      "epoch": 0.05874185520273875,
      "grad_norm": 4.0564703941345215,
      "learning_rate": 2.9745303226206172e-05,
      "loss": 1.9422,
      "step": 30165
    },
    {
      "epoch": 0.05877106547384902,
      "grad_norm": 1.928350806236267,
      "learning_rate": 2.974505057980277e-05,
      "loss": 1.8624,
      "step": 30180
    },
    {
      "epoch": 0.05880027574495928,
      "grad_norm": 2.1937801837921143,
      "learning_rate": 2.974479780922923e-05,
      "loss": 1.9617,
      "step": 30195
    },
    {
      "epoch": 0.05882948601606954,
      "grad_norm": 4.145959854125977,
      "learning_rate": 2.974454491448767e-05,
      "loss": 1.8212,
      "step": 30210
    },
    {
      "epoch": 0.05885869628717981,
      "grad_norm": 4.780306816101074,
      "learning_rate": 2.9744291895580222e-05,
      "loss": 1.7573,
      "step": 30225
    },
    {
      "epoch": 0.05888790655829007,
      "grad_norm": 3.6125333309173584,
      "learning_rate": 2.9744038752509016e-05,
      "loss": 1.9406,
      "step": 30240
    },
    {
      "epoch": 0.05891711682940033,
      "grad_norm": 2.673916816711426,
      "learning_rate": 2.974378548527619e-05,
      "loss": 1.8403,
      "step": 30255
    },
    {
      "epoch": 0.05894632710051059,
      "grad_norm": 2.3269338607788086,
      "learning_rate": 2.9743532093883877e-05,
      "loss": 2.1496,
      "step": 30270
    },
    {
      "epoch": 0.05897553737162086,
      "grad_norm": 4.606401443481445,
      "learning_rate": 2.9743278578334205e-05,
      "loss": 1.7679,
      "step": 30285
    },
    {
      "epoch": 0.05900474764273112,
      "grad_norm": 4.217224597930908,
      "learning_rate": 2.9743024938629303e-05,
      "loss": 1.8091,
      "step": 30300
    },
    {
      "epoch": 0.05903395791384138,
      "grad_norm": 2.7839462757110596,
      "learning_rate": 2.974277117477132e-05,
      "loss": 1.9216,
      "step": 30315
    },
    {
      "epoch": 0.05906316818495165,
      "grad_norm": 2.616410970687866,
      "learning_rate": 2.9742517286762386e-05,
      "loss": 1.7535,
      "step": 30330
    },
    {
      "epoch": 0.05909237845606191,
      "grad_norm": 2.646209239959717,
      "learning_rate": 2.974226327460464e-05,
      "loss": 1.7763,
      "step": 30345
    },
    {
      "epoch": 0.05912158872717217,
      "grad_norm": 4.532891750335693,
      "learning_rate": 2.9742009138300225e-05,
      "loss": 1.9677,
      "step": 30360
    },
    {
      "epoch": 0.05915079899828243,
      "grad_norm": 3.891831874847412,
      "learning_rate": 2.9741754877851277e-05,
      "loss": 1.8411,
      "step": 30375
    },
    {
      "epoch": 0.0591800092693927,
      "grad_norm": 2.234100580215454,
      "learning_rate": 2.9741500493259933e-05,
      "loss": 1.77,
      "step": 30390
    },
    {
      "epoch": 0.05920921954050296,
      "grad_norm": 3.4290971755981445,
      "learning_rate": 2.9741245984528342e-05,
      "loss": 1.9234,
      "step": 30405
    },
    {
      "epoch": 0.05923842981161322,
      "grad_norm": 2.175567150115967,
      "learning_rate": 2.974099135165865e-05,
      "loss": 2.1054,
      "step": 30420
    },
    {
      "epoch": 0.05926764008272349,
      "grad_norm": 2.412052631378174,
      "learning_rate": 2.9740736594652992e-05,
      "loss": 1.8607,
      "step": 30435
    },
    {
      "epoch": 0.05929685035383375,
      "grad_norm": 2.27789044380188,
      "learning_rate": 2.974048171351352e-05,
      "loss": 1.9491,
      "step": 30450
    },
    {
      "epoch": 0.05932606062494401,
      "grad_norm": 3.9195199012756348,
      "learning_rate": 2.9740226708242375e-05,
      "loss": 1.8516,
      "step": 30465
    },
    {
      "epoch": 0.059355270896054274,
      "grad_norm": 2.3743503093719482,
      "learning_rate": 2.973997157884171e-05,
      "loss": 2.1744,
      "step": 30480
    },
    {
      "epoch": 0.05938448116716454,
      "grad_norm": 4.110390663146973,
      "learning_rate": 2.9739716325313676e-05,
      "loss": 1.8803,
      "step": 30495
    },
    {
      "epoch": 0.0594136914382748,
      "grad_norm": 3.738240957260132,
      "learning_rate": 2.973946094766041e-05,
      "loss": 1.7978,
      "step": 30510
    },
    {
      "epoch": 0.059442901709385064,
      "grad_norm": 2.840871572494507,
      "learning_rate": 2.9739205445884075e-05,
      "loss": 1.8413,
      "step": 30525
    },
    {
      "epoch": 0.059472111980495325,
      "grad_norm": 2.931602954864502,
      "learning_rate": 2.9738949819986814e-05,
      "loss": 1.9035,
      "step": 30540
    },
    {
      "epoch": 0.05950132225160559,
      "grad_norm": 2.814918041229248,
      "learning_rate": 2.9738694069970787e-05,
      "loss": 1.9169,
      "step": 30555
    },
    {
      "epoch": 0.059530532522715854,
      "grad_norm": 4.824680805206299,
      "learning_rate": 2.9738438195838145e-05,
      "loss": 1.8696,
      "step": 30570
    },
    {
      "epoch": 0.059559742793826115,
      "grad_norm": 2.1868393421173096,
      "learning_rate": 2.9738182197591038e-05,
      "loss": 1.8737,
      "step": 30585
    },
    {
      "epoch": 0.05958895306493638,
      "grad_norm": 4.172234535217285,
      "learning_rate": 2.973792607523163e-05,
      "loss": 1.8148,
      "step": 30600
    },
    {
      "epoch": 0.059618163336046644,
      "grad_norm": 4.052221775054932,
      "learning_rate": 2.9737669828762074e-05,
      "loss": 1.8824,
      "step": 30615
    },
    {
      "epoch": 0.059647373607156905,
      "grad_norm": 3.6041007041931152,
      "learning_rate": 2.9737413458184522e-05,
      "loss": 1.8255,
      "step": 30630
    },
    {
      "epoch": 0.059676583878267166,
      "grad_norm": 4.06122350692749,
      "learning_rate": 2.9737156963501143e-05,
      "loss": 1.951,
      "step": 30645
    },
    {
      "epoch": 0.059705794149377434,
      "grad_norm": 2.3008453845977783,
      "learning_rate": 2.973690034471409e-05,
      "loss": 1.758,
      "step": 30660
    },
    {
      "epoch": 0.059735004420487695,
      "grad_norm": 3.7912003993988037,
      "learning_rate": 2.9736643601825532e-05,
      "loss": 1.8688,
      "step": 30675
    },
    {
      "epoch": 0.059764214691597956,
      "grad_norm": 3.7779648303985596,
      "learning_rate": 2.9736386734837616e-05,
      "loss": 1.7613,
      "step": 30690
    },
    {
      "epoch": 0.05979342496270822,
      "grad_norm": 4.632726669311523,
      "learning_rate": 2.9736129743752523e-05,
      "loss": 1.6212,
      "step": 30705
    },
    {
      "epoch": 0.059822635233818484,
      "grad_norm": 4.628555774688721,
      "learning_rate": 2.9735872628572404e-05,
      "loss": 1.8643,
      "step": 30720
    },
    {
      "epoch": 0.059851845504928745,
      "grad_norm": 3.8072681427001953,
      "learning_rate": 2.9735615389299434e-05,
      "loss": 1.7805,
      "step": 30735
    },
    {
      "epoch": 0.059881055776039006,
      "grad_norm": 3.9826717376708984,
      "learning_rate": 2.973535802593577e-05,
      "loss": 1.8594,
      "step": 30750
    },
    {
      "epoch": 0.059910266047149274,
      "grad_norm": 3.04962420463562,
