{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.055086175435697216,
  "eval_steps": 375,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.672411695713148e-05,
      "grad_norm": 0.20513708889484406,
      "learning_rate": 2e-05,
      "loss": 1.597,
      "step": 1
    },
    {
      "epoch": 7.344823391426296e-05,
      "grad_norm": 0.22414308786392212,
      "learning_rate": 4e-05,
      "loss": 1.6271,
      "step": 2
    },
    {
      "epoch": 0.00011017235087139444,
      "grad_norm": 0.24069412052631378,
      "learning_rate": 6e-05,
      "loss": 1.7479,
      "step": 3
    },
    {
      "epoch": 0.00014689646782852593,
      "grad_norm": 0.25528982281684875,
      "learning_rate": 8e-05,
      "loss": 1.755,
      "step": 4
    },
    {
      "epoch": 0.0001836205847856574,
      "grad_norm": 0.27266138792037964,
      "learning_rate": 0.0001,
      "loss": 1.8481,
      "step": 5
    },
    {
      "epoch": 0.00022034470174278888,
      "grad_norm": 0.2748166620731354,
      "learning_rate": 0.00012,
      "loss": 1.7316,
      "step": 6
    },
    {
      "epoch": 0.00025706881869992036,
      "grad_norm": 0.2660968005657196,
      "learning_rate": 0.00014,
      "loss": 1.5395,
      "step": 7
    },
    {
      "epoch": 0.00029379293565705186,
      "grad_norm": 0.27617958188056946,
      "learning_rate": 0.00016,
      "loss": 1.5831,
      "step": 8
    },
    {
      "epoch": 0.0003305170526141833,
      "grad_norm": 0.2743740677833557,
      "learning_rate": 0.00018,
      "loss": 1.5308,
      "step": 9
    },
    {
      "epoch": 0.0003672411695713148,
      "grad_norm": 0.3421348035335541,
      "learning_rate": 0.0002,
      "loss": 1.3795,
      "step": 10
    },
    {
      "epoch": 0.00040396528652844626,
      "grad_norm": 0.32813140749931335,
      "learning_rate": 0.00019999977772170748,
      "loss": 1.417,
      "step": 11
    },
    {
      "epoch": 0.00044068940348557776,
      "grad_norm": 0.3679603934288025,
      "learning_rate": 0.00019999911088781805,
      "loss": 1.5825,
      "step": 12
    },
    {
      "epoch": 0.0004774135204427092,
      "grad_norm": 0.3459932506084442,
      "learning_rate": 0.0001999979995012962,
      "loss": 1.6053,
      "step": 13
    },
    {
      "epoch": 0.0005141376373998407,
      "grad_norm": 0.40405407547950745,
      "learning_rate": 0.00019999644356708261,
      "loss": 1.4521,
      "step": 14
    },
    {
      "epoch": 0.0005508617543569722,
      "grad_norm": 0.43900588154792786,
      "learning_rate": 0.00019999444309209432,
      "loss": 1.5331,
      "step": 15
    },
    {
      "epoch": 0.0005875858713141037,
      "grad_norm": 0.40844762325286865,
      "learning_rate": 0.0001999919980852246,
      "loss": 1.4974,
      "step": 16
    },
    {
      "epoch": 0.0006243099882712351,
      "grad_norm": 0.42985180020332336,
      "learning_rate": 0.00019998910855734288,
      "loss": 1.3543,
      "step": 17
    },
    {
      "epoch": 0.0006610341052283666,
      "grad_norm": 0.4113822281360626,
      "learning_rate": 0.0001999857745212947,
      "loss": 1.3932,
      "step": 18
    },
    {
      "epoch": 0.0006977582221854981,
      "grad_norm": 0.4689984917640686,
      "learning_rate": 0.00019998199599190178,
      "loss": 1.5862,
      "step": 19
    },
    {
      "epoch": 0.0007344823391426296,
      "grad_norm": 0.42687124013900757,
      "learning_rate": 0.0001999777729859618,
      "loss": 1.2747,
      "step": 20
    },
    {
      "epoch": 0.000771206456099761,
      "grad_norm": 0.46551212668418884,
      "learning_rate": 0.00019997310552224846,
      "loss": 1.3753,
      "step": 21
    },
    {
      "epoch": 0.0008079305730568925,
      "grad_norm": 0.46449393033981323,
      "learning_rate": 0.00019996799362151122,
      "loss": 1.6019,
      "step": 22
    },
    {
      "epoch": 0.000844654690014024,
      "grad_norm": 0.5185602903366089,
      "learning_rate": 0.00019996243730647538,
      "loss": 1.2967,
      "step": 23
    },
    {
      "epoch": 0.0008813788069711555,
      "grad_norm": 0.5354712009429932,
      "learning_rate": 0.00019995643660184191,
      "loss": 1.3975,
      "step": 24
    },
    {
      "epoch": 0.000918102923928287,
      "grad_norm": 0.5164687633514404,
      "learning_rate": 0.00019994999153428737,
      "loss": 1.3142,
      "step": 25
    },
    {
      "epoch": 0.0009548270408854184,
      "grad_norm": 0.5585957765579224,
      "learning_rate": 0.00019994310213246368,
      "loss": 1.5114,
      "step": 26
    },
    {
      "epoch": 0.00099155115784255,
      "grad_norm": 0.560759425163269,
      "learning_rate": 0.00019993576842699816,
      "loss": 1.365,
      "step": 27
    },
    {
      "epoch": 0.0010282752747996814,
      "grad_norm": 0.6499810814857483,
      "learning_rate": 0.0001999279904504933,
      "loss": 1.4964,
      "step": 28
    },
    {
      "epoch": 0.0010649993917568128,
      "grad_norm": 0.619306206703186,
      "learning_rate": 0.00019991976823752653,
      "loss": 1.3245,
      "step": 29
    },
    {
      "epoch": 0.0011017235087139444,
      "grad_norm": 0.5648550987243652,
      "learning_rate": 0.00019991110182465032,
      "loss": 1.2674,
      "step": 30
    },
    {
      "epoch": 0.0011384476256710758,
      "grad_norm": 0.6652882695198059,
      "learning_rate": 0.00019990199125039174,
      "loss": 1.2976,
      "step": 31
    },
    {
      "epoch": 0.0011751717426282074,
      "grad_norm": 0.7157711982727051,
      "learning_rate": 0.00019989243655525247,
      "loss": 1.2119,
      "step": 32
    },
    {
      "epoch": 0.0012118958595853388,
      "grad_norm": 0.7031236886978149,
      "learning_rate": 0.00019988243778170853,
      "loss": 1.2599,
      "step": 33
    },
    {
      "epoch": 0.0012486199765424702,
      "grad_norm": 0.8519224524497986,
      "learning_rate": 0.0001998719949742101,
      "loss": 1.405,
      "step": 34
    },
    {
      "epoch": 0.0012853440934996018,
      "grad_norm": 0.7543824911117554,
      "learning_rate": 0.0001998611081791814,
      "loss": 1.4087,
      "step": 35
    },
    {
      "epoch": 0.0013220682104567332,
      "grad_norm": 0.8041827082633972,
      "learning_rate": 0.00019984977744502038,
      "loss": 1.3854,
      "step": 36
    },
    {
      "epoch": 0.0013587923274138646,
      "grad_norm": 0.9433043599128723,
      "learning_rate": 0.00019983800282209857,
      "loss": 1.4865,
      "step": 37
    },
    {
      "epoch": 0.0013955164443709962,
      "grad_norm": 0.8946244120597839,
      "learning_rate": 0.00019982578436276082,
      "loss": 1.1977,
      "step": 38
    },
    {
      "epoch": 0.0014322405613281276,
      "grad_norm": 1.0030922889709473,
      "learning_rate": 0.00019981312212132512,
      "loss": 1.4746,
      "step": 39
    },
    {
      "epoch": 0.0014689646782852592,
      "grad_norm": 1.024678349494934,
      "learning_rate": 0.00019980001615408228,
      "loss": 1.3357,
      "step": 40
    },
    {
      "epoch": 0.0015056887952423906,
      "grad_norm": 1.0329046249389648,
      "learning_rate": 0.00019978646651929572,
      "loss": 1.2657,
      "step": 41
    },
    {
      "epoch": 0.001542412912199522,
      "grad_norm": 0.8716040253639221,
      "learning_rate": 0.00019977247327720128,
      "loss": 1.3346,
      "step": 42
    },
    {
      "epoch": 0.0015791370291566536,
      "grad_norm": 1.2746973037719727,
      "learning_rate": 0.0001997580364900068,
      "loss": 1.1684,
      "step": 43
    },
    {
      "epoch": 0.001615861146113785,
      "grad_norm": 1.2592463493347168,
      "learning_rate": 0.000199743156221892,
      "loss": 1.3664,
      "step": 44
    },
    {
      "epoch": 0.0016525852630709166,
      "grad_norm": 1.1690689325332642,
      "learning_rate": 0.00019972783253900808,
      "loss": 1.1748,
      "step": 45
    },
    {
      "epoch": 0.001689309380028048,
      "grad_norm": 1.367287039756775,
      "learning_rate": 0.00019971206550947748,
      "loss": 1.1485,
      "step": 46
    },
    {
      "epoch": 0.0017260334969851794,
      "grad_norm": 1.5622020959854126,
      "learning_rate": 0.00019969585520339354,
      "loss": 1.2815,
      "step": 47
    },
    {
      "epoch": 0.001762757613942311,
      "grad_norm": 1.5811524391174316,
      "learning_rate": 0.0001996792016928203,
      "loss": 1.2898,
      "step": 48
    },
    {
      "epoch": 0.0017994817308994424,
      "grad_norm": 1.611449122428894,
      "learning_rate": 0.00019966210505179197,
      "loss": 1.1641,
      "step": 49
    },
    {
      "epoch": 0.001836205847856574,
      "grad_norm": 2.270111083984375,
      "learning_rate": 0.00019964456535631286,
      "loss": 1.5822,
      "step": 50
    },
    {
      "epoch": 0.0018729299648137054,
      "grad_norm": 1.0750924348831177,
      "learning_rate": 0.0001996265826843568,
      "loss": 1.7117,
      "step": 51
    },
    {
      "epoch": 0.0019096540817708368,
      "grad_norm": 1.1549891233444214,
      "learning_rate": 0.00019960815711586696,
      "loss": 1.8695,
      "step": 52
    },
    {
      "epoch": 0.0019463781987279685,
      "grad_norm": 0.8596017956733704,
      "learning_rate": 0.00019958928873275539,
      "loss": 1.7824,
      "step": 53
    },
    {
      "epoch": 0.0019831023156851,
      "grad_norm": 0.5342891812324524,
      "learning_rate": 0.00019956997761890277,
      "loss": 1.706,
      "step": 54
    },
    {
      "epoch": 0.0020198264326422312,
      "grad_norm": 0.3867919445037842,
      "learning_rate": 0.00019955022386015792,
      "loss": 1.5963,
      "step": 55
    },
    {
      "epoch": 0.002056550549599363,
      "grad_norm": 0.3200293183326721,
      "learning_rate": 0.00019953002754433743,
      "loss": 1.6127,
      "step": 56
    },
    {
      "epoch": 0.0020932746665564945,
      "grad_norm": 0.2999487817287445,
      "learning_rate": 0.00019950938876122542,
      "loss": 1.5714,
      "step": 57
    },
    {
      "epoch": 0.0021299987835136256,
      "grad_norm": 0.30229923129081726,
      "learning_rate": 0.00019948830760257291,
      "loss": 1.5561,
      "step": 58
    },
    {
      "epoch": 0.0021667229004707573,
      "grad_norm": 0.3074771463871002,
      "learning_rate": 0.0001994667841620976,
      "loss": 1.5753,
      "step": 59
    },
    {
      "epoch": 0.002203447017427889,
      "grad_norm": 0.34731152653694153,
      "learning_rate": 0.00019944481853548335,
      "loss": 1.6722,
      "step": 60
    },
    {
      "epoch": 0.00224017113438502,
      "grad_norm": 0.3669075667858124,
      "learning_rate": 0.00019942241082037982,
      "loss": 1.5859,
      "step": 61
    },
    {
      "epoch": 0.0022768952513421517,
      "grad_norm": 0.40429770946502686,
      "learning_rate": 0.00019939956111640197,
      "loss": 1.5168,
      "step": 62
    },
    {
      "epoch": 0.0023136193682992833,
      "grad_norm": 0.46783143281936646,
      "learning_rate": 0.00019937626952512964,
      "loss": 1.5518,
      "step": 63
    },
    {
      "epoch": 0.002350343485256415,
      "grad_norm": 0.4412369132041931,
      "learning_rate": 0.0001993525361501072,
      "loss": 1.5243,
      "step": 64
    },
    {
      "epoch": 0.002387067602213546,
      "grad_norm": 0.44865235686302185,
      "learning_rate": 0.00019932836109684286,
      "loss": 1.5518,
      "step": 65
    },
    {
      "epoch": 0.0024237917191706777,
      "grad_norm": 0.41685259342193604,
      "learning_rate": 0.00019930374447280845,
      "loss": 1.4616,
      "step": 66
    },
    {
      "epoch": 0.0024605158361278093,
      "grad_norm": 0.46259555220603943,
      "learning_rate": 0.00019927868638743875,
      "loss": 1.7916,
      "step": 67
    },
    {
      "epoch": 0.0024972399530849405,
      "grad_norm": 0.4173118770122528,
      "learning_rate": 0.0001992531869521312,
      "loss": 1.4857,
      "step": 68
    },
    {
      "epoch": 0.002533964070042072,
      "grad_norm": 0.42661041021347046,
      "learning_rate": 0.00019922724628024515,
      "loss": 1.6275,
      "step": 69
    },
    {
      "epoch": 0.0025706881869992037,
      "grad_norm": 0.40807750821113586,
      "learning_rate": 0.0001992008644871016,
      "loss": 1.3025,
      "step": 70
    },
    {
      "epoch": 0.002607412303956335,
      "grad_norm": 0.44255152344703674,
      "learning_rate": 0.00019917404168998256,
      "loss": 1.3303,
      "step": 71
    },
    {
      "epoch": 0.0026441364209134665,
      "grad_norm": 0.5059233903884888,
      "learning_rate": 0.0001991467780081305,
      "loss": 1.4523,
      "step": 72
    },
    {
      "epoch": 0.002680860537870598,
      "grad_norm": 0.47907745838165283,
      "learning_rate": 0.00019911907356274795,
      "loss": 1.4456,
      "step": 73
    },
    {
      "epoch": 0.0027175846548277293,
      "grad_norm": 0.507641077041626,
      "learning_rate": 0.00019909092847699683,
      "loss": 1.2867,
      "step": 74
    },
    {
      "epoch": 0.002754308771784861,
      "grad_norm": 0.514249324798584,
      "learning_rate": 0.00019906234287599798,
      "loss": 1.5424,
      "step": 75
    },
    {
      "epoch": 0.0027910328887419925,
      "grad_norm": 0.5781522393226624,
      "learning_rate": 0.00019903331688683057,
      "loss": 1.3849,
      "step": 76
    },
    {
      "epoch": 0.002827757005699124,
      "grad_norm": 0.5795761942863464,
      "learning_rate": 0.00019900385063853154,
      "loss": 1.2846,
      "step": 77
    },
    {
      "epoch": 0.0028644811226562553,
      "grad_norm": 0.5483823418617249,
      "learning_rate": 0.00019897394426209505,
      "loss": 1.4211,
      "step": 78
    },
    {
      "epoch": 0.002901205239613387,
      "grad_norm": 0.6278328895568848,
      "learning_rate": 0.00019894359789047187,
      "loss": 1.4692,
      "step": 79
    },
    {
      "epoch": 0.0029379293565705185,
      "grad_norm": 0.6398199796676636,
      "learning_rate": 0.00019891281165856873,
      "loss": 1.3862,
      "step": 80
    },
    {
      "epoch": 0.0029746534735276497,
      "grad_norm": 0.8465139865875244,
      "learning_rate": 0.00019888158570324795,
      "loss": 1.0812,
      "step": 81
    },
    {
      "epoch": 0.0030113775904847813,
      "grad_norm": 0.721346378326416,
      "learning_rate": 0.0001988499201633265,
      "loss": 1.4107,
      "step": 82
    },
    {
      "epoch": 0.003048101707441913,
      "grad_norm": 0.7719030976295471,
      "learning_rate": 0.00019881781517957562,
      "loss": 1.2957,
      "step": 83
    },
    {
      "epoch": 0.003084825824399044,
      "grad_norm": 0.7670479416847229,
      "learning_rate": 0.0001987852708947202,
      "loss": 1.1521,
      "step": 84
    },
    {
      "epoch": 0.0031215499413561757,
      "grad_norm": 0.7475246787071228,
      "learning_rate": 0.00019875228745343794,
      "loss": 1.1696,
      "step": 85
    },
    {
      "epoch": 0.0031582740583133073,
      "grad_norm": 0.9474829435348511,
      "learning_rate": 0.0001987188650023589,
      "loss": 1.388,
      "step": 86
    },
    {
      "epoch": 0.003194998175270439,
      "grad_norm": 0.8713952302932739,
      "learning_rate": 0.0001986850036900648,
      "loss": 1.2912,
      "step": 87
    },
    {
      "epoch": 0.00323172229222757,
      "grad_norm": 0.940808117389679,
      "learning_rate": 0.00019865070366708836,
      "loss": 1.2101,
      "step": 88
    },
    {
      "epoch": 0.0032684464091847017,
      "grad_norm": 0.981411874294281,
      "learning_rate": 0.00019861596508591255,
      "loss": 1.2999,
      "step": 89
    },
    {
      "epoch": 0.0033051705261418333,
      "grad_norm": 1.036609411239624,
      "learning_rate": 0.00019858078810097002,
      "loss": 1.3757,
      "step": 90
    },
    {
      "epoch": 0.0033418946430989645,
      "grad_norm": 0.9568408727645874,
      "learning_rate": 0.00019854517286864245,
      "loss": 1.1213,
      "step": 91
    },
    {
      "epoch": 0.003378618760056096,
      "grad_norm": 1.0471506118774414,
      "learning_rate": 0.0001985091195472596,
      "loss": 1.2401,
      "step": 92
    },
    {
      "epoch": 0.0034153428770132277,
      "grad_norm": 1.1496285200119019,
      "learning_rate": 0.0001984726282970989,
      "loss": 1.3033,
      "step": 93
    },
    {
      "epoch": 0.003452066993970359,
      "grad_norm": 1.1266779899597168,
      "learning_rate": 0.0001984356992803847,
      "loss": 1.0255,
      "step": 94
    },
    {
      "epoch": 0.0034887911109274905,
      "grad_norm": 1.1548618078231812,
      "learning_rate": 0.00019839833266128724,
      "loss": 1.1411,
      "step": 95
    },
    {
      "epoch": 0.003525515227884622,
      "grad_norm": 1.2729730606079102,
      "learning_rate": 0.00019836052860592237,
      "loss": 1.2182,
      "step": 96
    },
    {