      "learning_rate": 2.9735100538483587e-05,
      "loss": 1.9104,
      "step": 30765
    },
    {
      "epoch": 0.059939476318259535,
      "grad_norm": 2.8500261306762695,
      "learning_rate": 2.9734842926945048e-05,
      "loss": 1.717,
      "step": 30780
    },
    {
      "epoch": 0.059968686589369796,
      "grad_norm": 1.5368775129318237,
      "learning_rate": 2.9734585191322324e-05,
      "loss": 1.8905,
      "step": 30795
    },
    {
      "epoch": 0.059997896860480064,
      "grad_norm": 3.886584997177124,
      "learning_rate": 2.9734327331617588e-05,
      "loss": 1.804,
      "step": 30810
    },
    {
      "epoch": 0.060027107131590325,
      "grad_norm": 3.792663812637329,
      "learning_rate": 2.9734069347833005e-05,
      "loss": 1.9667,
      "step": 30825
    },
    {
      "epoch": 0.060056317402700586,
      "grad_norm": 3.7504067420959473,
      "learning_rate": 2.9733811239970756e-05,
      "loss": 1.9053,
      "step": 30840
    },
    {
      "epoch": 0.06008552767381085,
      "grad_norm": 2.617370367050171,
      "learning_rate": 2.973355300803301e-05,
      "loss": 1.8853,
      "step": 30855
    },
    {
      "epoch": 0.060114737944921115,
      "grad_norm": 3.2594456672668457,
      "learning_rate": 2.973329465202194e-05,
      "loss": 1.7045,
      "step": 30870
    },
    {
      "epoch": 0.060143948216031376,
      "grad_norm": 1.8166978359222412,
      "learning_rate": 2.9733036171939723e-05,
      "loss": 1.8431,
      "step": 30885
    },
    {
      "epoch": 0.06017315848714164,
      "grad_norm": 2.9280359745025635,
      "learning_rate": 2.973277756778854e-05,
      "loss": 1.9088,
      "step": 30900
    },
    {
      "epoch": 0.060202368758251905,
      "grad_norm": 4.446992874145508,
      "learning_rate": 2.973251883957056e-05,
      "loss": 1.7744,
      "step": 30915
    },
    {
      "epoch": 0.060231579029362166,
      "grad_norm": 4.279955863952637,
      "learning_rate": 2.9732259987287967e-05,
      "loss": 1.8309,
      "step": 30930
    },
    {
      "epoch": 0.06026078930047243,
      "grad_norm": 2.3727381229400635,
      "learning_rate": 2.9732001010942946e-05,
      "loss": 1.9247,
      "step": 30945
    },
    {
      "epoch": 0.06028999957158269,
      "grad_norm": 2.6014645099639893,
      "learning_rate": 2.9731741910537668e-05,
      "loss": 1.7653,
      "step": 30960
    },
    {
      "epoch": 0.060319209842692956,
      "grad_norm": 2.387563467025757,
      "learning_rate": 2.973148268607432e-05,
      "loss": 1.7814,
      "step": 30975
    },
    {
      "epoch": 0.06034842011380322,
      "grad_norm": 3.4683475494384766,
      "learning_rate": 2.9731223337555088e-05,
      "loss": 1.8522,
      "step": 30990
    },
    {
      "epoch": 0.06037763038491348,
      "grad_norm": 2.352069854736328,
      "learning_rate": 2.973096386498215e-05,
      "loss": 2.0015,
      "step": 31005
    },
    {
      "epoch": 0.06040684065602374,
      "grad_norm": 3.2257015705108643,
      "learning_rate": 2.9730704268357694e-05,
      "loss": 1.8176,
      "step": 31020
    },
    {
      "epoch": 0.06043605092713401,
      "grad_norm": 2.4939286708831787,
      "learning_rate": 2.97304445476839e-05,
      "loss": 1.7388,
      "step": 31035
    },
    {
      "epoch": 0.06046526119824427,
      "grad_norm": 3.216533660888672,
      "learning_rate": 2.9730184702962968e-05,
      "loss": 1.859,
      "step": 31050
    },
    {
      "epoch": 0.06049447146935453,
      "grad_norm": 3.4212570190429688,
      "learning_rate": 2.9729924734197077e-05,
      "loss": 1.7748,
      "step": 31065
    },
    {
      "epoch": 0.060523681740464796,
      "grad_norm": 3.6867172718048096,
      "learning_rate": 2.9729664641388417e-05,
      "loss": 2.0345,
      "step": 31080
    },
    {
      "epoch": 0.06055289201157506,
      "grad_norm": 3.243898630142212,
      "learning_rate": 2.9729404424539183e-05,
      "loss": 1.9114,
      "step": 31095
    },
    {
      "epoch": 0.06058210228268532,
      "grad_norm": 2.8866801261901855,
      "learning_rate": 2.9729144083651554e-05,
      "loss": 1.9833,
      "step": 31110
    },
    {
      "epoch": 0.06061131255379558,
      "grad_norm": 3.0051145553588867,
      "learning_rate": 2.9728883618727738e-05,
      "loss": 2.026,
      "step": 31125
    },
    {
      "epoch": 0.06064052282490585,
      "grad_norm": 3.0416595935821533,
      "learning_rate": 2.972862302976992e-05,
      "loss": 1.8541,
      "step": 31140
    },
    {
      "epoch": 0.06066973309601611,
      "grad_norm": 2.835458517074585,
      "learning_rate": 2.9728362316780294e-05,
      "loss": 1.8636,
      "step": 31155
    },
    {
      "epoch": 0.06069894336712637,
      "grad_norm": 2.40006947517395,
      "learning_rate": 2.972810147976106e-05,
      "loss": 1.9535,
      "step": 31170
    },
    {
      "epoch": 0.06072815363823664,
      "grad_norm": 2.7326056957244873,
      "learning_rate": 2.9727840518714415e-05,
      "loss": 1.7814,
      "step": 31185
    },
    {
      "epoch": 0.0607573639093469,
      "grad_norm": 4.403411388397217,
      "learning_rate": 2.972757943364255e-05,
      "loss": 1.7441,
      "step": 31200
    },
    {
      "epoch": 0.06078657418045716,
      "grad_norm": 2.259716272354126,
      "learning_rate": 2.9727318224547667e-05,
      "loss": 1.8447,
      "step": 31215
    },
    {
      "epoch": 0.06081578445156742,
      "grad_norm": 2.397702217102051,
      "learning_rate": 2.9727056891431966e-05,
      "loss": 1.7677,
      "step": 31230
    },
    {
      "epoch": 0.06084499472267769,
      "grad_norm": 3.840811014175415,
      "learning_rate": 2.9726795434297646e-05,
      "loss": 1.7271,
      "step": 31245
    },
    {
      "epoch": 0.06087420499378795,
      "grad_norm": 2.695345163345337,
      "learning_rate": 2.9726533853146913e-05,
      "loss": 1.9575,
      "step": 31260
    },
    {
      "epoch": 0.06090341526489821,
      "grad_norm": 2.330676555633545,
      "learning_rate": 2.972627214798197e-05,
      "loss": 1.9488,
      "step": 31275
    },
    {
      "epoch": 0.06093262553600848,
      "grad_norm": 3.987534999847412,
      "learning_rate": 2.9726010318805014e-05,
      "loss": 1.8628,
      "step": 31290
    },
    {
      "epoch": 0.06096183580711874,
      "grad_norm": 3.307753086090088,
      "learning_rate": 2.9725748365618252e-05,
      "loss": 1.9931,
      "step": 31305
    },
    {
      "epoch": 0.060991046078229,
      "grad_norm": 4.703588485717773,
      "learning_rate": 2.9725486288423894e-05,
      "loss": 1.8416,
      "step": 31320
    },
    {
      "epoch": 0.06102025634933926,
      "grad_norm": 2.01497220993042,
      "learning_rate": 2.9725224087224146e-05,
      "loss": 1.897,
      "step": 31335
    },
    {
      "epoch": 0.06104946662044953,
      "grad_norm": 2.9679431915283203,
      "learning_rate": 2.9724961762021215e-05,
      "loss": 1.951,
      "step": 31350
    },
    {
      "epoch": 0.06107867689155979,
      "grad_norm": 2.5453622341156006,
      "learning_rate": 2.972469931281731e-05,
      "loss": 1.8823,
      "step": 31365