      "epoch": 0.0035622393448417533,
      "grad_norm": 1.3738242387771606,
      "learning_rate": 0.0001983222872823505,
      "loss": 1.3431,
      "step": 97
    },
    {
      "epoch": 0.003598963461798885,
      "grad_norm": 1.3226675987243652,
      "learning_rate": 0.00019828360886057594,
      "loss": 1.1926,
      "step": 98
    },
    {
      "epoch": 0.0036356875787560165,
      "grad_norm": 1.5393412113189697,
      "learning_rate": 0.00019824449351254616,
      "loss": 0.971,
      "step": 99
    },
    {
      "epoch": 0.003672411695713148,
      "grad_norm": 2.251595973968506,
      "learning_rate": 0.00019820494141215104,
      "loss": 1.0961,
      "step": 100
    },
    {
      "epoch": 0.0037091358126702793,
      "grad_norm": 0.4639330804347992,
      "learning_rate": 0.000198164952735222,
      "loss": 1.7433,
      "step": 101
    },
    {
      "epoch": 0.003745859929627411,
      "grad_norm": 0.5886768102645874,
      "learning_rate": 0.00019812452765953135,
      "loss": 1.7479,
      "step": 102
    },
    {
      "epoch": 0.0037825840465845425,
      "grad_norm": 0.5448111295700073,
      "learning_rate": 0.00019808366636479147,
      "loss": 1.8777,
      "step": 103
    },
    {
      "epoch": 0.0038193081635416737,
      "grad_norm": 0.5520771741867065,
      "learning_rate": 0.00019804236903265388,
      "loss": 1.631,
      "step": 104
    },
    {
      "epoch": 0.0038560322804988053,
      "grad_norm": 0.4806782603263855,
      "learning_rate": 0.00019800063584670863,
      "loss": 1.7175,
      "step": 105
    },
    {
      "epoch": 0.003892756397455937,
      "grad_norm": 0.35501325130462646,
      "learning_rate": 0.00019795846699248332,
      "loss": 1.5065,
      "step": 106
    },
    {
      "epoch": 0.0039294805144130685,
      "grad_norm": 0.3451433479785919,
      "learning_rate": 0.00019791586265744237,
      "loss": 1.6491,
      "step": 107
    },
    {
      "epoch": 0.0039662046313702,
      "grad_norm": 0.3105389177799225,
      "learning_rate": 0.00019787282303098617,
      "loss": 1.526,
      "step": 108
    },
    {
      "epoch": 0.004002928748327331,
      "grad_norm": 0.3053218424320221,
      "learning_rate": 0.0001978293483044502,
      "loss": 1.398,
      "step": 109
    },
    {
      "epoch": 0.0040396528652844625,
      "grad_norm": 0.3019847869873047,
      "learning_rate": 0.00019778543867110426,
      "loss": 1.5692,
      "step": 110
    },
    {
      "epoch": 0.004076376982241594,
      "grad_norm": 0.31656384468078613,
      "learning_rate": 0.00019774109432615147,
      "loss": 1.5968,
      "step": 111
    },
    {
      "epoch": 0.004113101099198726,
      "grad_norm": 0.3558514714241028,
      "learning_rate": 0.00019769631546672756,
      "loss": 1.6489,
      "step": 112
    },
    {
      "epoch": 0.004149825216155857,
      "grad_norm": 0.4041827917098999,
      "learning_rate": 0.00019765110229189988,
      "loss": 1.6954,
      "step": 113
    },
    {
      "epoch": 0.004186549333112989,
      "grad_norm": 0.37753960490226746,
      "learning_rate": 0.00019760545500266657,
      "loss": 1.4188,
      "step": 114
    },
    {
      "epoch": 0.0042232734500701205,
      "grad_norm": 0.3455388844013214,
      "learning_rate": 0.00019755937380195568,
      "loss": 1.6167,
      "step": 115
    },
    {
      "epoch": 0.004259997567027251,
      "grad_norm": 0.40142956376075745,
      "learning_rate": 0.00019751285889462423,
      "loss": 1.6301,
      "step": 116
    },
    {
      "epoch": 0.004296721683984383,
      "grad_norm": 0.4501782953739166,
      "learning_rate": 0.0001974659104874573,
      "loss": 1.5667,
      "step": 117
    },
    {
      "epoch": 0.0043334458009415145,
      "grad_norm": 0.43178555369377136,
      "learning_rate": 0.0001974185287891671,
      "loss": 1.4848,
      "step": 118
    },
    {
      "epoch": 0.004370169917898646,
      "grad_norm": 0.42376014590263367,
      "learning_rate": 0.0001973707140103921,
      "loss": 1.383,
      "step": 119
    },
    {
      "epoch": 0.004406894034855778,
      "grad_norm": 0.44833165407180786,
      "learning_rate": 0.00019732246636369605,
      "loss": 1.4163,
      "step": 120
    },
    {
      "epoch": 0.004443618151812909,
      "grad_norm": 0.4841060936450958,
      "learning_rate": 0.00019727378606356703,
      "loss": 1.4926,
      "step": 121
    },
    {
      "epoch": 0.00448034226877004,
      "grad_norm": 0.45483699440956116,
      "learning_rate": 0.00019722467332641656,
      "loss": 1.3195,
      "step": 122
    },
    {
      "epoch": 0.004517066385727172,
      "grad_norm": 0.4739449620246887,
      "learning_rate": 0.00019717512837057855,
      "loss": 1.604,
      "step": 123
    },
    {
      "epoch": 0.004553790502684303,
      "grad_norm": 0.562369704246521,
      "learning_rate": 0.0001971251514163083,
      "loss": 1.4788,
      "step": 124
    },
    {
      "epoch": 0.004590514619641435,
      "grad_norm": 0.5533655881881714,
      "learning_rate": 0.0001970747426857817,
      "loss": 1.2155,
      "step": 125
    },
    {
      "epoch": 0.0046272387365985665,
      "grad_norm": 0.5289334058761597,
      "learning_rate": 0.00019702390240309404,
      "loss": 1.3699,
      "step": 126
    },
    {
      "epoch": 0.004663962853555698,
      "grad_norm": 0.4955001175403595,
      "learning_rate": 0.0001969726307942592,
      "loss": 1.2211,
      "step": 127
    },
    {
      "epoch": 0.00470068697051283,
      "grad_norm": 0.6196656823158264,
      "learning_rate": 0.00019692092808720846,
      "loss": 1.3253,
      "step": 128
    },
    {
      "epoch": 0.0047374110874699605,
      "grad_norm": 0.5415480136871338,
      "learning_rate": 0.0001968687945117896,
      "loss": 1.1904,
      "step": 129
    },
    {
      "epoch": 0.004774135204427092,
      "grad_norm": 0.7057448029518127,
      "learning_rate": 0.00019681623029976588,
      "loss": 1.2524,
      "step": 130
    },
    {
      "epoch": 0.004810859321384224,
      "grad_norm": 0.7077358961105347,
      "learning_rate": 0.00019676323568481498,
      "loss": 1.4879,
      "step": 131
    },
    {
      "epoch": 0.004847583438341355,
      "grad_norm": 0.9498615860939026,
      "learning_rate": 0.00019670981090252792,
      "loss": 1.3681,
      "step": 132
    },
    {
      "epoch": 0.004884307555298487,
      "grad_norm": 0.7702960968017578,
      "learning_rate": 0.00019665595619040808,
      "loss": 1.181,
      "step": 133
    },
    {
      "epoch": 0.0049210316722556185,
      "grad_norm": 0.6912301182746887,
      "learning_rate": 0.0001966016717878702,
      "loss": 0.9197,
      "step": 134
    },
    {
      "epoch": 0.004957755789212749,
      "grad_norm": 0.7591871023178101,
      "learning_rate": 0.00019654695793623907,
      "loss": 0.9831,
      "step": 135
    },
    {
      "epoch": 0.004994479906169881,
      "grad_norm": 0.8321182131767273,
      "learning_rate": 0.0001964918148787488,
      "loss": 1.4199,
      "step": 136
    },
    {
      "epoch": 0.0050312040231270125,
      "grad_norm": 0.9303563833236694,
      "learning_rate": 0.00019643624286054144,
      "loss": 1.1338,
      "step": 137
    },
    {
      "epoch": 0.005067928140084144,
      "grad_norm": 0.9619457721710205,
      "learning_rate": 0.00019638024212866606,
      "loss": 1.1613,
      "step": 138
    },
    {
      "epoch": 0.005104652257041276,
      "grad_norm": 0.8900859355926514,
      "learning_rate": 0.0001963238129320776,
      "loss": 0.907,
      "step": 139
    },
    {
      "epoch": 0.005141376373998407,
      "grad_norm": 0.8651755452156067,
      "learning_rate": 0.00019626695552163578,
      "loss": 1.1722,
      "step": 140
    },
    {
      "epoch": 0.005178100490955539,
      "grad_norm": 1.0805082321166992,
      "learning_rate": 0.00019620967015010395,
      "loss": 1.2615,
      "step": 141
    },
    {
      "epoch": 0.00521482460791267,
      "grad_norm": 1.3124961853027344,
      "learning_rate": 0.00019615195707214803,
      "loss": 1.3879,
      "step": 142
    },
    {
      "epoch": 0.005251548724869801,
      "grad_norm": 0.9659666419029236,
      "learning_rate": 0.0001960938165443353,
      "loss": 1.0666,
      "step": 143
    },
    {
      "epoch": 0.005288272841826933,
      "grad_norm": 1.118941068649292,
      "learning_rate": 0.00019603524882513327,
      "loss": 1.1902,
      "step": 144
    },
    {
      "epoch": 0.0053249969587840645,
      "grad_norm": 1.4917659759521484,
      "learning_rate": 0.0001959762541749086,
      "loss": 1.5368,
      "step": 145
    },
    {
      "epoch": 0.005361721075741196,
      "grad_norm": 0.9861322045326233,
      "learning_rate": 0.00019591683285592593,
      "loss": 0.9165,
      "step": 146
    },
    {
      "epoch": 0.005398445192698328,
      "grad_norm": 1.0425766706466675,
      "learning_rate": 0.00019585698513234663,
      "loss": 1.0769,
      "step": 147
    },
    {
      "epoch": 0.0054351693096554585,
      "grad_norm": 1.1993998289108276,
      "learning_rate": 0.0001957967112702277,
      "loss": 0.8454,
      "step": 148
    },
    {
      "epoch": 0.00547189342661259,
      "grad_norm": 1.1993870735168457,
      "learning_rate": 0.00019573601153752052,
      "loss": 1.4025,
      "step": 149
    },
    {
      "epoch": 0.005508617543569722,
      "grad_norm": 2.824781894683838,
      "learning_rate": 0.00019567488620406983,
      "loss": 1.844,
      "step": 150
    },
    {
      "epoch": 0.005545341660526853,
      "grad_norm": 0.3162299692630768,
      "learning_rate": 0.00019561333554161224,
      "loss": 1.4276,
      "step": 151
    },
    {
      "epoch": 0.005582065777483985,
      "grad_norm": 0.48461535573005676,
      "learning_rate": 0.0001955513598237753,
      "loss": 1.5173,
      "step": 152
    },
    {
      "epoch": 0.0056187898944411166,
      "grad_norm": 0.4895917475223541,
      "learning_rate": 0.00019548895932607621,
      "loss": 1.8674,
      "step": 153
    },
    {
      "epoch": 0.005655514011398248,
      "grad_norm": 0.45790672302246094,
      "learning_rate": 0.00019542613432592038,
      "loss": 1.7434,
      "step": 154
    },
    {
      "epoch": 0.005692238128355379,
      "grad_norm": 0.34740132093429565,
      "learning_rate": 0.00019536288510260056,
      "loss": 1.6913,
      "step": 155
    },
    {
      "epoch": 0.0057289622453125105,
      "grad_norm": 0.42053771018981934,
      "learning_rate": 0.00019529921193729534,
      "loss": 1.6131,
      "step": 156
    },
    {
      "epoch": 0.005765686362269642,
      "grad_norm": 0.35702255368232727,
      "learning_rate": 0.00019523511511306793,
      "loss": 1.5921,
      "step": 157
    },
    {
      "epoch": 0.005802410479226774,
      "grad_norm": 0.3384164571762085,
      "learning_rate": 0.000195170594914865,
      "loss": 1.5233,
      "step": 158
    },
    {
      "epoch": 0.005839134596183905,
      "grad_norm": 0.3720633387565613,
      "learning_rate": 0.00019510565162951537,
      "loss": 1.6443,
      "step": 159
    },
    {
      "epoch": 0.005875858713141037,
      "grad_norm": 0.3458070755004883,
      "learning_rate": 0.00019504028554572864,
      "loss": 1.4968,
      "step": 160
    },
    {
      "epoch": 0.005912582830098169,
      "grad_norm": 0.3496636152267456,
      "learning_rate": 0.00019497449695409408,
      "loss": 1.6608,
      "step": 161
    },
    {
      "epoch": 0.005949306947055299,
      "grad_norm": 0.33955851197242737,
      "learning_rate": 0.00019490828614707916,
      "loss": 1.6084,
      "step": 162
    },
    {
      "epoch": 0.005986031064012431,
      "grad_norm": 0.34924259781837463,
      "learning_rate": 0.00019484165341902845,
      "loss": 1.5904,
      "step": 163
    },
    {
      "epoch": 0.0060227551809695625,
      "grad_norm": 0.34039485454559326,
      "learning_rate": 0.00019477459906616206,
      "loss": 1.4567,
      "step": 164
    },
    {
      "epoch": 0.006059479297926694,
      "grad_norm": 0.3943594694137573,
      "learning_rate": 0.00019470712338657458,
      "loss": 1.6373,
      "step": 165
    },
    {
      "epoch": 0.006096203414883826,
      "grad_norm": 0.3953979015350342,
      "learning_rate": 0.0001946392266802336,
      "loss": 1.3525,
      "step": 166
    },
    {
      "epoch": 0.006132927531840957,
      "grad_norm": 0.41088470816612244,
      "learning_rate": 0.0001945709092489783,
      "loss": 1.3018,
      "step": 167
    },
    {
      "epoch": 0.006169651648798088,
      "grad_norm": 0.422065794467926,
      "learning_rate": 0.00019450217139651844,
      "loss": 1.4741,
      "step": 168
    },
    {
      "epoch": 0.00620637576575522,
      "grad_norm": 0.4211116433143616,
      "learning_rate": 0.0001944330134284326,
      "loss": 1.3508,
      "step": 169
    },
    {
      "epoch": 0.006243099882712351,
      "grad_norm": 0.42598092555999756,
      "learning_rate": 0.00019436343565216711,
      "loss": 1.3221,
      "step": 170
    },
    {
      "epoch": 0.006279823999669483,
      "grad_norm": 0.4171473979949951,
      "learning_rate": 0.00019429343837703455,
      "loss": 1.4506,
      "step": 171
    },
    {
      "epoch": 0.006316548116626615,
      "grad_norm": 0.47473809123039246,
      "learning_rate": 0.0001942230219142124,
      "loss": 1.634,
      "step": 172
    },
    {
      "epoch": 0.006353272233583746,
      "grad_norm": 0.5350157618522644,
      "learning_rate": 0.0001941521865767417,
      "loss": 1.3339,
      "step": 173
    },
    {
      "epoch": 0.006389996350540878,
      "grad_norm": 0.5004799365997314,
      "learning_rate": 0.0001940809326795256,
      "loss": 1.431,
      "step": 174
    },
    {
      "epoch": 0.0064267204674980085,
      "grad_norm": 0.46311405301094055,
      "learning_rate": 0.000194009260539328,
      "loss": 1.3441,
      "step": 175
    },
    {
      "epoch": 0.00646344458445514,
      "grad_norm": 0.5131816864013672,
      "learning_rate": 0.0001939371704747721,
      "loss": 1.3648,
      "step": 176
    },
    {
      "epoch": 0.006500168701412272,
      "grad_norm": 0.5859532952308655,
      "learning_rate": 0.00019386466280633906,
      "loss": 1.2863,
      "step": 177
    },
    {
      "epoch": 0.006536892818369403,
      "grad_norm": 0.5633041262626648,
      "learning_rate": 0.00019379173785636646,
      "loss": 1.1503,
      "step": 178
    },
    {
      "epoch": 0.006573616935326535,
      "grad_norm": 0.5649035573005676,
      "learning_rate": 0.000193718395949047,
      "loss": 1.491,
      "step": 179
    },
    {
      "epoch": 0.006610341052283667,
      "grad_norm": 0.5964656472206116,
      "learning_rate": 0.00019364463741042694,
      "loss": 1.1172,
      "step": 180
    },
    {
      "epoch": 0.006647065169240797,
      "grad_norm": 0.6336075067520142,
      "learning_rate": 0.00019357046256840473,
      "loss": 1.2763,
      "step": 181
    },
    {
      "epoch": 0.006683789286197929,
      "grad_norm": 0.6734123229980469,
      "learning_rate": 0.00019349587175272948,
      "loss": 1.2599,
      "step": 182
    },
    {
      "epoch": 0.0067205134031550606,
      "grad_norm": 0.706652820110321,
      "learning_rate": 0.0001934208652949996,
      "loss": 1.1628,
      "step": 183
    },
    {
      "epoch": 0.006757237520112192,
      "grad_norm": 0.7219396233558655,
      "learning_rate": 0.00019334544352866127,
      "loss": 1.4062,
      "step": 184
    },
    {
      "epoch": 0.006793961637069324,
      "grad_norm": 0.802462637424469,
      "learning_rate": 0.00019326960678900688,
      "loss": 1.0637,
      "step": 185
    },
    {
      "epoch": 0.006830685754026455,
      "grad_norm": 0.7378774285316467,
      "learning_rate": 0.00019319335541317361,
      "loss": 1.2669,
      "step": 186
    },
    {
      "epoch": 0.006867409870983587,
      "grad_norm": 0.8295735120773315,
      "learning_rate": 0.00019311668974014208,
      "loss": 1.3364,
      "step": 187
    },
    {
      "epoch": 0.006904133987940718,
      "grad_norm": 0.8445268273353577,
      "learning_rate": 0.00019303961011073447,
      "loss": 1.2424,
      "step": 188
    },
    {
      "epoch": 0.006940858104897849,
      "grad_norm": 0.8897864818572998,
      "learning_rate": 0.00019296211686761346,
      "loss": 1.5806,
      "step": 189
    },
    {
      "epoch": 0.006977582221854981,
      "grad_norm": 0.8445175886154175,
      "learning_rate": 0.00019288421035528028,
      "loss": 1.0133,
      "step": 190
    },
    {
      "epoch": 0.007014306338812113,
      "grad_norm": 0.996290922164917,
      "learning_rate": 0.00019280589092007352,
      "loss": 1.2268,
      "step": 191
    },
    {
      "epoch": 0.007051030455769244,
      "grad_norm": 0.991658627986908,
      "learning_rate": 0.00019272715891016735,
      "loss": 1.0738,
      "step": 192
    },
    {
      "epoch": 0.007087754572726376,
      "grad_norm": 1.2388745546340942,
      "learning_rate": 0.00019264801467557007,
      "loss": 1.0838,
      "step": 193
    },
    {
      "epoch": 0.0071244786896835065,
      "grad_norm": 1.130258321762085,
      "learning_rate": 0.00019256845856812266,
      "loss": 0.9665,
      "step": 194
    },
    {
      "epoch": 0.007161202806640638,
      "grad_norm": 1.1169642210006714,
      "learning_rate": 0.000192488490941497,
      "loss": 1.2167,
      "step": 195
    },
    {
      "epoch": 0.00719792692359777,
      "grad_norm": 1.2247196435928345,
      "learning_rate": 0.00019240811215119448,
      "loss": 1.1034,
      "step": 196
    },
    {
      "epoch": 0.007234651040554901,
      "grad_norm": 1.1018201112747192,
      "learning_rate": 0.00019232732255454422,
      "loss": 1.0397,
      "step": 197
    },
    {
      "epoch": 0.007271375157512033,
      "grad_norm": 1.2593382596969604,
      "learning_rate": 0.00019224612251070175,
      "loss": 1.2624,
      "step": 198
    },
    {
      "epoch": 0.007308099274469165,
      "grad_norm": 1.4708625078201294,
      "learning_rate": 0.0001921645123806472,
      "loss": 1.3028,
      "step": 199
    },
    {
      "epoch": 0.007344823391426296,
      "grad_norm": 2.4146103858947754,
      "learning_rate": 0.0001920824925271838,
      "loss": 1.3279,
      "step": 200
    },
    {
      "epoch": 0.007381547508383427,
      "grad_norm": 0.21593758463859558,
      "learning_rate": 0.0001920000633149362,
      "loss": 1.3266,
      "step": 201
    },
    {
      "epoch": 0.007418271625340559,
      "grad_norm": 0.32144275307655334,
      "learning_rate": 0.00019191722511034884,
      "loss": 1.8219,
      "step": 202
    },
    {
      "epoch": 0.00745499574229769,
      "grad_norm": 0.34392163157463074,
      "learning_rate": 0.00019183397828168448,
      "loss": 1.8002,
      "step": 203
    },
    {
      "epoch": 0.007491719859254822,
      "grad_norm": 0.3341860771179199,
      "learning_rate": 0.00019175032319902234,
      "loss": 1.5016,
      "step": 204
    },
    {
      "epoch": 0.007528443976211953,
      "grad_norm": 0.3221505284309387,
      "learning_rate": 0.00019166626023425662,
      "loss": 1.6324,
      "step": 205
    },
    {
      "epoch": 0.007565168093169085,
      "grad_norm": 0.3108210861682892,
      "learning_rate": 0.00019158178976109476,
      "loss": 1.5967,
      "step": 206
    },
    {
      "epoch": 0.007601892210126217,
      "grad_norm": 0.37417441606521606,
      "learning_rate": 0.0001914969121550558,
      "loss": 1.6194,
      "step": 207
    },
    {
      "epoch": 0.007638616327083347,
      "grad_norm": 0.35099324584007263,
      "learning_rate": 0.00019141162779346874,
      "loss": 1.6411,
      "step": 208
    },
    {
      "epoch": 0.007675340444040479,
      "grad_norm": 0.38678109645843506,
      "learning_rate": 0.00019132593705547082,
      "loss": 1.4553,
      "step": 209
    },
    {
      "epoch": 0.007712064560997611,
      "grad_norm": 0.41745078563690186,
      "learning_rate": 0.00019123984032200586,
      "loss": 1.5553,
      "step": 210
    },
    {
      "epoch": 0.007748788677954742,
      "grad_norm": 0.43167099356651306,
      "learning_rate": 0.00019115333797582254,
      "loss": 1.6822,
      "step": 211
    },
    {
      "epoch": 0.007785512794911874,
      "grad_norm": 0.3153819143772125,
      "learning_rate": 0.00019106643040147278,
      "loss": 1.4385,
      "step": 212
    },
    {
      "epoch": 0.007822236911869005,
      "grad_norm": 0.342204213142395,
      "learning_rate": 0.00019097911798530987,
      "loss": 1.6237,
      "step": 213
    },
    {
      "epoch": 0.007858961028826137,
      "grad_norm": 0.3589281439781189,
      "learning_rate": 0.00019089140111548696,
      "loss": 1.6182,
      "step": 214
    },
    {
      "epoch": 0.007895685145783269,
      "grad_norm": 0.39386796951293945,
      "learning_rate": 0.00019080328018195513,
      "loss": 1.3046,
      "step": 215
    },
    {
      "epoch": 0.0079324092627404,
      "grad_norm": 0.36143559217453003,
      "learning_rate": 0.0001907147555764618,
      "loss": 1.4274,
      "step": 216
    },
    {
      "epoch": 0.00796913337969753,
      "grad_norm": 0.37651386857032776,
      "learning_rate": 0.00019062582769254895,
      "loss": 1.5752,
      "step": 217
    },
    {
      "epoch": 0.008005857496654662,
      "grad_norm": 0.38396772742271423,
      "learning_rate": 0.00019053649692555135,
      "loss": 1.3698,
      "step": 218
    },
    {
      "epoch": 0.008042581613611793,
      "grad_norm": 0.4823744297027588,
      "learning_rate": 0.00019044676367259476,
      "loss": 1.3599,
      "step": 219
    },
    {
      "epoch": 0.008079305730568925,
      "grad_norm": 0.4878006875514984,
      "learning_rate": 0.00019035662833259432,
      "loss": 1.4728,
      "step": 220
    },
    {
      "epoch": 0.008116029847526057,
      "grad_norm": 0.47330114245414734,
      "learning_rate": 0.00019026609130625257,
      "loss": 1.3208,
      "step": 221
    },
    {
      "epoch": 0.008152753964483188,
      "grad_norm": 0.49724292755126953,
      "learning_rate": 0.00019017515299605788,
      "loss": 1.2265,
      "step": 222
    },
    {
      "epoch": 0.00818947808144032,
      "grad_norm": 0.46729010343551636,
      "learning_rate": 0.00019008381380628247,
      "loss": 1.3524,
      "step": 223
    },
    {
      "epoch": 0.008226202198397451,
      "grad_norm": 0.5844569206237793,
      "learning_rate": 0.00018999207414298067,
      "loss": 1.4708,
      "step": 224
    },
    {
      "epoch": 0.008262926315354583,
      "grad_norm": 0.5950741171836853,
      "learning_rate": 0.00018989993441398726,
      "loss": 1.3631,
      "step": 225
    },
    {
      "epoch": 0.008299650432311715,
      "grad_norm": 0.5740101933479309,
      "learning_rate": 0.00018980739502891546,
      "loss": 1.5136,
      "step": 226
    },
    {
      "epoch": 0.008336374549268846,
      "grad_norm": 0.4856874644756317,
      "learning_rate": 0.0001897144563991552,
      "loss": 1.4501,
      "step": 227
    },
    {
      "epoch": 0.008373098666225978,
      "grad_norm": 0.6637743711471558,
      "learning_rate": 0.00018962111893787128,
      "loss": 1.5492,
      "step": 228
    },
    {
      "epoch": 0.00840982278318311,
      "grad_norm": 0.5797656178474426,
      "learning_rate": 0.00018952738306000151,
      "loss": 1.1211,
      "step": 229
    },
    {
      "epoch": 0.008446546900140241,
      "grad_norm": 0.6373082995414734,
      "learning_rate": 0.00018943324918225494,
      "loss": 1.4005,
      "step": 230
    },
    {
      "epoch": 0.008483271017097371,
      "grad_norm": 0.6246806979179382,
      "learning_rate": 0.0001893387177231099,
      "loss": 1.4602,
      "step": 231
    },
    {
      "epoch": 0.008519995134054503,
      "grad_norm": 0.718519389629364,
      "learning_rate": 0.0001892437891028122,
      "loss": 1.1933,
      "step": 232
    },
    {
      "epoch": 0.008556719251011634,
      "grad_norm": 0.869508683681488,
      "learning_rate": 0.0001891484637433733,
      "loss": 1.384,
      "step": 233
    },
    {
      "epoch": 0.008593443367968766,
      "grad_norm": 0.7499548196792603,
      "learning_rate": 0.00018905274206856837,
      "loss": 1.1006,
      "step": 234
    },
    {
      "epoch": 0.008630167484925897,
      "grad_norm": 0.6303340196609497,
      "learning_rate": 0.00018895662450393438,
      "loss": 1.005,
      "step": 235
    },
    {
      "epoch": 0.008666891601883029,
      "grad_norm": 0.8143547773361206,
      "learning_rate": 0.00018886011147676833,
      "loss": 1.1803,
      "step": 236
    },
    {
      "epoch": 0.00870361571884016,
      "grad_norm": 0.6528473496437073,
      "learning_rate": 0.00018876320341612522,
      "loss": 1.0515,
      "step": 237
    },
    {
      "epoch": 0.008740339835797292,
      "grad_norm": 0.8889027833938599,
      "learning_rate": 0.00018866590075281624,
      "loss": 1.141,
      "step": 238
    },
    {
      "epoch": 0.008777063952754424,
      "grad_norm": 0.905524730682373,
      "learning_rate": 0.00018856820391940674,
      "loss": 0.9512,
      "step": 239
    },
    {
      "epoch": 0.008813788069711555,
      "grad_norm": 1.0086828470230103,
      "learning_rate": 0.00018847011335021449,
      "loss": 0.8296,
      "step": 240
    },
    {
      "epoch": 0.008850512186668687,
      "grad_norm": 1.0967466831207275,
      "learning_rate": 0.00018837162948130752,
      "loss": 1.4463,
      "step": 241
    },
    {
      "epoch": 0.008887236303625819,
      "grad_norm": 1.037026286125183,
      "learning_rate": 0.00018827275275050233,
      "loss": 1.1008,
      "step": 242
    },
    {
      "epoch": 0.00892396042058295,
      "grad_norm": 0.8901092410087585,
      "learning_rate": 0.00018817348359736203,
      "loss": 1.0084,
      "step": 243
    },
    {
      "epoch": 0.00896068453754008,
      "grad_norm": 1.1297985315322876,
      "learning_rate": 0.00018807382246319412,
      "loss": 1.0629,
      "step": 244
    },
    {
      "epoch": 0.008997408654497212,
      "grad_norm": 1.1219151020050049,
      "learning_rate": 0.00018797376979104872,
      "loss": 1.1223,
      "step": 245
    },
    {
      "epoch": 0.009034132771454343,
      "grad_norm": 1.319037675857544,
      "learning_rate": 0.00018787332602571662,
      "loss": 1.1906,
      "step": 246
    },
    {
      "epoch": 0.009070856888411475,
      "grad_norm": 1.1289669275283813,
      "learning_rate": 0.00018777249161372713,
      "loss": 1.1872,
      "step": 247
    },
    {
      "epoch": 0.009107581005368607,
      "grad_norm": 1.8450517654418945,
      "learning_rate": 0.00018767126700334634,
      "loss": 1.4198,
      "step": 248
    },
    {
      "epoch": 0.009144305122325738,
      "grad_norm": 1.2915611267089844,
      "learning_rate": 0.0001875696526445749,
      "loss": 1.1705,
      "step": 249
    },
    {
      "epoch": 0.00918102923928287,
      "grad_norm": 1.5633907318115234,
      "learning_rate": 0.0001874676489891461,
      "loss": 1.1454,
      "step": 250
    },
    {
      "epoch": 0.009217753356240001,
      "grad_norm": 0.21041421592235565,
      "learning_rate": 0.00018736525649052394,
      "loss": 1.7956,
      "step": 251
    },
    {
      "epoch": 0.009254477473197133,
      "grad_norm": 0.23667362332344055,
      "learning_rate": 0.00018726247560390099,
      "loss": 1.5941,
      "step": 252
    },
    {
      "epoch": 0.009291201590154265,
      "grad_norm": 0.2861754596233368,
      "learning_rate": 0.00018715930678619644,
      "loss": 1.5927,
      "step": 253
    },
    {
      "epoch": 0.009327925707111396,
      "grad_norm": 0.33346420526504517,
      "learning_rate": 0.00018705575049605413,
      "loss": 1.7128,
      "step": 254
    },
    {
      "epoch": 0.009364649824068528,
      "grad_norm": 0.31159916520118713,
      "learning_rate": 0.00018695180719384029,
      "loss": 1.5633,
      "step": 255
    },
    {
      "epoch": 0.00940137394102566,
      "grad_norm": 0.33967325091362,
      "learning_rate": 0.00018684747734164177,
      "loss": 1.7932,
      "step": 256
    },
    {
      "epoch": 0.00943809805798279,
      "grad_norm": 0.37867024540901184,
      "learning_rate": 0.00018674276140326376,
      "loss": 1.4598,
      "step": 257
    },
    {
      "epoch": 0.009474822174939921,
      "grad_norm": 0.29999667406082153,
      "learning_rate": 0.00018663765984422786,
      "loss": 1.2748,
      "step": 258
    },
    {
      "epoch": 0.009511546291897053,
      "grad_norm": 0.34820878505706787,
      "learning_rate": 0.00018653217313177004,
      "loss": 1.3325,
      "step": 259
    },
    {
      "epoch": 0.009548270408854184,
      "grad_norm": 0.37458816170692444,
      "learning_rate": 0.00018642630173483832,
      "loss": 1.6293,
      "step": 260
    },
    {
      "epoch": 0.009584994525811316,
      "grad_norm": 0.35919415950775146,
      "learning_rate": 0.00018632004612409103,
      "loss": 1.4407,
      "step": 261
    },
    {
      "epoch": 0.009621718642768447,
      "grad_norm": 0.31860947608947754,
      "learning_rate": 0.00018621340677189453,
      "loss": 1.4238,
      "step": 262
    },
    {
      "epoch": 0.009658442759725579,
      "grad_norm": 0.3451169729232788,
      "learning_rate": 0.00018610638415232097,
      "loss": 1.4651,
      "step": 263
    },
    {
      "epoch": 0.00969516687668271,
      "grad_norm": 0.34404557943344116,
      "learning_rate": 0.00018599897874114652,
      "loss": 1.4537,
      "step": 264
    },
    {
      "epoch": 0.009731890993639842,
      "grad_norm": 0.3328021466732025,
      "learning_rate": 0.00018589119101584898,
      "loss": 1.3376,
      "step": 265
    },
    {
      "epoch": 0.009768615110596974,
      "grad_norm": 0.366354763507843,
      "learning_rate": 0.00018578302145560584,
      "loss": 1.4524,
      "step": 266
    },
    {
      "epoch": 0.009805339227554105,
      "grad_norm": 0.36403486132621765,
      "learning_rate": 0.00018567447054129195,
      "loss": 1.4067,
      "step": 267
    },
    {
      "epoch": 0.009842063344511237,
      "grad_norm": 0.39728018641471863,
      "learning_rate": 0.00018556553875547754,
      "loss": 1.3385,
      "step": 268
    },
    {
      "epoch": 0.009878787461468369,
      "grad_norm": 0.4437379837036133,
      "learning_rate": 0.00018545622658242607,
      "loss": 1.4817,
      "step": 269
    },
    {
      "epoch": 0.009915511578425499,
      "grad_norm": 0.45578524470329285,
      "learning_rate": 0.00018534653450809197,
      "loss": 1.2447,
      "step": 270
    },
    {
      "epoch": 0.00995223569538263,
      "grad_norm": 0.4467160105705261,
      "learning_rate": 0.00018523646302011867,
      "loss": 1.3505,
      "step": 271
    },
    {
      "epoch": 0.009988959812339762,
      "grad_norm": 0.5396415591239929,
      "learning_rate": 0.00018512601260783606,
      "loss": 1.4303,
      "step": 272
    },
    {
      "epoch": 0.010025683929296893,
      "grad_norm": 0.5345490574836731,
      "learning_rate": 0.00018501518376225887,
      "loss": 1.4428,
      "step": 273
    },
    {
      "epoch": 0.010062408046254025,
      "grad_norm": 0.5290482044219971,
      "learning_rate": 0.00018490397697608395,
      "loss": 1.2741,
      "step": 274
    },
    {
      "epoch": 0.010099132163211157,
      "grad_norm": 0.5261471271514893,
      "learning_rate": 0.0001847923927436884,
      "loss": 1.3481,
      "step": 275
    },
    {
      "epoch": 0.010135856280168288,
      "grad_norm": 0.5443547368049622,
      "learning_rate": 0.00018468043156112728,
      "loss": 1.4052,
      "step": 276
    },
    {
      "epoch": 0.01017258039712542,
      "grad_norm": 0.6845529675483704,
      "learning_rate": 0.0001845680939261314,
      "loss": 1.3127,
      "step": 277
    },
    {
      "epoch": 0.010209304514082551,
      "grad_norm": 0.5482890009880066,
      "learning_rate": 0.00018445538033810515,
      "loss": 1.3504,
      "step": 278
    },
    {
      "epoch": 0.010246028631039683,
      "grad_norm": 0.5635014176368713,
      "learning_rate": 0.00018434229129812418,
      "loss": 0.9958,
      "step": 279
    },
    {
      "epoch": 0.010282752747996815,
      "grad_norm": 0.6552486419677734,
      "learning_rate": 0.0001842288273089332,
      "loss": 1.1248,
      "step": 280
    },
    {
      "epoch": 0.010319476864953946,
      "grad_norm": 0.6528019905090332,
      "learning_rate": 0.00018411498887494396,
      "loss": 1.2559,
      "step": 281
    },
    {
      "epoch": 0.010356200981911078,
      "grad_norm": 0.8731454610824585,
      "learning_rate": 0.00018400077650223263,
      "loss": 1.27,
      "step": 282
    },
    {
      "epoch": 0.010392925098868208,
      "grad_norm": 0.6625577211380005,
      "learning_rate": 0.0001838861906985379,
      "loss": 1.2527,
      "step": 283
    },
    {
      "epoch": 0.01042964921582534,
      "grad_norm": 0.6907861232757568,
      "learning_rate": 0.00018377123197325842,
      "loss": 1.042,
      "step": 284
    },
    {
      "epoch": 0.010466373332782471,
      "grad_norm": 0.858329713344574,
      "learning_rate": 0.00018365590083745085,
      "loss": 1.2466,
      "step": 285
    },
    {
      "epoch": 0.010503097449739603,
      "grad_norm": 0.9771177768707275,
      "learning_rate": 0.00018354019780382735,
      "loss": 1.0942,
      "step": 286
    },
    {
      "epoch": 0.010539821566696734,
      "grad_norm": 0.8735194802284241,
      "learning_rate": 0.0001834241233867533,
      "loss": 1.2432,
      "step": 287
    },
    {
      "epoch": 0.010576545683653866,
      "grad_norm": 0.8684554696083069,
      "learning_rate": 0.00018330767810224524,