    },
    {
      "epoch": 0.06110788716267005,
      "grad_norm": 4.135504245758057,
      "learning_rate": 2.9724436739614643e-05,
      "loss": 1.9061,
      "step": 31380
    },
    {
      "epoch": 0.06113709743378032,
      "grad_norm": 1.99344003200531,
      "learning_rate": 2.9724174042415417e-05,
      "loss": 1.986,
      "step": 31395
    },
    {
      "epoch": 0.06116630770489058,
      "grad_norm": 4.097934722900391,
      "learning_rate": 2.9723911221221857e-05,
      "loss": 1.8222,
      "step": 31410
    },
    {
      "epoch": 0.06119551797600084,
      "grad_norm": 4.836818695068359,
      "learning_rate": 2.9723648276036165e-05,
      "loss": 1.9038,
      "step": 31425
    },
    {
      "epoch": 0.0612247282471111,
      "grad_norm": 2.4198572635650635,
      "learning_rate": 2.972338520686056e-05,
      "loss": 1.9228,
      "step": 31440
    },
    {
      "epoch": 0.06125393851822137,
      "grad_norm": 4.546353816986084,
      "learning_rate": 2.9723122013697265e-05,
      "loss": 1.6837,
      "step": 31455
    },
    {
      "epoch": 0.06128314878933163,
      "grad_norm": 3.9125819206237793,
      "learning_rate": 2.972285869654848e-05,
      "loss": 1.9057,
      "step": 31470
    },
    {
      "epoch": 0.06131235906044189,
      "grad_norm": 2.7409276962280273,
      "learning_rate": 2.9722595255416435e-05,
      "loss": 1.6805,
      "step": 31485
    },
    {
      "epoch": 0.06134156933155216,
      "grad_norm": 3.4497761726379395,
      "learning_rate": 2.9722331690303344e-05,
      "loss": 1.7006,
      "step": 31500
    },
    {
      "epoch": 0.06137077960266242,
      "grad_norm": 3.8101327419281006,
      "learning_rate": 2.9722068001211427e-05,
      "loss": 1.8445,
      "step": 31515
    },
    {
      "epoch": 0.06139998987377268,
      "grad_norm": 4.296467304229736,
      "learning_rate": 2.9721804188142906e-05,
      "loss": 1.9453,
      "step": 31530
    },
    {
      "epoch": 0.06142920014488294,
      "grad_norm": 4.05743408203125,
      "learning_rate": 2.9721540251100003e-05,
      "loss": 1.6627,
      "step": 31545
    },
    {
      "epoch": 0.06145841041599321,
      "grad_norm": 1.896415114402771,
      "learning_rate": 2.9721276190084934e-05,
      "loss": 2.046,
      "step": 31560
    },
    {
      "epoch": 0.06148762068710347,
      "grad_norm": 3.3542966842651367,
      "learning_rate": 2.9721012005099933e-05,
      "loss": 1.7947,
      "step": 31575
    },
    {
      "epoch": 0.06151683095821373,
      "grad_norm": 2.018230676651001,
      "learning_rate": 2.9720747696147214e-05,
      "loss": 1.8102,
      "step": 31590
    },
    {
      "epoch": 0.06154604122932399,
      "grad_norm": 2.8734261989593506,
      "learning_rate": 2.9720483263229012e-05,
      "loss": 1.9102,
      "step": 31605
    },
    {
      "epoch": 0.06157525150043426,
      "grad_norm": 2.8697266578674316,
      "learning_rate": 2.972021870634755e-05,
      "loss": 1.9179,
      "step": 31620
    },
    {
      "epoch": 0.06160446177154452,
      "grad_norm": 2.5872962474823,
      "learning_rate": 2.9719954025505054e-05,
      "loss": 1.8555,
      "step": 31635
    },
    {
      "epoch": 0.06163367204265478,
      "grad_norm": 3.8212106227874756,
      "learning_rate": 2.9719689220703758e-05,
      "loss": 1.7516,
      "step": 31650
    },
    {
      "epoch": 0.06166288231376505,
      "grad_norm": 2.7884206771850586,
      "learning_rate": 2.9719424291945885e-05,
      "loss": 1.8213,
      "step": 31665
    },
    {
      "epoch": 0.06169209258487531,
      "grad_norm": 2.5221076011657715,
      "learning_rate": 2.971915923923367e-05,
      "loss": 1.6586,
      "step": 31680
    },
    {
      "epoch": 0.06172130285598557,
      "grad_norm": 2.4005963802337646,
      "learning_rate": 2.971889406256935e-05,
      "loss": 2.0582,
      "step": 31695
    },
    {
      "epoch": 0.061750513127095834,
      "grad_norm": 2.02565860748291,
      "learning_rate": 2.9718628761955146e-05,
      "loss": 1.8941,
      "step": 31710
    },
    {
      "epoch": 0.0617797233982061,
      "grad_norm": 3.2867822647094727,
      "learning_rate": 2.9718363337393303e-05,
      "loss": 1.7149,
      "step": 31725
    },
    {
      "epoch": 0.06180893366931636,
      "grad_norm": 2.4237873554229736,
      "learning_rate": 2.9718097788886054e-05,
      "loss": 1.8685,
      "step": 31740
    },
    {
      "epoch": 0.061838143940426624,
      "grad_norm": 2.000887155532837,
      "learning_rate": 2.9717832116435632e-05,
      "loss": 1.7489,
      "step": 31755
    },
    {
      "epoch": 0.06186735421153689,
      "grad_norm": 2.2382023334503174,
      "learning_rate": 2.9717566320044276e-05,
      "loss": 1.8845,
      "step": 31770
    },
    {
      "epoch": 0.06189656448264715,
      "grad_norm": 3.9158990383148193,
      "learning_rate": 2.9717300399714222e-05,
      "loss": 1.689,
      "step": 31785
    },
    {
      "epoch": 0.061925774753757414,
      "grad_norm": 4.004560470581055,
      "learning_rate": 2.9717034355447713e-05,
      "loss": 1.9279,
      "step": 31800
    },
    {
      "epoch": 0.061954985024867675,
      "grad_norm": 2.9310123920440674,
      "learning_rate": 2.9716768187246986e-05,
      "loss": 1.771,
      "step": 31815
    },
    {
      "epoch": 0.06198419529597794,
      "grad_norm": 2.5769150257110596,
      "learning_rate": 2.9716501895114287e-05,
      "loss": 1.9365,
      "step": 31830
    },
    {
      "epoch": 0.0620134055670882,
      "grad_norm": 3.767500162124634,
      "learning_rate": 2.9716235479051858e-05,
      "loss": 1.8358,
      "step": 31845
    },
    {
      "epoch": 0.062042615838198464,
      "grad_norm": 4.410051345825195,
      "learning_rate": 2.9715968939061932e-05,
      "loss": 1.8382,
      "step": 31860
    },
    {
      "epoch": 0.06207182610930873,
      "grad_norm": 2.5900702476501465,
      "learning_rate": 2.9715702275146767e-05,
      "loss": 1.8717,
      "step": 31875
    },
    {
      "epoch": 0.06210103638041899,
      "grad_norm": 3.7013213634490967,
      "learning_rate": 2.9715435487308608e-05,
      "loss": 1.8794,
      "step": 31890
    },
    {
      "epoch": 0.062130246651529254,
      "grad_norm": 2.48606276512146,
      "learning_rate": 2.9715168575549688e-05,
      "loss": 1.9964,
      "step": 31905
    },
    {
      "epoch": 0.062159456922639515,
      "grad_norm": 2.358415365219116,
      "learning_rate": 2.9714901539872267e-05,
      "loss": 1.902,
      "step": 31920
    },
    {
      "epoch": 0.06218866719374978,
      "grad_norm": 3.140083074569702,
      "learning_rate": 2.9714634380278594e-05,
      "loss": 1.8696,
      "step": 31935
    },
    {
      "epoch": 0.062217877464860044,
      "grad_norm": 3.2135777473449707,
      "learning_rate": 2.9714367096770913e-05,
      "loss": 2.0002,
      "step": 31950
    },
    {
      "epoch": 0.062247087735970305,
      "grad_norm": 3.2304131984710693,
      "learning_rate": 2.9714099689351477e-05,
      "loss": 1.8721,
      "step": 31965
    },
    {
      "epoch": 0.06227629800708057,
      "grad_norm": 2.206209659576416,