      "loss": 1.3073,
      "step": 288
    },
    {
      "epoch": 0.010613269800610997,
      "grad_norm": 0.8898025751113892,
      "learning_rate": 0.0001831908624679683,
      "loss": 1.151,
      "step": 289
    },
    {
      "epoch": 0.010649993917568129,
      "grad_norm": 1.0928819179534912,
      "learning_rate": 0.0001830736770032341,
      "loss": 1.0203,
      "step": 290
    },
    {
      "epoch": 0.01068671803452526,
      "grad_norm": 1.0231902599334717,
      "learning_rate": 0.0001829561222289984,
      "loss": 1.0107,
      "step": 291
    },
    {
      "epoch": 0.010723442151482392,
      "grad_norm": 0.9310503602027893,
      "learning_rate": 0.00018283819866785853,
      "loss": 1.1733,
      "step": 292
    },
    {
      "epoch": 0.010760166268439524,
      "grad_norm": 1.2396451234817505,
      "learning_rate": 0.0001827199068440516,
      "loss": 1.4462,
      "step": 293
    },
    {
      "epoch": 0.010796890385396656,
      "grad_norm": 1.0380631685256958,
      "learning_rate": 0.00018260124728345162,
      "loss": 1.2387,
      "step": 294
    },
    {
      "epoch": 0.010833614502353787,
      "grad_norm": 1.0441265106201172,
      "learning_rate": 0.00018248222051356754,
      "loss": 0.9914,
      "step": 295
    },
    {
      "epoch": 0.010870338619310917,
      "grad_norm": 0.9357383847236633,
      "learning_rate": 0.00018236282706354063,
      "loss": 1.0629,
      "step": 296
    },
    {
      "epoch": 0.010907062736268049,
      "grad_norm": 1.2338109016418457,
      "learning_rate": 0.00018224306746414238,
      "loss": 1.1174,
      "step": 297
    },
    {
      "epoch": 0.01094378685322518,
      "grad_norm": 1.3758505582809448,
      "learning_rate": 0.00018212294224777197,
      "loss": 1.3553,
      "step": 298
    },
    {
      "epoch": 0.010980510970182312,
      "grad_norm": 1.25632643699646,
      "learning_rate": 0.00018200245194845399,
      "loss": 1.0981,
      "step": 299
    },
    {
      "epoch": 0.011017235087139443,
      "grad_norm": 1.888166069984436,
      "learning_rate": 0.00018188159710183594,
      "loss": 1.3899,
      "step": 300
    },
    {
      "epoch": 0.011053959204096575,
      "grad_norm": 0.19148558378219604,
      "learning_rate": 0.000181760378245186,
      "loss": 1.4595,
      "step": 301
    },
    {
      "epoch": 0.011090683321053707,
      "grad_norm": 0.25866252183914185,
      "learning_rate": 0.00018163879591739067,
      "loss": 1.6216,
      "step": 302
    },
    {
      "epoch": 0.011127407438010838,
      "grad_norm": 0.29589152336120605,
      "learning_rate": 0.0001815168506589521,
      "loss": 1.7321,
      "step": 303
    },
    {
      "epoch": 0.01116413155496797,
      "grad_norm": 0.34040263295173645,
      "learning_rate": 0.000181394543011986,
      "loss": 1.7168,
      "step": 304
    },
    {
      "epoch": 0.011200855671925102,
      "grad_norm": 0.3252499997615814,
      "learning_rate": 0.00018127187352021907,
      "loss": 1.6335,
      "step": 305
    },
    {
      "epoch": 0.011237579788882233,
      "grad_norm": 0.3420853316783905,
      "learning_rate": 0.0001811488427289866,
      "loss": 1.4776,
      "step": 306
    },
    {
      "epoch": 0.011274303905839365,
      "grad_norm": 0.32042768597602844,
      "learning_rate": 0.00018102545118523007,
      "loss": 1.4981,
      "step": 307
    },
    {
      "epoch": 0.011311028022796496,
      "grad_norm": 0.38823747634887695,
      "learning_rate": 0.00018090169943749476,
      "loss": 1.5228,
      "step": 308
    },
    {
      "epoch": 0.011347752139753626,
      "grad_norm": 0.32528162002563477,
      "learning_rate": 0.00018077758803592718,
      "loss": 1.4219,
      "step": 309
    },
    {
      "epoch": 0.011384476256710758,
      "grad_norm": 0.43534666299819946,
      "learning_rate": 0.00018065311753227273,
      "loss": 1.4017,
      "step": 310
    },
    {
      "epoch": 0.01142120037366789,
      "grad_norm": 0.3710077404975891,
      "learning_rate": 0.0001805282884798732,
      "loss": 1.3608,
      "step": 311
    },
    {
      "epoch": 0.011457924490625021,
      "grad_norm": 0.36862894892692566,
      "learning_rate": 0.00018040310143366446,
      "loss": 1.466,
      "step": 312
    },
    {
      "epoch": 0.011494648607582153,
      "grad_norm": 0.3496580719947815,
      "learning_rate": 0.00018027755695017368,
      "loss": 1.3394,
      "step": 313
    },
    {
      "epoch": 0.011531372724539284,
      "grad_norm": 0.3310506045818329,
      "learning_rate": 0.00018015165558751717,
      "loss": 1.5432,
      "step": 314
    },
    {
      "epoch": 0.011568096841496416,
      "grad_norm": 0.3637956976890564,
      "learning_rate": 0.00018002539790539773,
      "loss": 1.6284,
      "step": 315
    },
    {
      "epoch": 0.011604820958453547,
      "grad_norm": 0.35876187682151794,
      "learning_rate": 0.00017989878446510215,
      "loss": 1.4233,
      "step": 316
    },
    {
      "epoch": 0.011641545075410679,
      "grad_norm": 0.3751535415649414,
      "learning_rate": 0.00017977181582949888,
      "loss": 1.4608,
      "step": 317
    },
    {
      "epoch": 0.01167826919236781,
      "grad_norm": 0.39340418577194214,
      "learning_rate": 0.0001796444925630353,
      "loss": 1.4141,
      "step": 318
    },
    {
      "epoch": 0.011714993309324942,
      "grad_norm": 0.42349880933761597,
      "learning_rate": 0.00017951681523173542,
      "loss": 1.5051,
      "step": 319
    },
    {
      "epoch": 0.011751717426282074,
      "grad_norm": 0.40073323249816895,
      "learning_rate": 0.0001793887844031972,
      "loss": 1.2607,
      "step": 320
    },
    {
      "epoch": 0.011788441543239206,
      "grad_norm": 0.4488166868686676,
      "learning_rate": 0.00017926040064659014,
      "loss": 1.3061,
      "step": 321
    },
    {
      "epoch": 0.011825165660196337,
      "grad_norm": 0.4523330628871918,
      "learning_rate": 0.0001791316645326526,
      "loss": 1.2599,
      "step": 322
    },
    {
      "epoch": 0.011861889777153467,
      "grad_norm": 0.5272870659828186,
      "learning_rate": 0.00017900257663368963,
      "loss": 1.4327,
      "step": 323
    },
    {
      "epoch": 0.011898613894110599,
      "grad_norm": 0.524665892124176,
      "learning_rate": 0.0001788731375235698,
      "loss": 1.4883,
      "step": 324
    },
    {
      "epoch": 0.01193533801106773,
      "grad_norm": 0.559048593044281,
      "learning_rate": 0.00017874334777772327,
      "loss": 1.2511,
      "step": 325
    },
    {
      "epoch": 0.011972062128024862,
      "grad_norm": 0.5605241060256958,
      "learning_rate": 0.00017861320797313892,
      "loss": 1.423,
      "step": 326
    },
    {
      "epoch": 0.012008786244981993,
      "grad_norm": 0.5054120421409607,
      "learning_rate": 0.0001784827186883618,
      "loss": 1.2819,
      "step": 327
    },
    {
      "epoch": 0.012045510361939125,
      "grad_norm": 0.525787353515625,
      "learning_rate": 0.00017835188050349064,
      "loss": 1.2644,
      "step": 328
    },
    {
      "epoch": 0.012082234478896257,
      "grad_norm": 0.5941430330276489,
      "learning_rate": 0.00017822069400017516,
      "loss": 1.2731,
      "step": 329
    },
    {
      "epoch": 0.012118958595853388,
      "grad_norm": 0.6347975730895996,
      "learning_rate": 0.00017808915976161362,
      "loss": 1.447,
      "step": 330
    },
    {
      "epoch": 0.01215568271281052,
      "grad_norm": 0.6294612288475037,
      "learning_rate": 0.00017795727837255015,
      "loss": 1.4093,
      "step": 331
    },
    {
      "epoch": 0.012192406829767652,
      "grad_norm": 0.6097142696380615,
      "learning_rate": 0.00017782505041927216,
      "loss": 1.0599,
      "step": 332
    },
    {
      "epoch": 0.012229130946724783,
      "grad_norm": 0.7774625420570374,
      "learning_rate": 0.00017769247648960774,
      "loss": 1.3103,
      "step": 333
    },
    {
      "epoch": 0.012265855063681915,
      "grad_norm": 0.6937221884727478,
      "learning_rate": 0.00017755955717292296,
      "loss": 1.1891,
      "step": 334
    },
    {
      "epoch": 0.012302579180639046,
      "grad_norm": 0.7517911791801453,
      "learning_rate": 0.00017742629306011944,
      "loss": 1.4556,
      "step": 335
    },
    {
      "epoch": 0.012339303297596176,
      "grad_norm": 0.7351686954498291,
      "learning_rate": 0.00017729268474363154,
      "loss": 1.1185,
      "step": 336
    },
    {
      "epoch": 0.012376027414553308,
      "grad_norm": 0.7667162418365479,
      "learning_rate": 0.0001771587328174239,
      "loss": 1.2215,
      "step": 337
    },
    {
      "epoch": 0.01241275153151044,
      "grad_norm": 0.961482584476471,
      "learning_rate": 0.0001770244378769885,
      "loss": 1.0115,
      "step": 338
    },
    {
      "epoch": 0.012449475648467571,
      "grad_norm": 0.8362442851066589,
      "learning_rate": 0.0001768898005193425,
      "loss": 1.1829,
      "step": 339
    },
    {
      "epoch": 0.012486199765424703,
      "grad_norm": 0.8960280418395996,
      "learning_rate": 0.000176754821343025,
      "loss": 1.4065,
      "step": 340
    },
    {
      "epoch": 0.012522923882381834,
      "grad_norm": 0.9358709454536438,
      "learning_rate": 0.0001766195009480949,
      "loss": 1.3015,
      "step": 341
    },
    {
      "epoch": 0.012559647999338966,
      "grad_norm": 0.9351969957351685,
      "learning_rate": 0.0001764838399361279,
      "loss": 1.13,
      "step": 342
    },
    {
      "epoch": 0.012596372116296098,
      "grad_norm": 1.611729621887207,
      "learning_rate": 0.00017634783891021393,
      "loss": 1.0757,
      "step": 343
    },
    {
      "epoch": 0.01263309623325323,
      "grad_norm": 0.9735264182090759,
      "learning_rate": 0.00017621149847495458,
      "loss": 1.3144,
      "step": 344
    },
    {
      "epoch": 0.01266982035021036,
      "grad_norm": 1.1304185390472412,
      "learning_rate": 0.00017607481923646016,
      "loss": 0.9156,
      "step": 345
    },
    {
      "epoch": 0.012706544467167492,
      "grad_norm": 1.2708826065063477,
      "learning_rate": 0.0001759378018023473,
      "loss": 0.9941,
      "step": 346
    },
    {
      "epoch": 0.012743268584124624,
      "grad_norm": 1.0765823125839233,
      "learning_rate": 0.00017580044678173592,
      "loss": 1.0142,
      "step": 347
    },
    {
      "epoch": 0.012779992701081756,
      "grad_norm": 1.248100996017456,
      "learning_rate": 0.00017566275478524693,
      "loss": 1.4131,
      "step": 348
    },
    {
      "epoch": 0.012816716818038885,
      "grad_norm": 1.4159555435180664,
      "learning_rate": 0.0001755247264249991,
      "loss": 1.3563,
      "step": 349
    },
    {
      "epoch": 0.012853440934996017,
      "grad_norm": 1.8039625883102417,
      "learning_rate": 0.0001753863623146066,
      "loss": 1.2605,
      "step": 350
    },
    {
      "epoch": 0.012890165051953149,
      "grad_norm": 0.19830720126628876,
      "learning_rate": 0.00017524766306917618,
      "loss": 1.5317,
      "step": 351
    },
    {
      "epoch": 0.01292688916891028,
      "grad_norm": 0.28898677229881287,
      "learning_rate": 0.0001751086293053045,
      "loss": 1.8923,
      "step": 352
    },
    {
      "epoch": 0.012963613285867412,
      "grad_norm": 0.32439830899238586,
      "learning_rate": 0.0001749692616410753,
      "loss": 1.5959,
      "step": 353
    },
    {
      "epoch": 0.013000337402824544,
      "grad_norm": 0.2880374491214752,
      "learning_rate": 0.00017482956069605668,
      "loss": 1.7155,
      "step": 354
    },
    {
      "epoch": 0.013037061519781675,
      "grad_norm": 0.32204726338386536,
      "learning_rate": 0.00017468952709129846,
      "loss": 1.6509,
      "step": 355
    },
    {
      "epoch": 0.013073785636738807,
      "grad_norm": 0.3474827706813812,
      "learning_rate": 0.00017454916144932922,
      "loss": 1.723,
      "step": 356
    },
    {
      "epoch": 0.013110509753695938,
      "grad_norm": 0.3080861270427704,
      "learning_rate": 0.0001744084643941536,
      "loss": 1.4913,
      "step": 357
    },
    {
      "epoch": 0.01314723387065307,
      "grad_norm": 0.3363988697528839,
      "learning_rate": 0.00017426743655124974,
      "loss": 1.4274,
      "step": 358
    },
    {
      "epoch": 0.013183957987610202,
      "grad_norm": 0.38495108485221863,
      "learning_rate": 0.0001741260785475661,
      "loss": 1.455,
      "step": 359
    },
    {
      "epoch": 0.013220682104567333,
      "grad_norm": 0.35908249020576477,
      "learning_rate": 0.00017398439101151905,
      "loss": 1.5102,
      "step": 360
    },
    {
      "epoch": 0.013257406221524465,
      "grad_norm": 0.344539999961853,
      "learning_rate": 0.00017384237457298987,
      "loss": 1.4615,
      "step": 361
    },
    {
      "epoch": 0.013294130338481595,
      "grad_norm": 0.34393495321273804,
      "learning_rate": 0.00017370002986332193,
      "loss": 1.3628,
      "step": 362
    },
    {
      "epoch": 0.013330854455438726,
      "grad_norm": 0.3426048457622528,
      "learning_rate": 0.00017355735751531807,
      "loss": 1.391,
      "step": 363
    },
    {
      "epoch": 0.013367578572395858,
      "grad_norm": 0.3551957607269287,
      "learning_rate": 0.00017341435816323756,
      "loss": 1.5873,
      "step": 364
    },
    {
      "epoch": 0.01340430268935299,
      "grad_norm": 0.45659440755844116,
      "learning_rate": 0.00017327103244279348,
      "loss": 1.5477,
      "step": 365
    },
    {
      "epoch": 0.013441026806310121,
      "grad_norm": 0.3504604399204254,
      "learning_rate": 0.00017312738099114973,
      "loss": 1.4575,
      "step": 366
    },
    {
      "epoch": 0.013477750923267253,
      "grad_norm": 0.3787093162536621,
      "learning_rate": 0.00017298340444691835,
      "loss": 1.4236,
      "step": 367
    },
    {
      "epoch": 0.013514475040224384,
      "grad_norm": 0.3764216899871826,
      "learning_rate": 0.00017283910345015647,
      "loss": 1.4799,
      "step": 368
    },
    {
      "epoch": 0.013551199157181516,
      "grad_norm": 0.4056175947189331,
      "learning_rate": 0.0001726944786423637,
      "loss": 1.2662,
      "step": 369
    },
    {
      "epoch": 0.013587923274138648,
      "grad_norm": 0.43907254934310913,
      "learning_rate": 0.00017254953066647913,
      "loss": 1.2174,
      "step": 370
    },
    {
      "epoch": 0.01362464739109578,
      "grad_norm": 0.48604634404182434,
      "learning_rate": 0.00017240426016687863,
      "loss": 1.5519,
      "step": 371
    },
    {
      "epoch": 0.01366137150805291,
      "grad_norm": 0.45083242654800415,
      "learning_rate": 0.00017225866778937165,
      "loss": 1.4578,
      "step": 372
    },
    {
      "epoch": 0.013698095625010042,
      "grad_norm": 0.5116049647331238,
      "learning_rate": 0.00017211275418119876,
      "loss": 1.3196,
      "step": 373
    },
    {
      "epoch": 0.013734819741967174,
      "grad_norm": 0.5021712183952332,
      "learning_rate": 0.0001719665199910285,
      "loss": 1.318,
      "step": 374
    },
    {
      "epoch": 0.013771543858924304,
      "grad_norm": 0.5341774225234985,
      "learning_rate": 0.00017181996586895454,
      "loss": 1.3344,
      "step": 375
    },
    {
      "epoch": 0.013771543858924304,
      "eval_loss": 1.3082693815231323,
      "eval_runtime": 1107.4453,
      "eval_samples_per_second": 41.412,
      "eval_steps_per_second": 20.706,
      "step": 375
    },
    {
      "epoch": 0.013808267975881435,
      "grad_norm": 0.6293779611587524,
      "learning_rate": 0.00017167309246649297,
      "loss": 1.4041,
      "step": 376
    },
    {
      "epoch": 0.013844992092838567,
      "grad_norm": 0.5708191990852356,
      "learning_rate": 0.0001715259004365791,
      "loss": 1.1751,
      "step": 377
    },
    {
      "epoch": 0.013881716209795699,
      "grad_norm": 0.615359365940094,
      "learning_rate": 0.00017137839043356484,
      "loss": 1.35,
      "step": 378
    },
    {
      "epoch": 0.01391844032675283,
      "grad_norm": 0.6823323369026184,
      "learning_rate": 0.00017123056311321562,
      "loss": 1.413,
      "step": 379
    },
    {
      "epoch": 0.013955164443709962,
      "grad_norm": 0.7502759695053101,
      "learning_rate": 0.0001710824191327075,
      "loss": 0.9556,
      "step": 380
    },
    {
      "epoch": 0.013991888560667094,
      "grad_norm": 0.9459863305091858,
      "learning_rate": 0.00017093395915062428,
      "loss": 1.0845,
      "step": 381
    },
    {
      "epoch": 0.014028612677624225,
      "grad_norm": 0.709890604019165,
      "learning_rate": 0.00017078518382695465,
      "loss": 1.0385,
      "step": 382
    },
    {
      "epoch": 0.014065336794581357,
      "grad_norm": 0.7171956896781921,
      "learning_rate": 0.00017063609382308908,
      "loss": 1.4663,
      "step": 383
    },
    {
      "epoch": 0.014102060911538488,
      "grad_norm": 0.6555240154266357,
      "learning_rate": 0.00017048668980181698,
      "loss": 1.3887,
      "step": 384
    },
    {
      "epoch": 0.01413878502849562,
      "grad_norm": 0.8395636081695557,
      "learning_rate": 0.00017033697242732377,
      "loss": 1.3444,
      "step": 385
    },
    {
      "epoch": 0.014175509145452752,
      "grad_norm": 0.7988017201423645,
      "learning_rate": 0.0001701869423651879,
      "loss": 1.3596,
      "step": 386
    },
    {
      "epoch": 0.014212233262409883,
      "grad_norm": 0.9169744849205017,
      "learning_rate": 0.00017003660028237793,
      "loss": 1.2652,
      "step": 387
    },
    {
      "epoch": 0.014248957379367013,
      "grad_norm": 0.8666754364967346,
      "learning_rate": 0.00016988594684724947,
      "loss": 1.1979,
      "step": 388
    },
    {
      "epoch": 0.014285681496324145,
      "grad_norm": 0.8198935985565186,
      "learning_rate": 0.00016973498272954222,
      "loss": 1.2049,
      "step": 389
    },
    {
      "epoch": 0.014322405613281276,
      "grad_norm": 0.8048646450042725,
      "learning_rate": 0.00016958370860037717,
      "loss": 1.1675,
      "step": 390
    },
    {
      "epoch": 0.014359129730238408,
      "grad_norm": 0.9369916319847107,
      "learning_rate": 0.00016943212513225345,
      "loss": 1.2479,
      "step": 391
    },
    {
      "epoch": 0.01439585384719554,
      "grad_norm": 0.9568859338760376,
      "learning_rate": 0.00016928023299904533,
      "loss": 1.0459,
      "step": 392
    },
    {
      "epoch": 0.014432577964152671,
      "grad_norm": 1.0061014890670776,
      "learning_rate": 0.0001691280328759992,
      "loss": 1.0307,
      "step": 393
    },
    {
      "epoch": 0.014469302081109803,
      "grad_norm": 1.1863815784454346,
      "learning_rate": 0.00016897552543973084,
      "loss": 1.4652,
      "step": 394
    },
    {
      "epoch": 0.014506026198066934,
      "grad_norm": 0.9680814743041992,
      "learning_rate": 0.00016882271136822206,
      "loss": 0.969,
      "step": 395
    },
    {
      "epoch": 0.014542750315024066,
      "grad_norm": 0.9966139793395996,
      "learning_rate": 0.0001686695913408179,
      "loss": 0.6784,
      "step": 396
    },
    {
      "epoch": 0.014579474431981198,
      "grad_norm": 1.4734877347946167,
      "learning_rate": 0.0001685161660382235,
      "loss": 1.4118,
      "step": 397
    },
    {
      "epoch": 0.01461619854893833,
      "grad_norm": 1.12624192237854,
      "learning_rate": 0.00016836243614250113,
      "loss": 0.9607,
      "step": 398
    },
    {
      "epoch": 0.01465292266589546,
      "grad_norm": 1.1937803030014038,
      "learning_rate": 0.00016820840233706719,
      "loss": 1.012,
      "step": 399
    },
    {
      "epoch": 0.014689646782852592,
      "grad_norm": 2.4200687408447266,
      "learning_rate": 0.0001680540653066891,
      "loss": 1.1553,
      "step": 400
    },
    {
      "epoch": 0.014726370899809722,
      "grad_norm": 0.16357336938381195,
      "learning_rate": 0.00016789942573748232,
      "loss": 1.538,
      "step": 401
    },
    {
      "epoch": 0.014763095016766854,
      "grad_norm": 0.1853504478931427,
      "learning_rate": 0.0001677444843169072,
      "loss": 1.4081,
      "step": 402
    },
    {
      "epoch": 0.014799819133723986,
      "grad_norm": 0.2039957493543625,
      "learning_rate": 0.00016758924173376603,
      "loss": 1.5763,
      "step": 403
    },
    {
      "epoch": 0.014836543250681117,
      "grad_norm": 0.2490047961473465,
      "learning_rate": 0.0001674336986781999,
      "loss": 1.5742,
      "step": 404
    },
    {
      "epoch": 0.014873267367638249,
      "grad_norm": 0.24538454413414001,
      "learning_rate": 0.00016727785584168581,
      "loss": 1.6601,
      "step": 405
    },
    {
      "epoch": 0.01490999148459538,
      "grad_norm": 0.2629493772983551,
      "learning_rate": 0.0001671217139170333,
      "loss": 1.5322,
      "step": 406
    },
    {
      "epoch": 0.014946715601552512,
      "grad_norm": 0.2834603786468506,
      "learning_rate": 0.00016696527359838154,
      "loss": 1.5609,
      "step": 407
    },
    {
      "epoch": 0.014983439718509644,
      "grad_norm": 0.2989986836910248,
      "learning_rate": 0.00016680853558119632,
      "loss": 1.5298,
      "step": 408
    },
    {
      "epoch": 0.015020163835466775,
      "grad_norm": 0.3203519582748413,
      "learning_rate": 0.0001666515005622668,
      "loss": 1.3994,
      "step": 409
    },
    {
      "epoch": 0.015056887952423907,
      "grad_norm": 0.36469900608062744,
      "learning_rate": 0.0001664941692397025,
      "loss": 1.5907,
      "step": 410
    },
    {
      "epoch": 0.015093612069381038,
      "grad_norm": 0.32446998357772827,
      "learning_rate": 0.00016633654231293013,
      "loss": 1.312,
      "step": 411
    },
    {
      "epoch": 0.01513033618633817,
      "grad_norm": 0.3682743012905121,
      "learning_rate": 0.00016617862048269065,
      "loss": 1.5026,
      "step": 412
    },
    {
      "epoch": 0.015167060303295302,
      "grad_norm": 0.37367111444473267,
      "learning_rate": 0.00016602040445103588,
      "loss": 1.6049,
      "step": 413
    },
    {
      "epoch": 0.015203784420252433,
      "grad_norm": 0.3751118779182434,
      "learning_rate": 0.00016586189492132566,
      "loss": 1.4049,
      "step": 414
    },
    {
      "epoch": 0.015240508537209563,
      "grad_norm": 0.346225380897522,
      "learning_rate": 0.00016570309259822453,
      "loss": 1.3977,
      "step": 415
    },
    {
      "epoch": 0.015277232654166695,
      "grad_norm": 0.3749098777770996,
      "learning_rate": 0.0001655439981876987,
      "loss": 1.3914,
      "step": 416
    },
    {
      "epoch": 0.015313956771123826,
      "grad_norm": 0.3564872443675995,
      "learning_rate": 0.00016538461239701277,
      "loss": 1.2481,
      "step": 417
    },
    {
      "epoch": 0.015350680888080958,
      "grad_norm": 0.3836904764175415,
      "learning_rate": 0.00016522493593472683,
      "loss": 1.5005,
      "step": 418
    },
    {
      "epoch": 0.01538740500503809,
      "grad_norm": 0.3882698118686676,
      "learning_rate": 0.0001650649695106931,
      "loss": 1.4291,
      "step": 419
    },
    {
      "epoch": 0.015424129121995221,
      "grad_norm": 0.41749534010887146,
      "learning_rate": 0.00016490471383605288,
      "loss": 1.307,
      "step": 420
    },
    {
      "epoch": 0.015460853238952353,
      "grad_norm": 0.48331576585769653,
      "learning_rate": 0.00016474416962323325,
      "loss": 1.4176,
      "step": 421
    },
    {
      "epoch": 0.015497577355909484,
      "grad_norm": 0.44741758704185486,
      "learning_rate": 0.00016458333758594414,
      "loss": 1.4716,
      "step": 422
    },
    {
      "epoch": 0.015534301472866616,
      "grad_norm": 0.4632352590560913,
      "learning_rate": 0.00016442221843917496,
      "loss": 1.3569,
      "step": 423
    },
    {
      "epoch": 0.015571025589823748,
      "grad_norm": 0.5382906794548035,
      "learning_rate": 0.00016426081289919143,
      "loss": 1.3076,
      "step": 424
    },
    {
      "epoch": 0.01560774970678088,
      "grad_norm": 0.5663718581199646,
      "learning_rate": 0.0001640991216835326,
      "loss": 1.1943,
      "step": 425
    },
    {
      "epoch": 0.01564447382373801,
      "grad_norm": 0.5091210007667542,
      "learning_rate": 0.00016393714551100734,
      "loss": 1.5461,
      "step": 426
    },
    {
      "epoch": 0.01568119794069514,
      "grad_norm": 0.5791824460029602,
      "learning_rate": 0.0001637748851016914,
      "loss": 1.4127,
      "step": 427
    },
    {
      "epoch": 0.015717922057652274,
      "grad_norm": 0.6630942225456238,
      "learning_rate": 0.00016361234117692413,
      "loss": 1.428,
      "step": 428
    },
    {
      "epoch": 0.015754646174609404,
      "grad_norm": 0.6061805486679077,
      "learning_rate": 0.00016344951445930526,
      "loss": 1.2227,
      "step": 429
    },
    {
      "epoch": 0.015791370291566537,
      "grad_norm": 0.6326902508735657,
      "learning_rate": 0.0001632864056726917,
      "loss": 1.243,
      "step": 430
    },
    {
      "epoch": 0.015828094408523667,
      "grad_norm": 0.5544697642326355,
      "learning_rate": 0.00016312301554219426,
      "loss": 1.2969,
      "step": 431
    },
    {
      "epoch": 0.0158648185254808,
      "grad_norm": 0.7134488224983215,
      "learning_rate": 0.00016295934479417453,
      "loss": 1.3122,
      "step": 432
    },
    {
      "epoch": 0.01590154264243793,
      "grad_norm": 0.6425052285194397,
      "learning_rate": 0.00016279539415624164,
      "loss": 0.9618,
      "step": 433
    },
    {
      "epoch": 0.01593826675939506,
      "grad_norm": 0.6341208815574646,
      "learning_rate": 0.0001626311643572489,
      "loss": 1.0879,
      "step": 434
    },
    {
      "epoch": 0.015974990876352194,
      "grad_norm": 0.8205564618110657,
      "learning_rate": 0.00016246665612729074,
      "loss": 0.9603,
      "step": 435
    },
    {
      "epoch": 0.016011714993309323,
      "grad_norm": 0.6695818305015564,
      "learning_rate": 0.00016230187019769928,
      "loss": 1.2907,
      "step": 436
    },
    {
      "epoch": 0.016048439110266457,
      "grad_norm": 0.8008434772491455,
      "learning_rate": 0.00016213680730104124,
      "loss": 1.1543,
      "step": 437
    },
    {
      "epoch": 0.016085163227223587,
      "grad_norm": 0.71556556224823,
      "learning_rate": 0.0001619714681711146,
      "loss": 0.963,
      "step": 438
    },
    {
      "epoch": 0.01612188734418072,
      "grad_norm": 1.0705739259719849,
      "learning_rate": 0.00016180585354294536,
      "loss": 0.9378,
      "step": 439
    },
    {
      "epoch": 0.01615861146113785,
      "grad_norm": 0.9722197651863098,
      "learning_rate": 0.00016163996415278424,
      "loss": 1.1929,
      "step": 440
    },
    {
      "epoch": 0.016195335578094983,
      "grad_norm": 0.8566421866416931,
      "learning_rate": 0.00016147380073810346,
      "loss": 0.9153,
      "step": 441
    },
    {
      "epoch": 0.016232059695052113,
      "grad_norm": 1.011698842048645,
      "learning_rate": 0.0001613073640375934,
      "loss": 0.9762,
      "step": 442
    },
    {
      "epoch": 0.016268783812009247,
      "grad_norm": 0.8080694675445557,
      "learning_rate": 0.00016114065479115946,
      "loss": 0.8245,
      "step": 443
    },
    {
      "epoch": 0.016305507928966376,
      "grad_norm": 1.1907299757003784,
      "learning_rate": 0.00016097367373991842,
      "loss": 1.4124,
      "step": 444
    },
    {
      "epoch": 0.01634223204592351,
      "grad_norm": 1.0441313982009888,
      "learning_rate": 0.00016080642162619565,
      "loss": 1.3574,
      "step": 445
    },
    {
      "epoch": 0.01637895616288064,
      "grad_norm": 1.0910389423370361,
      "learning_rate": 0.0001606388991935214,
      "loss": 1.0868,
      "step": 446
    },
    {
      "epoch": 0.016415680279837773,
      "grad_norm": 1.2694731950759888,
      "learning_rate": 0.0001604711071866277,
      "loss": 1.4078,
      "step": 447
    },
    {
      "epoch": 0.016452404396794903,
      "grad_norm": 1.233826756477356,
      "learning_rate": 0.00016030304635144494,
      "loss": 1.076,
      "step": 448
    },
    {
      "epoch": 0.016489128513752033,
      "grad_norm": 1.4833847284317017,
      "learning_rate": 0.00016013471743509862,
      "loss": 1.1485,
      "step": 449
    },
    {
      "epoch": 0.016525852630709166,
      "grad_norm": 1.6127415895462036,
      "learning_rate": 0.00015996612118590603,
      "loss": 1.1406,
      "step": 450
    },
    {
      "epoch": 0.016562576747666296,
      "grad_norm": 0.1769523173570633,
      "learning_rate": 0.00015979725835337294,
      "loss": 1.5624,
      "step": 451
    },
    {
      "epoch": 0.01659930086462343,
      "grad_norm": 0.19608841836452484,
      "learning_rate": 0.00015962812968819016,
      "loss": 1.6618,
      "step": 452
    },
    {
      "epoch": 0.01663602498158056,
      "grad_norm": 0.2193925380706787,
      "learning_rate": 0.0001594587359422303,
      "loss": 1.5073,
      "step": 453
    },
    {
      "epoch": 0.016672749098537692,
      "grad_norm": 0.24558556079864502,
      "learning_rate": 0.0001592890778685444,
      "loss": 1.5806,
      "step": 454
    },
    {
      "epoch": 0.016709473215494822,
      "grad_norm": 0.24949359893798828,
      "learning_rate": 0.00015911915622135862,
      "loss": 1.5047,
      "step": 455
    },
    {
      "epoch": 0.016746197332451956,
      "grad_norm": 0.25603586435317993,
      "learning_rate": 0.00015894897175607086,
      "loss": 1.5165,
      "step": 456
    },
    {
      "epoch": 0.016782921449409086,
      "grad_norm": 0.3430022895336151,
      "learning_rate": 0.00015877852522924732,
      "loss": 1.528,
      "step": 457
    },
    {
      "epoch": 0.01681964556636622,
      "grad_norm": 0.2846647799015045,
      "learning_rate": 0.00015860781739861928,
      "loss": 1.5558,
      "step": 458
    },
    {
      "epoch": 0.01685636968332335,
      "grad_norm": 0.28785619139671326,
      "learning_rate": 0.00015843684902307962,
      "loss": 1.5034,
      "step": 459
    },
    {
      "epoch": 0.016893093800280482,
      "grad_norm": 0.35725685954093933,
      "learning_rate": 0.00015826562086267956,
      "loss": 1.5568,
      "step": 460
    },
    {
      "epoch": 0.016929817917237612,
      "grad_norm": 0.3280180096626282,
      "learning_rate": 0.00015809413367862512,
      "loss": 1.6265,
      "step": 461
    },
    {
      "epoch": 0.016966542034194742,
      "grad_norm": 0.3238905072212219,
      "learning_rate": 0.00015792238823327388,
      "loss": 1.39,
      "step": 462
    },
    {
      "epoch": 0.017003266151151875,
      "grad_norm": 0.40948182344436646,
      "learning_rate": 0.00015775038529013152,
      "loss": 1.354,
      "step": 463
    },
    {
      "epoch": 0.017039990268109005,
      "grad_norm": 0.3718741536140442,
      "learning_rate": 0.0001575781256138485,
      "loss": 1.3932,
      "step": 464
    },
    {
      "epoch": 0.01707671438506614,
      "grad_norm": 0.3844534754753113,
      "learning_rate": 0.00015740560997021648,
      "loss": 1.3301,
      "step": 465
    },
    {
      "epoch": 0.01711343850202327,
      "grad_norm": 0.3812541663646698,
      "learning_rate": 0.00015723283912616513,
      "loss": 1.386,
      "step": 466
    },
    {
      "epoch": 0.0171501626189804,
      "grad_norm": 0.3776322305202484,
      "learning_rate": 0.00015705981384975866,
      "loss": 1.5339,
      "step": 467
    },
    {
      "epoch": 0.01718688673593753,
      "grad_norm": 0.3965727984905243,
      "learning_rate": 0.0001568865349101923,
      "loss": 1.4599,
      "step": 468
    },
    {
      "epoch": 0.017223610852894665,
      "grad_norm": 0.42512449622154236,
      "learning_rate": 0.00015671300307778898,
      "loss": 1.2251,
      "step": 469
    },
    {
      "epoch": 0.017260334969851795,
      "grad_norm": 0.40490153431892395,
      "learning_rate": 0.00015653921912399589,
      "loss": 1.3921,
      "step": 470
    },
    {
      "epoch": 0.017297059086808928,
      "grad_norm": 0.4677824079990387,
      "learning_rate": 0.00015636518382138107,
      "loss": 1.518,
      "step": 471
    },
    {
      "epoch": 0.017333783203766058,
      "grad_norm": 0.48965519666671753,
      "learning_rate": 0.0001561908979436299,
      "loss": 1.3384,
      "step": 472
    },
    {
      "epoch": 0.01737050732072319,
      "grad_norm": 0.4480048716068268,
      "learning_rate": 0.00015601636226554168,
      "loss": 1.0987,
      "step": 473
    },
    {
      "epoch": 0.01740723143768032,
      "grad_norm": 0.4561201333999634,
      "learning_rate": 0.00015584157756302634,
      "loss": 1.4265,
      "step": 474
    },
    {
      "epoch": 0.01744395555463745,
      "grad_norm": 0.5285091400146484,
      "learning_rate": 0.0001556665446131007,
      "loss": 1.3597,
      "step": 475
    },
    {
      "epoch": 0.017480679671594584,
      "grad_norm": 0.5413166880607605,
      "learning_rate": 0.00015549126419388536,
      "loss": 1.478,
      "step": 476
    },
    {
      "epoch": 0.017517403788551714,
      "grad_norm": 0.5521997809410095,
      "learning_rate": 0.0001553157370846009,
      "loss": 0.9438,
      "step": 477
    },
    {
      "epoch": 0.017554127905508848,
      "grad_norm": 0.7720620632171631,
      "learning_rate": 0.00015513996406556465,
      "loss": 1.4747,
      "step": 478
    },
    {
      "epoch": 0.017590852022465978,
      "grad_norm": 0.652715265750885,