      "learning_rate": 2.9713832158022535e-05,
      "loss": 1.9788,
      "step": 31980
    },
    {
      "epoch": 0.062305508278190834,
      "grad_norm": 3.055828332901001,
      "learning_rate": 2.9713564502786348e-05,
      "loss": 1.7976,
      "step": 31995
    },
    {
      "epoch": 0.062334718549301095,
      "grad_norm": 3.891845226287842,
      "learning_rate": 2.9713296723645165e-05,
      "loss": 1.7772,
      "step": 32010
    },
    {
      "epoch": 0.062363928820411356,
      "grad_norm": 1.8532415628433228,
      "learning_rate": 2.9713028820601238e-05,
      "loss": 1.9817,
      "step": 32025
    },
    {
      "epoch": 0.062393139091521624,
      "grad_norm": 1.650211215019226,
      "learning_rate": 2.9712760793656826e-05,
      "loss": 2.1243,
      "step": 32040
    },
    {
      "epoch": 0.062422349362631885,
      "grad_norm": 2.663621187210083,
      "learning_rate": 2.971249264281419e-05,
      "loss": 2.0554,
      "step": 32055
    },
    {
      "epoch": 0.062451559633742146,
      "grad_norm": 3.2437026500701904,
      "learning_rate": 2.9712224368075578e-05,
      "loss": 1.765,
      "step": 32070
    },
    {
      "epoch": 0.06248076990485241,
      "grad_norm": 4.013802528381348,
      "learning_rate": 2.971195596944326e-05,
      "loss": 1.8195,
      "step": 32085
    },
    {
      "epoch": 0.06250998017596267,
      "grad_norm": 4.0295610427856445,
      "learning_rate": 2.971168744691949e-05,
      "loss": 2.036,
      "step": 32100
    },
    {
      "epoch": 0.06253919044707294,
      "grad_norm": 2.830064296722412,
      "learning_rate": 2.9711418800506533e-05,
      "loss": 1.8499,
      "step": 32115
    },
    {
      "epoch": 0.0625684007181832,
      "grad_norm": 3.896538019180298,
      "learning_rate": 2.9711150030206643e-05,
      "loss": 2.0089,
      "step": 32130
    },
    {
      "epoch": 0.06259761098929346,
      "grad_norm": 3.6802754402160645,
      "learning_rate": 2.97108811360221e-05,
      "loss": 1.9039,
      "step": 32145
    },
    {
      "epoch": 0.06262682126040373,
      "grad_norm": 2.8547956943511963,
      "learning_rate": 2.9710612117955148e-05,
      "loss": 1.9065,
      "step": 32160
    },
    {
      "epoch": 0.06265603153151399,
      "grad_norm": 2.0819807052612305,
      "learning_rate": 2.9710342976008066e-05,
      "loss": 1.7786,
      "step": 32175
    },
    {
      "epoch": 0.06268524180262425,
      "grad_norm": 2.173609972000122,
      "learning_rate": 2.9710073710183118e-05,
      "loss": 1.8011,
      "step": 32190
    },
    {
      "epoch": 0.06271445207373451,
      "grad_norm": 2.2233803272247314,
      "learning_rate": 2.970980432048257e-05,
      "loss": 1.8308,
      "step": 32205
    },
    {
      "epoch": 0.06274366234484478,
      "grad_norm": 2.227229356765747,
      "learning_rate": 2.9709534806908692e-05,
      "loss": 1.9227,
      "step": 32220
    },
    {
      "epoch": 0.06277287261595504,
      "grad_norm": 3.8464114665985107,
      "learning_rate": 2.970926516946375e-05,
      "loss": 1.9808,
      "step": 32235
    },
    {
      "epoch": 0.0628020828870653,
      "grad_norm": 4.511869430541992,
      "learning_rate": 2.9708995408150018e-05,
      "loss": 1.9856,
      "step": 32250
    },
    {
      "epoch": 0.06283129315817557,
      "grad_norm": 2.0209896564483643,
      "learning_rate": 2.9708725522969767e-05,
      "loss": 1.8766,
      "step": 32265
    },
    {
      "epoch": 0.06286050342928583,
      "grad_norm": 4.369567394256592,
      "learning_rate": 2.9708455513925273e-05,
      "loss": 1.8103,
      "step": 32280
    },
    {
      "epoch": 0.06288971370039609,
      "grad_norm": 4.502676486968994,
      "learning_rate": 2.97081853810188e-05,
      "loss": 1.8489,
      "step": 32295
    },
    {
      "epoch": 0.06291892397150635,
      "grad_norm": 2.288998603820801,
      "learning_rate": 2.9707915124252637e-05,
      "loss": 1.9542,
      "step": 32310
    },
    {
      "epoch": 0.06294813424261662,
      "grad_norm": 2.6359355449676514,
      "learning_rate": 2.9707644743629046e-05,
      "loss": 1.7883,
      "step": 32325
    },
    {
      "epoch": 0.06297734451372689,
      "grad_norm": 4.356678485870361,
      "learning_rate": 2.970737423915031e-05,
      "loss": 1.8937,
      "step": 32340
    },
    {
      "epoch": 0.06300655478483715,
      "grad_norm": 3.2190353870391846,
      "learning_rate": 2.970710361081871e-05,
      "loss": 1.9758,
      "step": 32355
    },
    {
      "epoch": 0.06303576505594741,
      "grad_norm": 2.522303819656372,
      "learning_rate": 2.970683285863652e-05,
      "loss": 1.7315,
      "step": 32370
    },
    {
      "epoch": 0.06306497532705767,
      "grad_norm": 2.500927448272705,
      "learning_rate": 2.9706561982606023e-05,
      "loss": 1.9004,
      "step": 32385
    },
    {
      "epoch": 0.06309418559816793,
      "grad_norm": 3.656970262527466,
      "learning_rate": 2.97062909827295e-05,
      "loss": 1.9246,
      "step": 32400
    },
    {
      "epoch": 0.06312339586927819,
      "grad_norm": 4.1837158203125,
      "learning_rate": 2.970601985900923e-05,
      "loss": 1.6453,
      "step": 32415
    },
    {
      "epoch": 0.06315260614038845,
      "grad_norm": 5.141902446746826,
      "learning_rate": 2.9705748611447498e-05,
      "loss": 1.7215,
      "step": 32430
    },
    {
      "epoch": 0.06318181641149873,
      "grad_norm": 5.384919166564941,
      "learning_rate": 2.9705477240046595e-05,
      "loss": 2.0638,
      "step": 32445
    },
    {
      "epoch": 0.06321102668260899,
      "grad_norm": 2.812014579772949,
      "learning_rate": 2.9705205744808795e-05,
      "loss": 1.8133,
      "step": 32460
    },
    {
      "epoch": 0.06324023695371925,
      "grad_norm": 4.618744850158691,
      "learning_rate": 2.970493412573639e-05,
      "loss": 1.7221,
      "step": 32475
    },
    {
      "epoch": 0.06326944722482951,
      "grad_norm": 2.0948140621185303,
      "learning_rate": 2.9704662382831665e-05,
      "loss": 1.7292,
      "step": 32490
    },
    {
      "epoch": 0.06329865749593977,
      "grad_norm": 2.2021284103393555,
      "learning_rate": 2.970439051609691e-05,
      "loss": 1.6685,
      "step": 32505
    },
    {
      "epoch": 0.06332786776705003,
      "grad_norm": 4.773664474487305,
      "learning_rate": 2.9704118525534414e-05,
      "loss": 1.8922,
      "step": 32520
    },
    {
      "epoch": 0.06335707803816029,
      "grad_norm": 2.716444253921509,
      "learning_rate": 2.970384641114647e-05,
      "loss": 1.816,
      "step": 32535
    },
    {
      "epoch": 0.06338628830927057,
      "grad_norm": 4.086103916168213,
      "learning_rate": 2.9703574172935366e-05,
      "loss": 1.8084,
      "step": 32550
    },
    {
      "epoch": 0.06341549858038083,
      "grad_norm": 2.4854936599731445,
      "learning_rate": 2.9703301810903397e-05,
      "loss": 1.7871,
      "step": 32565
    },
    {
      "epoch": 0.06344470885149109,
      "grad_norm": 1.8130111694335938,
      "learning_rate": 2.9703029325052857e-05,
      "loss": 1.8795,
      "step": 32580
    },
    {
      "epoch": 0.06347391912260135,
      "grad_norm": 3.8686211109161377,
      "learning_rate": 2.970275671538604e-05,
      "loss": 1.9867,
      "step": 32595
    },
    {
      "epoch": 0.06350312939371161,
      "grad_norm": 2.5253164768218994,
      "learning_rate": 2.9702483981905233e-05,
      "loss": 1.875,
      "step": 32610
    },
    {
      "epoch": 0.06353233966482187,
      "grad_norm": 3.0097036361694336,
      "learning_rate": 2.9702211124612745e-05,
      "loss": 1.9521,
      "step": 32625
    },
    {
      "epoch": 0.06356154993593213,
      "grad_norm": 3.3433878421783447,
      "learning_rate": 2.9701938143510873e-05,
      "loss": 1.806,
      "step": 32640
    },
    {
      "epoch": 0.06359076020704241,
      "grad_norm": 1.4688777923583984,
      "learning_rate": 2.970166503860191e-05,
      "loss": 1.8248,
      "step": 32655
    },
    {
      "epoch": 0.06361997047815267,
      "grad_norm": 2.365694522857666,
      "learning_rate": 2.9701391809888156e-05,
      "loss": 1.7152,
      "step": 32670
    },
    {
      "epoch": 0.06364918074926293,
      "grad_norm": 2.6821203231811523,
      "learning_rate": 2.9701118457371915e-05,
      "loss": 1.858,
      "step": 32685
    },
    {
      "epoch": 0.06367839102037319,
      "grad_norm": 1.842124342918396,
      "learning_rate": 2.9700844981055486e-05,
      "loss": 1.7689,
      "step": 32700
    },
    {
      "epoch": 0.06370760129148345,
      "grad_norm": 2.2430474758148193,
      "learning_rate": 2.9700571380941178e-05,
      "loss": 2.0056,
      "step": 32715
    },
    {
      "epoch": 0.06373681156259371,
      "grad_norm": 3.1308460235595703,
      "learning_rate": 2.970029765703129e-05,
      "loss": 1.9163,
      "step": 32730
    },
    {
      "epoch": 0.06376602183370397,
      "grad_norm": 2.6456706523895264,
      "learning_rate": 2.9700023809328122e-05,
      "loss": 1.8002,
      "step": 32745
    },
    {
      "epoch": 0.06379523210481425,
      "grad_norm": 3.659024953842163,
      "learning_rate": 2.969974983783399e-05,
      "loss": 1.9222,
      "step": 32760
    },
    {
      "epoch": 0.06382444237592451,
      "grad_norm": 2.3271381855010986,
      "learning_rate": 2.96994757425512e-05,
      "loss": 1.7778,
      "step": 32775
    },
    {
      "epoch": 0.06385365264703477,
      "grad_norm": 3.411533832550049,
      "learning_rate": 2.9699201523482057e-05,
      "loss": 1.778,
      "step": 32790
    },
    {
      "epoch": 0.06388286291814503,
      "grad_norm": 2.649052143096924,
      "learning_rate": 2.9698927180628866e-05,
      "loss": 1.8199,
      "step": 32805
    },
    {
      "epoch": 0.06391207318925529,
      "grad_norm": 2.8163363933563232,
      "learning_rate": 2.9698652713993947e-05,
      "loss": 1.8579,
      "step": 32820
    },
    {
      "epoch": 0.06394128346036555,
      "grad_norm": 3.319955348968506,
      "learning_rate": 2.9698378123579603e-05,
      "loss": 1.9385,
      "step": 32835
    },
    {
      "epoch": 0.06397049373147581,
      "grad_norm": 2.217761754989624,
      "learning_rate": 2.969810340938815e-05,
      "loss": 2.0007,
      "step": 32850
    },
    {
      "epoch": 0.06399970400258609,
      "grad_norm": 4.801873683929443,
      "learning_rate": 2.96978285714219e-05,
      "loss": 1.9876,
      "step": 32865
    },
    {
      "epoch": 0.06402891427369635,
      "grad_norm": 3.5146069526672363,
      "learning_rate": 2.969755360968317e-05,
      "loss": 1.8267,
      "step": 32880
    },
    {
      "epoch": 0.06405812454480661,
      "grad_norm": 3.407545328140259,
      "learning_rate": 2.9697278524174275e-05,
      "loss": 1.743,
      "step": 32895
    },
    {
      "epoch": 0.06408733481591687,
      "grad_norm": 1.410946011543274,
      "learning_rate": 2.969700331489753e-05,
      "loss": 1.9281,
      "step": 32910
    },
    {
      "epoch": 0.06411654508702713,
      "grad_norm": 2.602207899093628,
      "learning_rate": 2.9696727981855253e-05,
      "loss": 1.8532,
      "step": 32925
    },
    {
      "epoch": 0.0641457553581374,
      "grad_norm": 4.358025550842285,
      "learning_rate": 2.969645252504976e-05,
      "loss": 1.7457,
      "step": 32940
    },
    {
      "epoch": 0.06417496562924765,
      "grad_norm": 3.8209457397460938,
      "learning_rate": 2.9696176944483373e-05,
      "loss": 2.0736,
      "step": 32955
    },
    {
      "epoch": 0.06420417590035793,
      "grad_norm": 3.317567825317383,
      "learning_rate": 2.9695901240158415e-05,
      "loss": 1.9322,
      "step": 32970
    },
    {
      "epoch": 0.06423338617146819,
      "grad_norm": 4.668582439422607,
      "learning_rate": 2.9695625412077208e-05,
      "loss": 2.0152,
      "step": 32985
    },
    {
      "epoch": 0.06426259644257845,
      "grad_norm": 2.6789300441741943,
      "learning_rate": 2.969534946024207e-05,
      "loss": 1.7129,
      "step": 33000
    },
    {
      "epoch": 0.06429180671368871,
      "grad_norm": 2.073744297027588,
      "learning_rate": 2.9695073384655326e-05,
      "loss": 1.6952,
      "step": 33015
    },
    {
      "epoch": 0.06432101698479897,
      "grad_norm": 3.3017473220825195,
      "learning_rate": 2.9694797185319304e-05,
      "loss": 1.8158,
      "step": 33030
    },
    {
      "epoch": 0.06435022725590923,
      "grad_norm": 4.441155433654785,
      "learning_rate": 2.969452086223633e-05,
      "loss": 1.9204,
      "step": 33045
    },
    {
      "epoch": 0.0643794375270195,
      "grad_norm": 2.9764716625213623,
      "learning_rate": 2.9694244415408727e-05,
      "loss": 1.8952,
      "step": 33060
    },
    {
      "epoch": 0.06440864779812977,
      "grad_norm": 4.1646575927734375,
      "learning_rate": 2.9693967844838827e-05,
      "loss": 1.7692,
      "step": 33075
    },
    {
      "epoch": 0.06443785806924003,
      "grad_norm": 4.056210994720459,
      "learning_rate": 2.9693691150528955e-05,
      "loss": 1.855,
      "step": 33090
    },
    {
      "epoch": 0.06446706834035029,
      "grad_norm": 2.6098761558532715,
      "learning_rate": 2.9693414332481443e-05,
      "loss": 1.8284,
      "step": 33105
    },
    {
      "epoch": 0.06449627861146055,
      "grad_norm": 4.3388166427612305,
      "learning_rate": 2.9693137390698622e-05,
      "loss": 1.8155,
      "step": 33120
    },
    {
      "epoch": 0.06452548888257081,
      "grad_norm": 3.3735921382904053,
      "learning_rate": 2.9692860325182825e-05,
      "loss": 1.7765,
      "step": 33135
    },
    {
      "epoch": 0.06455469915368107,
      "grad_norm": 3.9341137409210205,
      "learning_rate": 2.9692583135936385e-05,
      "loss": 1.7622,
      "step": 33150
    },
    {
      "epoch": 0.06458390942479134,
      "grad_norm": 5.620511531829834,
      "learning_rate": 2.9692305822961637e-05,
      "loss": 1.7926,
      "step": 33165
    },
    {
      "epoch": 0.06461311969590161,
      "grad_norm": 1.8949609994888306,
      "learning_rate": 2.9692028386260915e-05,
      "loss": 1.9158,
      "step": 33180
    },
    {
      "epoch": 0.06464232996701187,
      "grad_norm": 2.183438539505005,