      "learning_rate": 0.00015496394591818716,
      "loss": 1.1152,
      "step": 479
    },
    {
      "epoch": 0.01762757613942311,
      "grad_norm": 0.6476910710334778,
      "learning_rate": 0.0001547876834249687,
      "loss": 1.5823,
      "step": 480
    },
    {
      "epoch": 0.01766430025638024,
      "grad_norm": 0.6067008972167969,
      "learning_rate": 0.00015461117736949577,
      "loss": 1.2522,
      "step": 481
    },
    {
      "epoch": 0.017701024373337374,
      "grad_norm": 0.7004554867744446,
      "learning_rate": 0.00015443442853643762,
      "loss": 1.0938,
      "step": 482
    },
    {
      "epoch": 0.017737748490294504,
      "grad_norm": 0.7463083267211914,
      "learning_rate": 0.00015425743771154294,
      "loss": 1.3231,
      "step": 483
    },
    {
      "epoch": 0.017774472607251637,
      "grad_norm": 0.7362497448921204,
      "learning_rate": 0.00015408020568163602,
      "loss": 1.176,
      "step": 484
    },
    {
      "epoch": 0.017811196724208767,
      "grad_norm": 0.7758181095123291,
      "learning_rate": 0.00015390273323461352,
      "loss": 1.115,
      "step": 485
    },
    {
      "epoch": 0.0178479208411659,
      "grad_norm": 0.8141542077064514,
      "learning_rate": 0.0001537250211594409,
      "loss": 1.2076,
      "step": 486
    },
    {
      "epoch": 0.01788464495812303,
      "grad_norm": 0.7685984373092651,
      "learning_rate": 0.0001535470702461489,
      "loss": 1.0848,
      "step": 487
    },
    {
      "epoch": 0.01792136907508016,
      "grad_norm": 0.9328698515892029,
      "learning_rate": 0.00015336888128583,
      "loss": 1.05,
      "step": 488
    },
    {
      "epoch": 0.017958093192037294,
      "grad_norm": 0.8681111335754395,
      "learning_rate": 0.000153190455070635,
      "loss": 1.3879,
      "step": 489
    },
    {
      "epoch": 0.017994817308994424,
      "grad_norm": 0.9153539538383484,
      "learning_rate": 0.00015301179239376938,
      "loss": 1.2555,
      "step": 490
    },
    {
      "epoch": 0.018031541425951557,
      "grad_norm": 0.9349640011787415,
      "learning_rate": 0.00015283289404948976,
      "loss": 1.0775,
      "step": 491
    },
    {
      "epoch": 0.018068265542908687,
      "grad_norm": 0.8827584385871887,
      "learning_rate": 0.0001526537608331006,
      "loss": 1.1739,
      "step": 492
    },
    {
      "epoch": 0.01810498965986582,
      "grad_norm": 0.9011739492416382,
      "learning_rate": 0.00015247439354095041,
      "loss": 1.0797,
      "step": 493
    },
    {
      "epoch": 0.01814171377682295,
      "grad_norm": 1.1915018558502197,
      "learning_rate": 0.00015229479297042823,
      "loss": 1.1857,
      "step": 494
    },
    {
      "epoch": 0.018178437893780083,
      "grad_norm": 1.03218412399292,
      "learning_rate": 0.00015211495991996027,
      "loss": 0.9331,
      "step": 495
    },
    {
      "epoch": 0.018215162010737213,
      "grad_norm": 0.9765962362289429,
      "learning_rate": 0.0001519348951890062,
      "loss": 0.9965,
      "step": 496
    },
    {
      "epoch": 0.018251886127694347,
      "grad_norm": 1.1422512531280518,
      "learning_rate": 0.0001517545995780556,
      "loss": 1.1346,
      "step": 497
    },
    {
      "epoch": 0.018288610244651476,
      "grad_norm": 1.3611609935760498,
      "learning_rate": 0.00015157407388862452,
      "loss": 1.2754,
      "step": 498
    },
    {
      "epoch": 0.01832533436160861,
      "grad_norm": 1.4425441026687622,
      "learning_rate": 0.00015139331892325179,
      "loss": 1.3542,
      "step": 499
    },
    {
      "epoch": 0.01836205847856574,
      "grad_norm": 1.4951629638671875,
      "learning_rate": 0.0001512123354854955,
      "loss": 1.2259,
      "step": 500
    },
    {
      "epoch": 0.01839878259552287,
      "grad_norm": 0.18715369701385498,
      "learning_rate": 0.0001510311243799295,
      "loss": 1.4844,
      "step": 501
    },
    {
      "epoch": 0.018435506712480003,
      "grad_norm": 0.19282080233097076,
      "learning_rate": 0.00015084968641213958,
      "loss": 1.6516,
      "step": 502
    },
    {
      "epoch": 0.018472230829437133,
      "grad_norm": 0.20482544600963593,
      "learning_rate": 0.00015066802238872023,
      "loss": 1.5296,
      "step": 503
    },
    {
      "epoch": 0.018508954946394266,
      "grad_norm": 0.22331880033016205,
      "learning_rate": 0.0001504861331172709,
      "loss": 1.6436,
      "step": 504
    },
    {
      "epoch": 0.018545679063351396,
      "grad_norm": 0.23592646420001984,
      "learning_rate": 0.0001503040194063922,
      "loss": 1.4528,
      "step": 505
    },
    {
      "epoch": 0.01858240318030853,
      "grad_norm": 0.24901770055294037,
      "learning_rate": 0.00015012168206568268,
      "loss": 1.6135,
      "step": 506
    },
    {
      "epoch": 0.01861912729726566,
      "grad_norm": 0.29037392139434814,
      "learning_rate": 0.00014993912190573505,
      "loss": 1.4802,
      "step": 507
    },
    {
      "epoch": 0.018655851414222793,
      "grad_norm": 0.3209822177886963,
      "learning_rate": 0.00014975633973813242,
      "loss": 1.8436,
      "step": 508
    },
    {
      "epoch": 0.018692575531179922,
      "grad_norm": 0.33240190148353577,
      "learning_rate": 0.00014957333637544503,
      "loss": 1.4927,
      "step": 509
    },
    {
      "epoch": 0.018729299648137056,
      "grad_norm": 0.34968650341033936,
      "learning_rate": 0.00014939011263122634,
      "loss": 1.2905,
      "step": 510
    },
    {
      "epoch": 0.018766023765094186,
      "grad_norm": 0.36854010820388794,
      "learning_rate": 0.0001492066693200096,
      "loss": 1.4524,
      "step": 511
    },
    {
      "epoch": 0.01880274788205132,
      "grad_norm": 0.35444486141204834,
      "learning_rate": 0.00014902300725730413,
      "loss": 1.4175,
      "step": 512
    },
    {
      "epoch": 0.01883947199900845,
      "grad_norm": 0.36686965823173523,
      "learning_rate": 0.00014883912725959167,
      "loss": 1.2651,
      "step": 513
    },
    {
      "epoch": 0.01887619611596558,
      "grad_norm": 0.3605664074420929,
      "learning_rate": 0.00014865503014432292,
      "loss": 1.3291,
      "step": 514
    },
    {
      "epoch": 0.018912920232922712,
      "grad_norm": 0.3801887333393097,
      "learning_rate": 0.00014847071672991367,
      "loss": 1.4952,
      "step": 515
    },
    {
      "epoch": 0.018949644349879842,
      "grad_norm": 0.3794371783733368,
      "learning_rate": 0.0001482861878357414,
      "loss": 1.477,
      "step": 516
    },
    {
      "epoch": 0.018986368466836975,
      "grad_norm": 0.43200942873954773,
      "learning_rate": 0.00014810144428214144,
      "loss": 1.3653,
      "step": 517
    },
    {
      "epoch": 0.019023092583794105,
      "grad_norm": 0.37317952513694763,
      "learning_rate": 0.0001479164868904034,
      "loss": 1.4011,
      "step": 518
    },
    {
      "epoch": 0.01905981670075124,
      "grad_norm": 0.4185098111629486,
      "learning_rate": 0.00014773131648276758,
      "loss": 1.4726,
      "step": 519
    },
    {
      "epoch": 0.01909654081770837,
      "grad_norm": 0.4448191523551941,
      "learning_rate": 0.00014754593388242117,
      "loss": 1.4114,
      "step": 520
    },
    {
      "epoch": 0.019133264934665502,
      "grad_norm": 0.41695013642311096,
      "learning_rate": 0.0001473603399134948,
      "loss": 1.4341,
      "step": 521
    },
    {
      "epoch": 0.01916998905162263,
      "grad_norm": 0.47701892256736755,
      "learning_rate": 0.0001471745354010586,
      "loss": 1.3506,
      "step": 522
    },
    {
      "epoch": 0.019206713168579765,
      "grad_norm": 0.521359920501709,
      "learning_rate": 0.00014698852117111884,
      "loss": 1.2664,
      "step": 523
    },
    {
      "epoch": 0.019243437285536895,
      "grad_norm": 0.45853301882743835,
      "learning_rate": 0.000146802298050614,
      "loss": 1.2361,
      "step": 524
    },
    {
      "epoch": 0.019280161402494028,
      "grad_norm": 0.5051366686820984,
      "learning_rate": 0.0001466158668674112,
      "loss": 1.2276,
      "step": 525
    },
    {
      "epoch": 0.019316885519451158,
      "grad_norm": 0.6021683812141418,
      "learning_rate": 0.00014642922845030257,
      "loss": 1.4695,
      "step": 526
    },
    {
      "epoch": 0.019353609636408288,
      "grad_norm": 0.6333509087562561,
      "learning_rate": 0.0001462423836290015,
      "loss": 1.2002,
      "step": 527
    },
    {
      "epoch": 0.01939033375336542,
      "grad_norm": 0.5938871502876282,
      "learning_rate": 0.00014605533323413887,
      "loss": 1.436,
      "step": 528
    },
    {
      "epoch": 0.01942705787032255,
      "grad_norm": 0.596309244632721,
      "learning_rate": 0.00014586807809725962,
      "loss": 1.0836,
      "step": 529
    },
    {
      "epoch": 0.019463781987279685,
      "grad_norm": 0.6632168889045715,
      "learning_rate": 0.00014568061905081875,
      "loss": 1.272,
      "step": 530
    },
    {
      "epoch": 0.019500506104236814,
      "grad_norm": 0.7848820090293884,
      "learning_rate": 0.00014549295692817778,
      "loss": 1.4626,
      "step": 531
    },
    {
      "epoch": 0.019537230221193948,
      "grad_norm": 0.7731841802597046,
      "learning_rate": 0.00014530509256360102,
      "loss": 1.388,
      "step": 532
    },
    {
      "epoch": 0.019573954338151078,
      "grad_norm": 0.8241812586784363,
      "learning_rate": 0.00014511702679225193,
      "loss": 1.1485,
      "step": 533
    },
    {
      "epoch": 0.01961067845510821,
      "grad_norm": 0.7125410437583923,
      "learning_rate": 0.0001449287604501893,
      "loss": 0.9962,
      "step": 534
    },
    {
      "epoch": 0.01964740257206534,
      "grad_norm": 0.7405867576599121,
      "learning_rate": 0.00014474029437436348,
      "loss": 1.1625,
      "step": 535
    },
    {
      "epoch": 0.019684126689022474,
      "grad_norm": 0.7960091829299927,
      "learning_rate": 0.00014455162940261285,
      "loss": 1.4028,
      "step": 536
    },
    {
      "epoch": 0.019720850805979604,
      "grad_norm": 1.0968146324157715,
      "learning_rate": 0.0001443627663736599,
      "loss": 1.5011,
      "step": 537
    },
    {
      "epoch": 0.019757574922936737,
      "grad_norm": 0.8453080058097839,
      "learning_rate": 0.00014417370612710778,
      "loss": 1.0241,
      "step": 538
    },
    {
      "epoch": 0.019794299039893867,
      "grad_norm": 0.8853964805603027,
      "learning_rate": 0.00014398444950343623,
      "loss": 1.1412,
      "step": 539
    },
    {
      "epoch": 0.019831023156850997,
      "grad_norm": 0.9515712261199951,
      "learning_rate": 0.00014379499734399798,
      "loss": 1.0634,
      "step": 540
    },
    {
      "epoch": 0.01986774727380813,
      "grad_norm": 0.9078505635261536,
      "learning_rate": 0.0001436053504910151,
      "loss": 1.2107,
      "step": 541
    },
    {
      "epoch": 0.01990447139076526,
      "grad_norm": 0.896154522895813,
      "learning_rate": 0.0001434155097875752,
      "loss": 0.9549,
      "step": 542
    },
    {
      "epoch": 0.019941195507722394,
      "grad_norm": 0.9410357475280762,
      "learning_rate": 0.00014322547607762762,
      "loss": 1.3356,
      "step": 543
    },
    {
      "epoch": 0.019977919624679524,
      "grad_norm": 0.8842859268188477,
      "learning_rate": 0.0001430352502059797,
      "loss": 0.9895,
      "step": 544
    },
    {
      "epoch": 0.020014643741636657,
      "grad_norm": 1.0675865411758423,
      "learning_rate": 0.0001428448330182931,
      "loss": 1.2787,
      "step": 545
    },
    {
      "epoch": 0.020051367858593787,
      "grad_norm": 1.3076605796813965,
      "learning_rate": 0.00014265422536107993,
      "loss": 1.1929,
      "step": 546
    },
    {
      "epoch": 0.02008809197555092,
      "grad_norm": 1.0128639936447144,
      "learning_rate": 0.00014246342808169914,
      "loss": 0.9906,
      "step": 547
    },
    {
      "epoch": 0.02012481609250805,
      "grad_norm": 1.0764323472976685,
      "learning_rate": 0.00014227244202835257,
      "loss": 1.0493,
      "step": 548
    },
    {
      "epoch": 0.020161540209465183,
      "grad_norm": 1.1746193170547485,
      "learning_rate": 0.0001420812680500813,
      "loss": 1.2294,
      "step": 549
    },
    {
      "epoch": 0.020198264326422313,
      "grad_norm": 1.706727147102356,
      "learning_rate": 0.00014188990699676184,
      "loss": 1.1193,
      "step": 550
    },
    {
      "epoch": 0.020234988443379447,
      "grad_norm": 0.16662216186523438,
      "learning_rate": 0.00014169835971910238,
      "loss": 1.4148,
      "step": 551
    },
    {
      "epoch": 0.020271712560336577,
      "grad_norm": 0.2052755057811737,
      "learning_rate": 0.0001415066270686389,
      "loss": 1.8365,
      "step": 552
    },
    {
      "epoch": 0.020308436677293706,
      "grad_norm": 0.21343569457530975,
      "learning_rate": 0.00014131470989773158,
      "loss": 1.5062,
      "step": 553
    },
    {
      "epoch": 0.02034516079425084,
      "grad_norm": 0.2197510302066803,
      "learning_rate": 0.0001411226090595608,
      "loss": 1.6103,
      "step": 554
    },
    {
      "epoch": 0.02038188491120797,
      "grad_norm": 0.2580277919769287,
      "learning_rate": 0.00014093032540812348,
      "loss": 1.6062,
      "step": 555
    },
    {
      "epoch": 0.020418609028165103,
      "grad_norm": 0.2760462164878845,
      "learning_rate": 0.0001407378597982293,
      "loss": 1.3687,
      "step": 556
    },
    {
      "epoch": 0.020455333145122233,
      "grad_norm": 0.2603081166744232,
      "learning_rate": 0.00014054521308549673,
      "loss": 1.3717,
      "step": 557
    },
    {
      "epoch": 0.020492057262079366,
      "grad_norm": 0.30551066994667053,
      "learning_rate": 0.0001403523861263495,
      "loss": 1.6391,
      "step": 558
    },
    {
      "epoch": 0.020528781379036496,
      "grad_norm": 0.27283573150634766,
      "learning_rate": 0.00014015937977801256,
      "loss": 1.5672,
      "step": 559
    },
    {
      "epoch": 0.02056550549599363,
      "grad_norm": 0.36117789149284363,
      "learning_rate": 0.00013996619489850822,
      "loss": 1.4592,
      "step": 560
    },
    {
      "epoch": 0.02060222961295076,
      "grad_norm": 0.3764129877090454,
      "learning_rate": 0.00013977283234665273,
      "loss": 1.4239,
      "step": 561
    },
    {
      "epoch": 0.020638953729907893,
      "grad_norm": 0.36206743121147156,
      "learning_rate": 0.00013957929298205195,
      "loss": 1.3832,
      "step": 562
    },
    {
      "epoch": 0.020675677846865022,
      "grad_norm": 0.32104671001434326,
      "learning_rate": 0.00013938557766509792,
      "loss": 1.3012,
      "step": 563
    },
    {
      "epoch": 0.020712401963822156,
      "grad_norm": 0.36413687467575073,
      "learning_rate": 0.0001391916872569648,
      "loss": 1.4532,
      "step": 564
    },
    {
      "epoch": 0.020749126080779286,
      "grad_norm": 0.38346680998802185,
      "learning_rate": 0.00013899762261960518,
      "loss": 1.5625,
      "step": 565
    },
    {
      "epoch": 0.020785850197736416,
      "grad_norm": 0.41821011900901794,
      "learning_rate": 0.0001388033846157462,
      "loss": 1.5106,
      "step": 566
    },
    {
      "epoch": 0.02082257431469355,
      "grad_norm": 0.37590596079826355,
      "learning_rate": 0.0001386089741088857,
      "loss": 1.4084,
      "step": 567
    },
    {
      "epoch": 0.02085929843165068,
      "grad_norm": 0.3924044966697693,
      "learning_rate": 0.00013841439196328836,
      "loss": 1.5348,
      "step": 568
    },
    {
      "epoch": 0.020896022548607812,
      "grad_norm": 0.3740192651748657,
      "learning_rate": 0.00013821963904398193,
      "loss": 1.334,
      "step": 569
    },
    {
      "epoch": 0.020932746665564942,
      "grad_norm": 0.4276060461997986,
      "learning_rate": 0.00013802471621675338,
      "loss": 1.3288,
      "step": 570
    },
    {
      "epoch": 0.020969470782522075,
      "grad_norm": 0.44933679699897766,
      "learning_rate": 0.00013782962434814492,
      "loss": 1.4203,
      "step": 571
    },
    {
      "epoch": 0.021006194899479205,
      "grad_norm": 0.49060145020484924,
      "learning_rate": 0.00013763436430545034,
      "loss": 1.5375,
      "step": 572
    },
    {
      "epoch": 0.02104291901643634,
      "grad_norm": 0.5104357600212097,
      "learning_rate": 0.00013743893695671096,
      "loss": 1.2615,
      "step": 573
    },
    {
      "epoch": 0.02107964313339347,
      "grad_norm": 0.5011386275291443,
      "learning_rate": 0.00013724334317071198,
      "loss": 1.0699,
      "step": 574
    },
    {
      "epoch": 0.021116367250350602,