      "learning_rate": 2.9691750825836557e-05,
      "loss": 1.9586,
      "step": 33195
    },
    {
      "epoch": 0.06467154023812213,
      "grad_norm": 2.05364727973938,
      "learning_rate": 2.9691473141690894e-05,
      "loss": 1.8952,
      "step": 33210
    },
    {
      "epoch": 0.0647007505092324,
      "grad_norm": 2.479680299758911,
      "learning_rate": 2.9691195333826277e-05,
      "loss": 1.7908,
      "step": 33225
    },
    {
      "epoch": 0.06472996078034265,
      "grad_norm": 2.0384159088134766,
      "learning_rate": 2.9690917402245034e-05,
      "loss": 1.8231,
      "step": 33240
    },
    {
      "epoch": 0.06475917105145292,
      "grad_norm": 2.3003928661346436,
      "learning_rate": 2.9690639346949503e-05,
      "loss": 1.6675,
      "step": 33255
    },
    {
      "epoch": 0.06478838132256318,
      "grad_norm": 2.9308221340179443,
      "learning_rate": 2.9690361167942042e-05,
      "loss": 1.9136,
      "step": 33270
    },
    {
      "epoch": 0.06481759159367345,
      "grad_norm": 2.3866639137268066,
      "learning_rate": 2.9690082865224975e-05,
      "loss": 1.8992,
      "step": 33285
    },
    {
      "epoch": 0.06484680186478371,
      "grad_norm": 4.399453163146973,
      "learning_rate": 2.968980443880066e-05,
      "loss": 1.9369,
      "step": 33300
    },
    {
      "epoch": 0.06487601213589397,
      "grad_norm": 3.06842041015625,
      "learning_rate": 2.9689525888671436e-05,
      "loss": 1.9624,
      "step": 33315
    },
    {
      "epoch": 0.06490522240700423,
      "grad_norm": 1.9636744260787964,
      "learning_rate": 2.9689247214839647e-05,
      "loss": 1.7785,
      "step": 33330
    },
    {
      "epoch": 0.0649344326781145,
      "grad_norm": 3.882155418395996,
      "learning_rate": 2.968896841730764e-05,
      "loss": 1.8328,
      "step": 33345
    },
    {
      "epoch": 0.06496364294922476,
      "grad_norm": 3.773200511932373,
      "learning_rate": 2.9688689496077764e-05,
      "loss": 1.9322,
      "step": 33360
    },
    {
      "epoch": 0.06499285322033502,
      "grad_norm": 2.331101417541504,
      "learning_rate": 2.968841045115237e-05,
      "loss": 1.7726,
      "step": 33375
    },
    {
      "epoch": 0.06502206349144529,
      "grad_norm": 3.3869986534118652,
      "learning_rate": 2.9688131282533802e-05,
      "loss": 1.9489,
      "step": 33390
    },
    {
      "epoch": 0.06505127376255555,
      "grad_norm": 2.627638578414917,
      "learning_rate": 2.968785199022442e-05,
      "loss": 1.7626,
      "step": 33405
    },
    {
      "epoch": 0.06508048403366581,
      "grad_norm": 2.031043291091919,
      "learning_rate": 2.9687572574226566e-05,
      "loss": 1.8948,
      "step": 33420
    },
    {
      "epoch": 0.06510969430477608,
      "grad_norm": 1.6995463371276855,
      "learning_rate": 2.96872930345426e-05,
      "loss": 1.8642,
      "step": 33435
    },
    {
      "epoch": 0.06513890457588634,
      "grad_norm": 3.292022943496704,
      "learning_rate": 2.9687013371174873e-05,
      "loss": 1.8344,
      "step": 33450
    },
    {
      "epoch": 0.0651681148469966,
      "grad_norm": 2.9283857345581055,
      "learning_rate": 2.968673358412574e-05,
      "loss": 2.0218,
      "step": 33465
    },
    {
      "epoch": 0.06519732511810686,
      "grad_norm": 3.2172300815582275,
      "learning_rate": 2.968645367339756e-05,
      "loss": 1.8049,
      "step": 33480
    },
    {
      "epoch": 0.06522653538921712,
      "grad_norm": 2.1529486179351807,
      "learning_rate": 2.9686173638992687e-05,
      "loss": 1.8755,
      "step": 33495
    },
    {
      "epoch": 0.0652557456603274,
      "grad_norm": 2.5349905490875244,
      "learning_rate": 2.9685893480913477e-05,
      "loss": 2.0173,
      "step": 33510
    },
    {
      "epoch": 0.06528495593143765,
      "grad_norm": 3.646724224090576,
      "learning_rate": 2.9685613199162296e-05,
      "loss": 1.9269,
      "step": 33525
    },
    {
      "epoch": 0.06531416620254792,
      "grad_norm": 3.3442795276641846,
      "learning_rate": 2.96853327937415e-05,
      "loss": 1.7363,
      "step": 33540
    },
    {
      "epoch": 0.06534337647365818,
      "grad_norm": 3.301544189453125,
      "learning_rate": 2.9685052264653452e-05,
      "loss": 1.9225,
      "step": 33555
    },
    {
      "epoch": 0.06537258674476844,
      "grad_norm": 1.919858455657959,
      "learning_rate": 2.9684771611900512e-05,
      "loss": 1.7969,
      "step": 33570
    },
    {
      "epoch": 0.0654017970158787,
      "grad_norm": 3.6086807250976562,
      "learning_rate": 2.9684490835485045e-05,
      "loss": 1.8075,
      "step": 33585
    },
    {
      "epoch": 0.06543100728698896,
      "grad_norm": 1.659013032913208,
      "learning_rate": 2.9684209935409418e-05,
      "loss": 1.7631,
      "step": 33600
    },
    {
      "epoch": 0.06546021755809923,
      "grad_norm": 4.616360187530518,
      "learning_rate": 2.968392891167599e-05,
      "loss": 1.817,
      "step": 33615
    },
    {
      "epoch": 0.0654894278292095,
      "grad_norm": 2.684124708175659,
      "learning_rate": 2.9683647764287136e-05,
      "loss": 1.8655,
      "step": 33630
    },
    {
      "epoch": 0.06551863810031976,
      "grad_norm": 2.5763068199157715,
      "learning_rate": 2.9683366493245213e-05,
      "loss": 1.851,
      "step": 33645
    },
    {
      "epoch": 0.06554784837143002,
      "grad_norm": 4.1813859939575195,
      "learning_rate": 2.96830850985526e-05,
      "loss": 1.8571,
      "step": 33660
    },
    {
      "epoch": 0.06557705864254028,
      "grad_norm": 3.5396945476531982,
      "learning_rate": 2.968280358021166e-05,
      "loss": 1.8414,
      "step": 33675
    },
    {
      "epoch": 0.06560626891365054,
      "grad_norm": 3.497781276702881,
      "learning_rate": 2.968252193822477e-05,
      "loss": 2.0295,
      "step": 33690
    },
    {
      "epoch": 0.0656354791847608,
      "grad_norm": 2.3261985778808594,
      "learning_rate": 2.9682240172594294e-05,
      "loss": 1.8092,
      "step": 33705
    },
    {
      "epoch": 0.06566468945587108,
      "grad_norm": 2.558701753616333,
      "learning_rate": 2.968195828332261e-05,
      "loss": 1.9174,
      "step": 33720
    },
    {
      "epoch": 0.06569389972698134,
      "grad_norm": 3.210392713546753,
      "learning_rate": 2.9681676270412092e-05,
      "loss": 1.8506,
      "step": 33735
    },
    {
      "epoch": 0.0657231099980916,
      "grad_norm": 2.7672812938690186,
      "learning_rate": 2.968139413386511e-05,
      "loss": 1.9361,
      "step": 33750
    },
    {
      "epoch": 0.06575232026920186,
      "grad_norm": 4.212212562561035,
      "learning_rate": 2.9681111873684046e-05,
      "loss": 1.7308,
      "step": 33765
    },
    {
      "epoch": 0.06578153054031212,
      "grad_norm": 1.9889777898788452,
      "learning_rate": 2.9680829489871274e-05,
      "loss": 1.8549,
      "step": 33780
    },
    {
      "epoch": 0.06581074081142238,
      "grad_norm": 1.7036499977111816,
      "learning_rate": 2.9680546982429166e-05,
      "loss": 1.877,
      "step": 33795
    },
    {
      "epoch": 0.06583995108253264,
      "grad_norm": 2.7053897380828857,
      "learning_rate": 2.9680264351360115e-05,
      "loss": 1.8077,
      "step": 33810
    },
    {
      "epoch": 0.06586916135364292,
      "grad_norm": 2.8420169353485107,
      "learning_rate": 2.967998159666649e-05,
      "loss": 1.8568,
      "step": 33825
    },
    {
      "epoch": 0.06589837162475318,
      "grad_norm": 2.4401204586029053,
      "learning_rate": 2.9679698718350673e-05,
      "loss": 1.8119,
      "step": 33840
    },
    {
      "epoch": 0.06592758189586344,
      "grad_norm": 3.5529043674468994,
      "learning_rate": 2.9679415716415053e-05,
      "loss": 1.755,
      "step": 33855
    },
    {
      "epoch": 0.0659567921669737,
      "grad_norm": 2.672156572341919,
      "learning_rate": 2.9679132590862004e-05,
      "loss": 1.9222,
      "step": 33870
    },
    {
      "epoch": 0.06598600243808396,
      "grad_norm": 1.9536501169204712,
      "learning_rate": 2.967884934169392e-05,
      "loss": 1.9098,
      "step": 33885
    },
    {
      "epoch": 0.06601521270919422,
      "grad_norm": 2.567133903503418,
      "learning_rate": 2.9678565968913177e-05,
      "loss": 1.924,
      "step": 33900
    },
    {
      "epoch": 0.06604442298030448,
      "grad_norm": 2.739780902862549,
      "learning_rate": 2.967828247252217e-05,
      "loss": 1.9591,
      "step": 33915
    },
    {
      "epoch": 0.06607363325141476,
      "grad_norm": 3.8824591636657715,
      "learning_rate": 2.9677998852523277e-05,
      "loss": 1.8305,
      "step": 33930
    },
    {
      "epoch": 0.06610284352252502,
      "grad_norm": 3.137244939804077,
      "learning_rate": 2.967771510891889e-05,
      "loss": 1.9673,
      "step": 33945
    },
    {
      "epoch": 0.06613205379363528,
      "grad_norm": 2.5533084869384766,
      "learning_rate": 2.9677431241711405e-05,
      "loss": 1.7838,
      "step": 33960
    },
    {
      "epoch": 0.06616126406474554,
      "grad_norm": 3.2840499877929688,
      "learning_rate": 2.9677147250903203e-05,
      "loss": 1.7223,
      "step": 33975
    },
    {
      "epoch": 0.0661904743358558,
      "grad_norm": 3.5756568908691406,
      "learning_rate": 2.9676863136496685e-05,
      "loss": 2.0505,
      "step": 33990
    },
    {
      "epoch": 0.06621968460696606,
      "grad_norm": 2.9339518547058105,
      "learning_rate": 2.967657889849423e-05,
      "loss": 1.8248,
      "step": 34005
    },
    {
      "epoch": 0.06624889487807632,
      "grad_norm": 2.587409496307373,
      "learning_rate": 2.9676294536898247e-05,
      "loss": 2.1007,
      "step": 34020
    },
    {
      "epoch": 0.0662781051491866,
      "grad_norm": 2.164442777633667,
      "learning_rate": 2.9676010051711123e-05,
      "loss": 1.6494,
      "step": 34035
    },
    {
      "epoch": 0.06630731542029686,
      "grad_norm": 2.022341012954712,
      "learning_rate": 2.9675725442935252e-05,
      "loss": 2.01,
      "step": 34050
    },
    {
      "epoch": 0.06633652569140712,
      "grad_norm": 2.1963977813720703,
      "learning_rate": 2.9675440710573036e-05,
      "loss": 1.7275,
      "step": 34065
    },
    {
      "epoch": 0.06636573596251738,
      "grad_norm": 2.7345070838928223,
      "learning_rate": 2.967515585462687e-05,
      "loss": 1.6542,
      "step": 34080
    },
    {
      "epoch": 0.06639494623362764,
      "grad_norm": 2.9531948566436768,
      "learning_rate": 2.9674870875099144e-05,
      "loss": 1.9653,
      "step": 34095
    },
    {
      "epoch": 0.0664241565047379,
      "grad_norm": 2.839843988418579,
      "learning_rate": 2.9674585771992277e-05,
      "loss": 1.8574,
      "step": 34110
    },
    {
      "epoch": 0.06645336677584816,
      "grad_norm": 2.976874351501465,
      "learning_rate": 2.967430054530865e-05,
      "loss": 2.0622,
      "step": 34125
    },
    {
      "epoch": 0.06648257704695844,
      "grad_norm": 2.1137330532073975,
      "learning_rate": 2.967401519505068e-05,
      "loss": 1.8188,
      "step": 34140
    },
    {
      "epoch": 0.0665117873180687,
      "grad_norm": 2.7060563564300537,
      "learning_rate": 2.9673729721220765e-05,
      "loss": 1.9227,
      "step": 34155
    },
    {
      "epoch": 0.06654099758917896,
      "grad_norm": 2.3645975589752197,
      "learning_rate": 2.9673444123821306e-05,
      "loss": 1.8725,
      "step": 34170
    },
    {
      "epoch": 0.06657020786028922,
      "grad_norm": 2.382066011428833,
      "learning_rate": 2.967315840285471e-05,
      "loss": 1.7719,
      "step": 34185
    },
    {
      "epoch": 0.06659941813139948,
      "grad_norm": 1.7729606628417969,
      "learning_rate": 2.9672872558323385e-05,
      "loss": 1.8138,
      "step": 34200
    },
    {
      "epoch": 0.06662862840250974,
      "grad_norm": 2.7769312858581543,
      "learning_rate": 2.9672586590229735e-05,
      "loss": 1.8957,
      "step": 34215
    },
    {
      "epoch": 0.06665783867362,
      "grad_norm": 4.402248382568359,
      "learning_rate": 2.9672300498576173e-05,
      "loss": 1.9404,
      "step": 34230
    },
    {
      "epoch": 0.06668704894473028,
      "grad_norm": 4.229050159454346,
      "learning_rate": 2.96720142833651e-05,
      "loss": 1.8112,
      "step": 34245
    },
    {
      "epoch": 0.06671625921584054,
      "grad_norm": 4.353719711303711,
      "learning_rate": 2.9671727944598935e-05,
      "loss": 1.8216,
      "step": 34260
    },
    {
      "epoch": 0.0667454694869508,
      "grad_norm": 3.652221918106079,
      "learning_rate": 2.9671441482280083e-05,
      "loss": 1.9569,
      "step": 34275
    },
    {
      "epoch": 0.06677467975806106,
      "grad_norm": 2.7157819271087646,
      "learning_rate": 2.9671154896410962e-05,
      "loss": 1.7512,
      "step": 34290
    },
    {
      "epoch": 0.06680389002917132,
      "grad_norm": 2.719609498977661,
      "learning_rate": 2.9670868186993982e-05,
      "loss": 1.9252,
      "step": 34305
    },
    {
      "epoch": 0.06683310030028158,
      "grad_norm": 2.834977865219116,
      "learning_rate": 2.967058135403155e-05,
      "loss": 1.7615,
      "step": 34320
    },
    {
      "epoch": 0.06686231057139184,
      "grad_norm": 2.27976393699646,
      "learning_rate": 2.9670294397526097e-05,
      "loss": 1.7578,
      "step": 34335
    },
    {
      "epoch": 0.06689152084250212,
      "grad_norm": 1.7607027292251587,
      "learning_rate": 2.967000731748003e-05,
      "loss": 1.8367,
      "step": 34350
    },
    {
      "epoch": 0.06692073111361238,
      "grad_norm": 3.453352451324463,
      "learning_rate": 2.9669720113895763e-05,
      "loss": 1.8645,
      "step": 34365
    },
    {
      "epoch": 0.06694994138472264,
      "grad_norm": 2.4427194595336914,
      "learning_rate": 2.9669432786775727e-05,
      "loss": 1.8592,
      "step": 34380
    },
    {
      "epoch": 0.0669791516558329,
      "grad_norm": 3.492604970932007,
      "learning_rate": 2.9669145336122335e-05,
      "loss": 1.8394,
      "step": 34395
    },
    {
      "epoch": 0.06700836192694316,
      "grad_norm": 2.3918869495391846,