      "grad_norm": 0.47830095887184143,
      "learning_rate": 0.00013704758381697844,
      "loss": 1.3299,
      "step": 575
    },
    {
      "epoch": 0.02115309136730773,
      "grad_norm": 0.5917227864265442,
      "learning_rate": 0.00013685165976577146,
      "loss": 1.2574,
      "step": 576
    },
    {
      "epoch": 0.021189815484264865,
      "grad_norm": 0.5975363850593567,
      "learning_rate": 0.0001366555718880843,
      "loss": 1.3073,
      "step": 577
    },
    {
      "epoch": 0.021226539601221995,
      "grad_norm": 0.5982011556625366,
      "learning_rate": 0.00013645932105563844,
      "loss": 1.1572,
      "step": 578
    },
    {
      "epoch": 0.021263263718179125,
      "grad_norm": 0.5494576692581177,
      "learning_rate": 0.00013626290814088005,
      "loss": 1.1604,
      "step": 579
    },
    {
      "epoch": 0.021299987835136258,
      "grad_norm": 0.5590533018112183,
      "learning_rate": 0.00013606633401697557,
      "loss": 1.2671,
      "step": 580
    },
    {
      "epoch": 0.021336711952093388,
      "grad_norm": 0.6850517392158508,
      "learning_rate": 0.00013586959955780824,
      "loss": 1.1631,
      "step": 581
    },
    {
      "epoch": 0.02137343606905052,
      "grad_norm": 0.6311553120613098,
      "learning_rate": 0.00013567270563797398,
      "loss": 1.2525,
      "step": 582
    },
    {
      "epoch": 0.02141016018600765,
      "grad_norm": 0.6627228260040283,
      "learning_rate": 0.00013547565313277776,
      "loss": 1.4404,
      "step": 583
    },
    {
      "epoch": 0.021446884302964785,
      "grad_norm": 0.6669113039970398,
      "learning_rate": 0.00013527844291822948,
      "loss": 1.0547,
      "step": 584
    },
    {
      "epoch": 0.021483608419921914,
      "grad_norm": 0.7531651854515076,
      "learning_rate": 0.0001350810758710401,
      "loss": 1.4637,
      "step": 585
    },
    {
      "epoch": 0.021520332536879048,
      "grad_norm": 0.7591425180435181,
      "learning_rate": 0.00013488355286861783,
      "loss": 1.0415,
      "step": 586
    },
    {
      "epoch": 0.021557056653836178,
      "grad_norm": 0.7605099081993103,
      "learning_rate": 0.0001346858747890642,
      "loss": 1.1771,
      "step": 587
    },
    {
      "epoch": 0.02159378077079331,
      "grad_norm": 0.7345492243766785,
      "learning_rate": 0.00013448804251117003,
      "loss": 1.1395,
      "step": 588
    },
    {
      "epoch": 0.02163050488775044,
      "grad_norm": 0.788847029209137,
      "learning_rate": 0.0001342900569144119,
      "loss": 1.3063,
      "step": 589
    },
    {
      "epoch": 0.021667229004707574,
      "grad_norm": 0.7377067804336548,
      "learning_rate": 0.0001340919188789477,
      "loss": 1.0561,
      "step": 590
    },
    {
      "epoch": 0.021703953121664704,
      "grad_norm": 0.7130348682403564,
      "learning_rate": 0.00013389362928561317,
      "loss": 0.7479,
      "step": 591
    },
    {
      "epoch": 0.021740677238621834,
      "grad_norm": 0.8610103726387024,
      "learning_rate": 0.00013369518901591772,
      "loss": 0.8646,
      "step": 592
    },
    {
      "epoch": 0.021777401355578967,
      "grad_norm": 0.9117506742477417,
      "learning_rate": 0.00013349659895204067,
      "loss": 1.0925,
      "step": 593
    },
    {
      "epoch": 0.021814125472536097,
      "grad_norm": 0.9404719471931458,
      "learning_rate": 0.0001332978599768272,
      "loss": 1.2433,
      "step": 594
    },
    {
      "epoch": 0.02185084958949323,
      "grad_norm": 0.888764500617981,
      "learning_rate": 0.00013309897297378455,
      "loss": 0.9794,
      "step": 595
    },
    {
      "epoch": 0.02188757370645036,
      "grad_norm": 1.0601435899734497,
      "learning_rate": 0.00013289993882707797,
      "loss": 0.9397,
      "step": 596
    },
    {
      "epoch": 0.021924297823407494,
      "grad_norm": 1.1301285028457642,
      "learning_rate": 0.00013270075842152678,
      "loss": 1.1752,
      "step": 597
    },
    {
      "epoch": 0.021961021940364624,
      "grad_norm": 1.1921364068984985,
      "learning_rate": 0.00013250143264260074,
      "loss": 0.7725,
      "step": 598
    },
    {
      "epoch": 0.021997746057321757,
      "grad_norm": 1.2796521186828613,
      "learning_rate": 0.0001323019623764156,
      "loss": 1.2066,
      "step": 599
    },
    {
      "epoch": 0.022034470174278887,
      "grad_norm": 1.4445984363555908,
      "learning_rate": 0.00013210234850972964,
      "loss": 1.0011,
      "step": 600
    },
    {
      "epoch": 0.02207119429123602,
      "grad_norm": 0.16283242404460907,
      "learning_rate": 0.0001319025919299394,
      "loss": 1.5617,
      "step": 601
    },
    {
      "epoch": 0.02210791840819315,
      "grad_norm": 0.21403326094150543,
      "learning_rate": 0.00013170269352507597,
      "loss": 1.5127,
      "step": 602
    },
    {
      "epoch": 0.022144642525150283,
      "grad_norm": 0.2155560702085495,
      "learning_rate": 0.0001315026541838008,
      "loss": 1.5917,
      "step": 603
    },
    {
      "epoch": 0.022181366642107413,
      "grad_norm": 0.2413771003484726,
      "learning_rate": 0.00013130247479540202,
      "loss": 1.5937,
      "step": 604
    },
    {
      "epoch": 0.022218090759064543,
      "grad_norm": 0.2648985981941223,
      "learning_rate": 0.00013110215624979025,
      "loss": 1.642,
      "step": 605
    },
    {
      "epoch": 0.022254814876021677,
      "grad_norm": 0.26700401306152344,
      "learning_rate": 0.00013090169943749476,
      "loss": 1.6316,
      "step": 606
    },
    {
      "epoch": 0.022291538992978806,
      "grad_norm": 0.2902841866016388,
      "learning_rate": 0.00013070110524965954,
      "loss": 1.3539,
      "step": 607
    },
    {
      "epoch": 0.02232826310993594,
      "grad_norm": 0.30102282762527466,
      "learning_rate": 0.00013050037457803924,
      "loss": 1.5971,
      "step": 608
    },
    {
      "epoch": 0.02236498722689307,
      "grad_norm": 0.3540296256542206,
      "learning_rate": 0.0001302995083149953,
      "loss": 1.6125,
      "step": 609
    },
    {
      "epoch": 0.022401711343850203,
      "grad_norm": 0.3202471435070038,
      "learning_rate": 0.0001300985073534919,
      "loss": 1.5303,
      "step": 610
    },
    {
      "epoch": 0.022438435460807333,
      "grad_norm": 0.35276538133621216,
      "learning_rate": 0.00012989737258709203,
      "loss": 1.5476,
      "step": 611
    },
    {
      "epoch": 0.022475159577764466,
      "grad_norm": 0.37919771671295166,
      "learning_rate": 0.00012969610490995358,
      "loss": 1.5535,
      "step": 612
    },
    {
      "epoch": 0.022511883694721596,
      "grad_norm": 0.36852240562438965,
      "learning_rate": 0.00012949470521682528,
      "loss": 1.1613,
      "step": 613
    },
    {
      "epoch": 0.02254860781167873,
      "grad_norm": 0.387608140707016,
      "learning_rate": 0.0001292931744030427,
      "loss": 1.4537,
      "step": 614
    },
    {
      "epoch": 0.02258533192863586,
      "grad_norm": 0.3829215168952942,
      "learning_rate": 0.0001290915133645243,
      "loss": 1.3919,
      "step": 615
    },
    {
      "epoch": 0.022622056045592993,
      "grad_norm": 0.3845890760421753,
      "learning_rate": 0.00012888972299776754,
      "loss": 1.5136,
      "step": 616
    },
    {
      "epoch": 0.022658780162550123,
      "grad_norm": 0.39009857177734375,
      "learning_rate": 0.00012868780419984482,
      "loss": 1.4686,
      "step": 617
    },
    {
      "epoch": 0.022695504279507252,
      "grad_norm": 0.3888898193836212,
      "learning_rate": 0.00012848575786839943,
      "loss": 1.442,
      "step": 618
    },
    {
      "epoch": 0.022732228396464386,
      "grad_norm": 0.4165363609790802,
      "learning_rate": 0.0001282835849016416,
      "loss": 1.5438,
      "step": 619
    },
    {
      "epoch": 0.022768952513421516,
      "grad_norm": 0.4076935052871704,
      "learning_rate": 0.00012808128619834461,
      "loss": 1.4592,
      "step": 620
    },
    {
      "epoch": 0.02280567663037865,
      "grad_norm": 0.48429492115974426,
      "learning_rate": 0.0001278788626578407,
      "loss": 1.4999,
      "step": 621
    },
    {
      "epoch": 0.02284240074733578,
      "grad_norm": 0.460707426071167,
      "learning_rate": 0.00012767631518001698,
      "loss": 1.5939,
      "step": 622
    },
    {
      "epoch": 0.022879124864292912,
      "grad_norm": 0.4455718994140625,
      "learning_rate": 0.00012747364466531163,
      "loss": 1.5002,
      "step": 623
    },
    {
      "epoch": 0.022915848981250042,
      "grad_norm": 0.48188769817352295,
      "learning_rate": 0.00012727085201470973,
      "loss": 1.2795,
      "step": 624
    },
    {
      "epoch": 0.022952573098207175,
      "grad_norm": 0.4929306209087372,
      "learning_rate": 0.00012706793812973941,
      "loss": 1.2996,
      "step": 625
    },
    {
      "epoch": 0.022989297215164305,
      "grad_norm": 0.503911554813385,
      "learning_rate": 0.0001268649039124677,
      "loss": 1.5771,
      "step": 626
    },
    {
      "epoch": 0.02302602133212144,
      "grad_norm": 0.5796214938163757,
      "learning_rate": 0.00012666175026549662,
      "loss": 1.3144,
      "step": 627
    },
    {
      "epoch": 0.02306274544907857,
      "grad_norm": 0.6723256707191467,
      "learning_rate": 0.000126458478091959,
      "loss": 1.2091,
      "step": 628
    },
    {
      "epoch": 0.023099469566035702,
      "grad_norm": 0.6558849811553955,
      "learning_rate": 0.00012625508829551473,
      "loss": 1.4037,
      "step": 629
    },
    {
      "epoch": 0.023136193682992832,
      "grad_norm": 0.5532969832420349,
      "learning_rate": 0.00012605158178034654,
      "loss": 1.1827,
      "step": 630
    },
    {
      "epoch": 0.023172917799949965,
      "grad_norm": 0.6064698100090027,
      "learning_rate": 0.00012584795945115603,
      "loss": 1.4832,
      "step": 631
    },
    {
      "epoch": 0.023209641916907095,
      "grad_norm": 0.5781803727149963,
      "learning_rate": 0.0001256442222131597,
      "loss": 1.2998,
      "step": 632
    },
    {
      "epoch": 0.023246366033864225,
      "grad_norm": 0.7046496868133545,
      "learning_rate": 0.0001254403709720848,
      "loss": 1.2692,
      "step": 633
    },
    {
      "epoch": 0.023283090150821358,
      "grad_norm": 0.8553188443183899,
      "learning_rate": 0.0001252364066341655,
      "loss": 1.3435,
      "step": 634
    },
    {
      "epoch": 0.023319814267778488,
      "grad_norm": 0.6920652985572815,
      "learning_rate": 0.00012503233010613865,
      "loss": 1.3254,
      "step": 635
    },
    {
      "epoch": 0.02335653838473562,
      "grad_norm": 0.7462049722671509,
      "learning_rate": 0.00012482814229523997,
      "loss": 1.1451,
      "step": 636
    },
    {
      "epoch": 0.02339326250169275,
      "grad_norm": 0.7453550696372986,
      "learning_rate": 0.00012462384410919975,
      "loss": 1.343,
      "step": 637
    },
    {
      "epoch": 0.023429986618649885,
      "grad_norm": 0.9004718065261841,
      "learning_rate": 0.00012441943645623903,
      "loss": 1.0797,
      "step": 638
    },
    {
      "epoch": 0.023466710735607015,
      "grad_norm": 0.9025092124938965,
      "learning_rate": 0.00012421492024506555,
      "loss": 1.2324,
      "step": 639
    },
    {
      "epoch": 0.023503434852564148,
      "grad_norm": 0.7931037545204163,
      "learning_rate": 0.00012401029638486953,
      "loss": 0.9252,
      "step": 640
    },
    {
      "epoch": 0.023540158969521278,
      "grad_norm": 0.8842725157737732,
      "learning_rate": 0.0001238055657853198,
      "loss": 1.0009,
      "step": 641
    },
    {
      "epoch": 0.02357688308647841,
      "grad_norm": 0.9652261734008789,
      "learning_rate": 0.00012360072935655982,
      "loss": 1.1185,
      "step": 642
    },
    {
      "epoch": 0.02361360720343554,
      "grad_norm": 0.9765187501907349,
      "learning_rate": 0.00012339578800920332,
      "loss": 1.0518,
      "step": 643
    },
    {
      "epoch": 0.023650331320392674,
      "grad_norm": 1.199384093284607,
      "learning_rate": 0.00012319074265433063,
      "loss": 0.8353,
      "step": 644
    },
    {
      "epoch": 0.023687055437349804,
      "grad_norm": 1.6428890228271484,
      "learning_rate": 0.00012298559420348437,
      "loss": 1.0458,
      "step": 645
    },
    {
      "epoch": 0.023723779554306934,
      "grad_norm": 1.3298590183258057,
      "learning_rate": 0.00012278034356866545,
      "loss": 1.0517,
      "step": 646
    },
    {
      "epoch": 0.023760503671264067,
      "grad_norm": 1.1145323514938354,
      "learning_rate": 0.00012257499166232907,
      "loss": 1.2029,
      "step": 647
    },
    {
      "epoch": 0.023797227788221197,
      "grad_norm": 1.1015548706054688,
      "learning_rate": 0.0001223695393973807,
      "loss": 1.1489,
      "step": 648
    },
    {
      "epoch": 0.02383395190517833,
      "grad_norm": 1.4857112169265747,
      "learning_rate": 0.0001221639876871719,
      "loss": 1.3863,
      "step": 649
    },
    {
      "epoch": 0.02387067602213546,
      "grad_norm": 2.150747299194336,
      "learning_rate": 0.0001219583374454963,
      "loss": 1.1509,
      "step": 650
    },
    {
      "epoch": 0.023907400139092594,
      "grad_norm": 0.18142010271549225,
      "learning_rate": 0.00012175258958658564,
      "loss": 1.5288,
      "step": 651
    },
    {
      "epoch": 0.023944124256049724,
      "grad_norm": 0.20369091629981995,
      "learning_rate": 0.00012154674502510555,
      "loss": 1.603,
      "step": 652
    },
    {
      "epoch": 0.023980848373006857,
      "grad_norm": 0.21081200242042542,
      "learning_rate": 0.00012134080467615159,
      "loss": 1.5871,
      "step": 653
    },
    {
      "epoch": 0.024017572489963987,
      "grad_norm": 0.2664206922054291,
      "learning_rate": 0.00012113476945524513,
      "loss": 1.6111,
      "step": 654
    },
    {
      "epoch": 0.02405429660692112,
      "grad_norm": 0.27232760190963745,
      "learning_rate": 0.00012092864027832933,
      "loss": 1.7799,
      "step": 655
    },
    {
      "epoch": 0.02409102072387825,
      "grad_norm": 0.262323260307312,
      "learning_rate": 0.000120722418061765,
      "loss": 1.5473,
      "step": 656
    },
    {
      "epoch": 0.024127744840835384,
      "grad_norm": 0.2722291946411133,
      "learning_rate": 0.0001205161037223266,
      "loss": 1.552,
      "step": 657
    },
    {
      "epoch": 0.024164468957792513,
      "grad_norm": 0.31498774886131287,
      "learning_rate": 0.00012030969817719808,
      "loss": 1.6091,
      "step": 658
    },
    {
      "epoch": 0.024201193074749643,
      "grad_norm": 0.29550325870513916,
      "learning_rate": 0.00012010320234396894,
      "loss": 1.4843,
      "step": 659
    },
    {
      "epoch": 0.024237917191706777,
      "grad_norm": 0.3069508969783783,
      "learning_rate": 0.00011989661714062999,
      "loss": 1.4666,
      "step": 660
    },
    {
      "epoch": 0.024274641308663907,
      "grad_norm": 0.3138614594936371,
      "learning_rate": 0.0001196899434855693,
      "loss": 1.6811,
      "step": 661
    },
    {
      "epoch": 0.02431136542562104,
      "grad_norm": 0.3409472405910492,
      "learning_rate": 0.00011948318229756827,
      "loss": 1.2844,
      "step": 662
    },
    {
      "epoch": 0.02434808954257817,
      "grad_norm": 0.36731666326522827,
      "learning_rate": 0.00011927633449579735,
      "loss": 1.2723,
      "step": 663
    },
    {
      "epoch": 0.024384813659535303,
      "grad_norm": 0.3378710448741913,
      "learning_rate": 0.0001190694009998121,
      "loss": 1.3536,
      "step": 664
    },
    {
      "epoch": 0.024421537776492433,
      "grad_norm": 0.3905196189880371,
      "learning_rate": 0.00011886238272954897,
      "loss": 1.5035,
      "step": 665
    },
    {
      "epoch": 0.024458261893449566,
      "grad_norm": 0.3912372887134552,
      "learning_rate": 0.00011865528060532127,
      "loss": 1.5337,
      "step": 666
    },
    {
      "epoch": 0.024494986010406696,
      "grad_norm": 0.4177848696708679,
      "learning_rate": 0.0001184480955478152,
      "loss": 1.5134,
      "step": 667
    },
    {
      "epoch": 0.02453171012736383,
      "grad_norm": 0.39146649837493896,
      "learning_rate": 0.00011824082847808558,
      "loss": 1.4287,
      "step": 668
    },
    {
      "epoch": 0.02456843424432096,
      "grad_norm": 0.38969284296035767,
      "learning_rate": 0.00011803348031755179,
      "loss": 1.3402,
      "step": 669
    },
    {
      "epoch": 0.024605158361278093,
      "grad_norm": 0.4288013279438019,
      "learning_rate": 0.0001178260519879937,
      "loss": 1.4364,
      "step": 670
    },
    {
      "epoch": 0.024641882478235223,