      "learning_rate": 2.9668857761938e-05,
      "loss": 1.7539,
      "step": 34410
    },
    {
      "epoch": 0.06703757219805342,
      "grad_norm": 1.826912760734558,
      "learning_rate": 2.9668570064225156e-05,
      "loss": 1.8747,
      "step": 34425
    },
    {
      "epoch": 0.06706678246916369,
      "grad_norm": 4.5764241218566895,
      "learning_rate": 2.966828224298622e-05,
      "loss": 1.8474,
      "step": 34440
    },
    {
      "epoch": 0.06709599274027396,
      "grad_norm": 3.9237124919891357,
      "learning_rate": 2.9667994298223612e-05,
      "loss": 1.8965,
      "step": 34455
    },
    {
      "epoch": 0.06712520301138422,
      "grad_norm": 3.064443826675415,
      "learning_rate": 2.9667706229939765e-05,
      "loss": 1.8549,
      "step": 34470
    },
    {
      "epoch": 0.06715441328249448,
      "grad_norm": 2.0611674785614014,
      "learning_rate": 2.96674180381371e-05,
      "loss": 1.8735,
      "step": 34485
    },
    {
      "epoch": 0.06718362355360474,
      "grad_norm": 3.887948513031006,
      "learning_rate": 2.9667129722818044e-05,
      "loss": 1.973,
      "step": 34500
    },
    {
      "epoch": 0.067212833824715,
      "grad_norm": 2.5983119010925293,
      "learning_rate": 2.966684128398503e-05,
      "loss": 1.8455,
      "step": 34515
    },
    {
      "epoch": 0.06724204409582527,
      "grad_norm": 3.6295909881591797,
      "learning_rate": 2.9666552721640474e-05,
      "loss": 1.738,
      "step": 34530
    },
    {
      "epoch": 0.06727125436693553,
      "grad_norm": 2.456125020980835,
      "learning_rate": 2.966626403578682e-05,
      "loss": 1.8651,
      "step": 34545
    },
    {
      "epoch": 0.06730046463804579,
      "grad_norm": 1.6152453422546387,
      "learning_rate": 2.966597522642649e-05,
      "loss": 1.8728,
      "step": 34560
    },
    {
      "epoch": 0.06732967490915606,
      "grad_norm": 2.789376974105835,
      "learning_rate": 2.966568629356193e-05,
      "loss": 1.8996,
      "step": 34575
    },
    {
      "epoch": 0.06735888518026632,
      "grad_norm": 3.9699041843414307,
      "learning_rate": 2.9665397237195555e-05,
      "loss": 1.7645,
      "step": 34590
    },
    {
      "epoch": 0.06738809545137658,
      "grad_norm": 2.7433955669403076,
      "learning_rate": 2.966510805732981e-05,
      "loss": 1.885,
      "step": 34605
    },
    {
      "epoch": 0.06741730572248684,
      "grad_norm": 3.254692316055298,
      "learning_rate": 2.9664818753967123e-05,
      "loss": 1.7664,
      "step": 34620
    },
    {
      "epoch": 0.0674465159935971,
      "grad_norm": 2.647752285003662,
      "learning_rate": 2.966452932710994e-05,
      "loss": 1.8037,
      "step": 34635
    },
    {
      "epoch": 0.06747572626470737,
      "grad_norm": 3.770737648010254,
      "learning_rate": 2.966423977676069e-05,
      "loss": 1.7053,
      "step": 34650
    },
    {
      "epoch": 0.06750493653581763,
      "grad_norm": 2.101461410522461,
      "learning_rate": 2.9663950102921814e-05,
      "loss": 1.8039,
      "step": 34665
    },
    {
      "epoch": 0.0675341468069279,
      "grad_norm": 4.246973514556885,
      "learning_rate": 2.9663660305595754e-05,
      "loss": 1.776,
      "step": 34680
    },
    {
      "epoch": 0.06756335707803816,
      "grad_norm": 6.2981157302856445,
      "learning_rate": 2.9663370384784946e-05,
      "loss": 1.7988,
      "step": 34695
    },
    {
      "epoch": 0.06759256734914842,
      "grad_norm": 2.8231287002563477,
      "learning_rate": 2.9663080340491838e-05,
      "loss": 1.9157,
      "step": 34710
    },
    {
      "epoch": 0.06762177762025869,
      "grad_norm": 2.1489720344543457,
      "learning_rate": 2.9662790172718867e-05,
      "loss": 1.7745,
      "step": 34725
    },
    {
      "epoch": 0.06765098789136895,
      "grad_norm": 2.714576482772827,
      "learning_rate": 2.9662499881468475e-05,
      "loss": 1.8596,
      "step": 34740
    },
    {
      "epoch": 0.06768019816247921,
      "grad_norm": 2.1026928424835205,
      "learning_rate": 2.966220946674311e-05,
      "loss": 1.9096,
      "step": 34755
    },
    {
      "epoch": 0.06770940843358947,
      "grad_norm": 4.316662311553955,
      "learning_rate": 2.9661918928545215e-05,
      "loss": 1.8735,
      "step": 34770
    },
    {
      "epoch": 0.06773861870469974,
      "grad_norm": 3.814514636993408,
      "learning_rate": 2.966162826687724e-05,
      "loss": 1.9701,
      "step": 34785
    },
    {
      "epoch": 0.06776782897581,
      "grad_norm": 4.129803657531738,
      "learning_rate": 2.966133748174164e-05,
      "loss": 1.9273,
      "step": 34800
    },
    {
      "epoch": 0.06779703924692027,
      "grad_norm": 1.9344481229782104,
      "learning_rate": 2.966104657314084e-05,
      "loss": 1.8738,
      "step": 34815
    },
    {
      "epoch": 0.06782624951803053,
      "grad_norm": 5.156272888183594,
      "learning_rate": 2.9660755541077314e-05,
      "loss": 1.9722,
      "step": 34830
    },
    {
      "epoch": 0.06785545978914079,
      "grad_norm": 2.4508838653564453,
      "learning_rate": 2.9660464385553504e-05,
      "loss": 1.8354,
      "step": 34845
    },
    {
      "epoch": 0.06788467006025105,
      "grad_norm": 4.646978855133057,
      "learning_rate": 2.9660173106571856e-05,
      "loss": 1.6924,
      "step": 34860
    },
    {
      "epoch": 0.06791388033136131,
      "grad_norm": 4.116873741149902,
      "learning_rate": 2.9659881704134835e-05,
      "loss": 1.9238,
      "step": 34875
    },
    {
      "epoch": 0.06794309060247158,
      "grad_norm": 2.2530601024627686,
      "learning_rate": 2.965959017824488e-05,
      "loss": 1.744,
      "step": 34890
    },
    {
      "epoch": 0.06797230087358185,
      "grad_norm": 3.1118083000183105,
      "learning_rate": 2.9659298528904462e-05,
      "loss": 1.7291,
      "step": 34905
    },
    {
      "epoch": 0.0680015111446921,
      "grad_norm": 4.10203218460083,
      "learning_rate": 2.9659006756116024e-05,
      "loss": 1.9676,
      "step": 34920
    },
    {
      "epoch": 0.06803072141580237,
      "grad_norm": 2.487515926361084,
      "learning_rate": 2.9658714859882033e-05,
      "loss": 1.755,
      "step": 34935
    },
    {
      "epoch": 0.06805993168691263,
      "grad_norm": 4.722667694091797,
      "learning_rate": 2.965842284020494e-05,
      "loss": 1.6132,
      "step": 34950
    },
    {
      "epoch": 0.06808914195802289,
      "grad_norm": 3.8074474334716797,
      "learning_rate": 2.9658130697087206e-05,
      "loss": 1.8949,
      "step": 34965
    },
    {
      "epoch": 0.06811835222913315,
      "grad_norm": 3.2412679195404053,
      "learning_rate": 2.9657838430531295e-05,
      "loss": 1.8062,
      "step": 34980
    },
    {
      "epoch": 0.06814756250024342,
      "grad_norm": 3.2179923057556152,
      "learning_rate": 2.9657546040539662e-05,
      "loss": 1.7753,
      "step": 34995
    }
  ],
  "logging_steps": 15,
  "max_steps": 513518,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 5.536910190624768e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}