      "grad_norm": 0.43082699179649353,
      "learning_rate": 0.00011761854441154767,
      "loss": 1.5866,
      "step": 671
    },
    {
      "epoch": 0.024678606595192353,
      "grad_norm": 0.41317757964134216,
      "learning_rate": 0.00011741095851070228,
      "loss": 1.2308,
      "step": 672
    },
    {
      "epoch": 0.024715330712149486,
      "grad_norm": 0.48538509011268616,
      "learning_rate": 0.00011720329520829429,
      "loss": 1.4215,
      "step": 673
    },
    {
      "epoch": 0.024752054829106616,
      "grad_norm": 0.47991839051246643,
      "learning_rate": 0.0001169955554275046,
      "loss": 1.2303,
      "step": 674
    },
    {
      "epoch": 0.02478877894606375,
      "grad_norm": 0.5367565155029297,
      "learning_rate": 0.0001167877400918541,
      "loss": 1.1713,
      "step": 675
    },
    {
      "epoch": 0.02482550306302088,
      "grad_norm": 0.632439911365509,
      "learning_rate": 0.00011657985012519952,
      "loss": 1.587,
      "step": 676
    },
    {
      "epoch": 0.024862227179978012,
      "grad_norm": 0.5963438153266907,
      "learning_rate": 0.00011637188645172944,
      "loss": 1.5088,
      "step": 677
    },
    {
      "epoch": 0.024898951296935142,
      "grad_norm": 0.6097551584243774,
      "learning_rate": 0.00011616384999596006,
      "loss": 1.2798,
      "step": 678
    },
    {
      "epoch": 0.024935675413892276,
      "grad_norm": 0.6214428544044495,
      "learning_rate": 0.00011595574168273111,
      "loss": 1.3446,
      "step": 679
    },
    {
      "epoch": 0.024972399530849405,
      "grad_norm": 0.7046561241149902,
      "learning_rate": 0.0001157475624372018,
      "loss": 1.2449,
      "step": 680
    },
    {
      "epoch": 0.02500912364780654,
      "grad_norm": 0.6154795289039612,
      "learning_rate": 0.0001155393131848467,
      "loss": 1.1613,
      "step": 681
    },
    {
      "epoch": 0.02504584776476367,
      "grad_norm": 0.5873011946678162,
      "learning_rate": 0.00011533099485145155,
      "loss": 1.2176,
      "step": 682
    },
    {
      "epoch": 0.025082571881720802,
      "grad_norm": 0.6725327968597412,
      "learning_rate": 0.00011512260836310924,
      "loss": 1.2244,
      "step": 683
    },
    {
      "epoch": 0.025119295998677932,
      "grad_norm": 0.6779379844665527,
      "learning_rate": 0.00011491415464621562,
      "loss": 1.3483,
      "step": 684
    },
    {
      "epoch": 0.02515602011563506,
      "grad_norm": 0.6670539379119873,
      "learning_rate": 0.00011470563462746541,
      "loss": 0.9212,
      "step": 685
    },
    {
      "epoch": 0.025192744232592195,
      "grad_norm": 0.7365601062774658,
      "learning_rate": 0.00011449704923384812,
      "loss": 0.9123,
      "step": 686
    },
    {
      "epoch": 0.025229468349549325,
      "grad_norm": 0.727044939994812,
      "learning_rate": 0.00011428839939264382,
      "loss": 1.0165,
      "step": 687
    },
    {
      "epoch": 0.02526619246650646,
      "grad_norm": 0.7276614308357239,
      "learning_rate": 0.0001140796860314191,
      "loss": 1.2586,
      "step": 688
    },
    {
      "epoch": 0.025302916583463588,
      "grad_norm": 0.9211059808731079,
      "learning_rate": 0.00011387091007802297,
      "loss": 1.2762,
      "step": 689
    },
    {
      "epoch": 0.02533964070042072,
      "grad_norm": 0.9361832737922668,
      "learning_rate": 0.0001136620724605827,
      "loss": 1.1681,
      "step": 690
    },
    {
      "epoch": 0.02537636481737785,
      "grad_norm": 0.9103788733482361,
      "learning_rate": 0.00011345317410749964,
      "loss": 1.6343,
      "step": 691
    },
    {
      "epoch": 0.025413088934334985,
      "grad_norm": 0.901527464389801,
      "learning_rate": 0.00011324421594744516,
      "loss": 1.2478,
      "step": 692
    },
    {
      "epoch": 0.025449813051292115,
      "grad_norm": 0.9870873689651489,
      "learning_rate": 0.00011303519890935656,
      "loss": 1.0414,
      "step": 693
    },
    {
      "epoch": 0.025486537168249248,
      "grad_norm": 0.9779725074768066,
      "learning_rate": 0.00011282612392243286,
      "loss": 1.1994,
      "step": 694
    },
    {
      "epoch": 0.025523261285206378,
      "grad_norm": 0.9034556150436401,
      "learning_rate": 0.00011261699191613066,
      "loss": 1.1086,
      "step": 695
    },
    {
      "epoch": 0.02555998540216351,
      "grad_norm": 1.065763235092163,
      "learning_rate": 0.00011240780382016005,
      "loss": 1.0693,
      "step": 696
    },
    {
      "epoch": 0.02559670951912064,
      "grad_norm": 0.9644365906715393,
      "learning_rate": 0.00011219856056448051,
      "loss": 0.8049,
      "step": 697
    },
    {
      "epoch": 0.02563343363607777,
      "grad_norm": 1.0780218839645386,
      "learning_rate": 0.00011198926307929664,
      "loss": 1.2169,
      "step": 698
    },
    {
      "epoch": 0.025670157753034904,
      "grad_norm": 1.1070480346679688,
      "learning_rate": 0.00011177991229505431,
      "loss": 1.1208,
      "step": 699
    },
    {
      "epoch": 0.025706881869992034,
      "grad_norm": 1.8106689453125,
      "learning_rate": 0.00011157050914243614,
      "loss": 1.5293,
      "step": 700
    },
    {
      "epoch": 0.025743605986949167,
      "grad_norm": 0.15589463710784912,
      "learning_rate": 0.00011136105455235766,
      "loss": 1.539,
      "step": 701
    },
    {
      "epoch": 0.025780330103906297,
      "grad_norm": 0.1818956434726715,
      "learning_rate": 0.00011115154945596305,
      "loss": 1.6633,
      "step": 702
    },
    {
      "epoch": 0.02581705422086343,
      "grad_norm": 0.22154124081134796,
      "learning_rate": 0.00011094199478462095,
      "loss": 1.7534,
      "step": 703
    },
    {
      "epoch": 0.02585377833782056,
      "grad_norm": 0.21808061003684998,
      "learning_rate": 0.00011073239146992054,
      "loss": 1.5555,
      "step": 704
    },
    {
      "epoch": 0.025890502454777694,
      "grad_norm": 0.2117631435394287,
      "learning_rate": 0.00011052274044366711,
      "loss": 1.5032,
      "step": 705
    },
    {
      "epoch": 0.025927226571734824,
      "grad_norm": 0.2357957512140274,
      "learning_rate": 0.00011031304263787812,
      "loss": 1.4831,
      "step": 706
    },
    {
      "epoch": 0.025963950688691957,
      "grad_norm": 0.24266815185546875,
      "learning_rate": 0.00011010329898477891,
      "loss": 1.3805,
      "step": 707
    },
    {
      "epoch": 0.026000674805649087,
      "grad_norm": 0.2782483398914337,
      "learning_rate": 0.0001098935104167988,
      "loss": 1.3342,
      "step": 708
    },
    {
      "epoch": 0.02603739892260622,
      "grad_norm": 0.2865573465824127,
      "learning_rate": 0.00010968367786656663,
      "loss": 1.5098,
      "step": 709
    },
    {
      "epoch": 0.02607412303956335,
      "grad_norm": 0.3134574890136719,
      "learning_rate": 0.00010947380226690684,
      "loss": 1.3941,
      "step": 710
    },
    {
      "epoch": 0.02611084715652048,
      "grad_norm": 0.3235221803188324,
      "learning_rate": 0.00010926388455083522,
      "loss": 1.5095,
      "step": 711
    },
    {
      "epoch": 0.026147571273477613,
      "grad_norm": 0.3691047430038452,
      "learning_rate": 0.00010905392565155477,
      "loss": 1.5345,
      "step": 712
    },
    {
      "epoch": 0.026184295390434743,
      "grad_norm": 0.3187674283981323,
      "learning_rate": 0.00010884392650245165,
      "loss": 1.5057,
      "step": 713
    },
    {
      "epoch": 0.026221019507391877,
      "grad_norm": 0.34957951307296753,
      "learning_rate": 0.00010863388803709089,
      "loss": 1.321,
      "step": 714
    },
    {
      "epoch": 0.026257743624349007,
      "grad_norm": 0.38225123286247253,
      "learning_rate": 0.00010842381118921232,
      "loss": 1.4884,
      "step": 715
    },
    {
      "epoch": 0.02629446774130614,
      "grad_norm": 0.379058301448822,
      "learning_rate": 0.00010821369689272638,
      "loss": 1.591,
      "step": 716
    },
    {
      "epoch": 0.02633119185826327,
      "grad_norm": 0.3939705491065979,
      "learning_rate": 0.00010800354608171003,
      "loss": 1.3692,
      "step": 717
    },
    {
      "epoch": 0.026367915975220403,
      "grad_norm": 0.3963218927383423,
      "learning_rate": 0.00010779335969040252,
      "loss": 1.5151,
      "step": 718
    },
    {
      "epoch": 0.026404640092177533,
      "grad_norm": 0.39338696002960205,
      "learning_rate": 0.00010758313865320134,
      "loss": 1.4341,
      "step": 719
    },
    {
      "epoch": 0.026441364209134666,
      "grad_norm": 0.41341230273246765,
      "learning_rate": 0.00010737288390465792,
      "loss": 1.3534,
      "step": 720
    },
    {
      "epoch": 0.026478088326091796,
      "grad_norm": 0.4581339657306671,
      "learning_rate": 0.00010716259637947357,
      "loss": 1.4227,
      "step": 721
    },
    {
      "epoch": 0.02651481244304893,
      "grad_norm": 0.4481271505355835,
      "learning_rate": 0.00010695227701249537,
      "loss": 1.4611,
      "step": 722
    },
    {
      "epoch": 0.02655153656000606,
      "grad_norm": 0.4743872284889221,
      "learning_rate": 0.00010674192673871191,
      "loss": 1.2814,
      "step": 723
    },
    {
      "epoch": 0.02658826067696319,
      "grad_norm": 0.48745524883270264,
      "learning_rate": 0.00010653154649324917,
      "loss": 1.2565,
      "step": 724
    },
    {
      "epoch": 0.026624984793920323,
      "grad_norm": 0.49790164828300476,
      "learning_rate": 0.00010632113721136636,
      "loss": 1.2832,
      "step": 725
    },
    {
      "epoch": 0.026661708910877453,
      "grad_norm": 0.5525292158126831,
      "learning_rate": 0.00010611069982845183,
      "loss": 1.4361,
      "step": 726
    },
    {
      "epoch": 0.026698433027834586,
      "grad_norm": 0.5680780410766602,
      "learning_rate": 0.00010590023528001884,
      "loss": 1.1937,
      "step": 727
    },
    {
      "epoch": 0.026735157144791716,
      "grad_norm": 0.5430833101272583,
      "learning_rate": 0.00010568974450170139,
      "loss": 1.1454,
      "step": 728
    },
    {
      "epoch": 0.02677188126174885,
      "grad_norm": 0.6157318353652954,
      "learning_rate": 0.00010547922842925008,
      "loss": 1.3081,
      "step": 729
    },
    {
      "epoch": 0.02680860537870598,
      "grad_norm": 0.689260721206665,
      "learning_rate": 0.00010526868799852796,
      "loss": 1.2269,
      "step": 730
    },
    {
      "epoch": 0.026845329495663112,
      "grad_norm": 0.6845170855522156,
      "learning_rate": 0.0001050581241455064,
      "loss": 1.4625,
      "step": 731
    },
    {
      "epoch": 0.026882053612620242,
      "grad_norm": 0.6399394869804382,
      "learning_rate": 0.00010484753780626089,
      "loss": 1.1899,
      "step": 732
    },
    {
      "epoch": 0.026918777729577376,
      "grad_norm": 0.7068498730659485,
      "learning_rate": 0.00010463692991696685,
      "loss": 1.1342,
      "step": 733
    },
    {
      "epoch": 0.026955501846534505,
      "grad_norm": 0.7933351993560791,
      "learning_rate": 0.00010442630141389549,
      "loss": 1.4175,
      "step": 734
    },
    {
      "epoch": 0.02699222596349164,
      "grad_norm": 0.6755648255348206,
      "learning_rate": 0.00010421565323340971,
      "loss": 1.2071,
      "step": 735
    },
    {
      "epoch": 0.02702895008044877,
      "grad_norm": 0.7793658375740051,
      "learning_rate": 0.00010400498631195992,
      "loss": 1.3134,
      "step": 736
    },
    {
      "epoch": 0.0270656741974059,
      "grad_norm": 0.7730168700218201,
      "learning_rate": 0.00010379430158607975,
      "loss": 1.3349,
      "step": 737
    },
    {
      "epoch": 0.027102398314363032,
      "grad_norm": 0.7361586689949036,
      "learning_rate": 0.000103583599992382,
      "loss": 1.0045,
      "step": 738
    },
    {
      "epoch": 0.027139122431320162,
      "grad_norm": 0.7865634560585022,
      "learning_rate": 0.0001033728824675545,
      "loss": 1.0729,
      "step": 739
    },
    {
      "epoch": 0.027175846548277295,
      "grad_norm": 0.7901663184165955,
      "learning_rate": 0.0001031621499483559,
      "loss": 0.94,
      "step": 740
    },
    {
      "epoch": 0.027212570665234425,
      "grad_norm": 1.0234339237213135,
      "learning_rate": 0.00010295140337161146,
      "loss": 1.249,
      "step": 741
    },
    {
      "epoch": 0.02724929478219156,
      "grad_norm": 0.839458703994751,
      "learning_rate": 0.00010274064367420897,
      "loss": 1.0009,
      "step": 742
    },
    {
      "epoch": 0.027286018899148688,
      "grad_norm": 0.9051604866981506,
      "learning_rate": 0.00010252987179309459,
      "loss": 1.0734,
      "step": 743
    },
    {
      "epoch": 0.02732274301610582,
      "grad_norm": 1.0820270776748657,
      "learning_rate": 0.00010231908866526851,
      "loss": 0.9465,
      "step": 744
    },
    {
      "epoch": 0.02735946713306295,
      "grad_norm": 0.944997251033783,
      "learning_rate": 0.00010210829522778111,
      "loss": 0.8389,
      "step": 745
    },
    {
      "epoch": 0.027396191250020085,
      "grad_norm": 0.9419519901275635,
      "learning_rate": 0.00010189749241772844,
      "loss": 0.8569,
      "step": 746
    },
    {
      "epoch": 0.027432915366977215,
      "grad_norm": 1.273436427116394,
      "learning_rate": 0.00010168668117224825,
      "loss": 1.1612,
      "step": 747
    },
    {
      "epoch": 0.027469639483934348,
      "grad_norm": 1.416648030281067,
      "learning_rate": 0.00010147586242851585,
      "loss": 1.2871,
      "step": 748
    },
    {
      "epoch": 0.027506363600891478,
      "grad_norm": 1.5401936769485474,
      "learning_rate": 0.00010126503712373982,
      "loss": 1.1061,
      "step": 749
    },
    {
      "epoch": 0.027543087717848608,
      "grad_norm": 1.9711471796035767,
      "learning_rate": 0.00010105420619515798,
      "loss": 1.2697,
      "step": 750
    },
    {
      "epoch": 0.027543087717848608,
      "eval_loss": 1.2755405902862549,
      "eval_runtime": 1107.2796,
      "eval_samples_per_second": 41.419,
      "eval_steps_per_second": 20.709,
      "step": 750
    },
    {
      "epoch": 0.02757981183480574,
      "grad_norm": 0.1593589037656784,
      "learning_rate": 0.00010084337058003303,
      "loss": 1.6991,
      "step": 751
    },
    {
      "epoch": 0.02761653595176287,
      "grad_norm": 0.19080309569835663,
      "learning_rate": 0.00010063253121564868,
      "loss": 1.6183,
      "step": 752
    },
    {
      "epoch": 0.027653260068720004,
      "grad_norm": 0.19537465274333954,
      "learning_rate": 0.00010042168903930514,
      "loss": 1.5824,
      "step": 753
    },
    {
      "epoch": 0.027689984185677134,
      "grad_norm": 0.20727789402008057,
      "learning_rate": 0.00010021084498831522,
      "loss": 1.5792,
      "step": 754
    },
    {
      "epoch": 0.027726708302634268,
      "grad_norm": 0.22856277227401733,
      "learning_rate": 0.0001,
      "loss": 1.7078,
      "step": 755
    },
    {
      "epoch": 0.027763432419591397,
      "grad_norm": 0.2358417958021164,
      "learning_rate": 9.97891550116848e-05,
      "loss": 1.5463,
      "step": 756
    },
    {
      "epoch": 0.02780015653654853,
      "grad_norm": 0.23272855579853058,
      "learning_rate": 9.957831096069488e-05,
      "loss": 1.4442,
      "step": 757
    },
    {
      "epoch": 0.02783688065350566,
      "grad_norm": 0.28245487809181213,
      "learning_rate": 9.936746878435136e-05,
      "loss": 1.3755,
      "step": 758
    },
    {
      "epoch": 0.027873604770462794,
      "grad_norm": 0.29861894249916077,
      "learning_rate": 9.915662941996699e-05,
      "loss": 1.3838,
      "step": 759
    },
    {
      "epoch": 0.027910328887419924,
      "grad_norm": 0.30934277176856995,
      "learning_rate": 9.894579380484204e-05,
      "loss": 1.4364,
      "step": 760
    },
    {
      "epoch": 0.027947053004377057,
      "grad_norm": 0.314209520816803,
      "learning_rate": 9.873496287626019e-05,
      "loss": 1.5728,
      "step": 761
    },
    {
      "epoch": 0.027983777121334187,
      "grad_norm": 0.3142238259315491,
      "learning_rate": 9.852413757148417e-05,
      "loss": 1.4214,
      "step": 762
    },
    {
      "epoch": 0.028020501238291317,
      "grad_norm": 0.3537854552268982,
      "learning_rate": 9.831331882775178e-05,
      "loss": 1.4793,
      "step": 763
    },
    {
      "epoch": 0.02805722535524845,
      "grad_norm": 0.36891040205955505,
      "learning_rate": 9.81025075822716e-05,
      "loss": 1.3237,
      "step": 764
    },
    {
      "epoch": 0.02809394947220558,
      "grad_norm": 0.44001129269599915,
      "learning_rate": 9.789170477221891e-05,
      "loss": 1.6051,
      "step": 765
    },
    {
      "epoch": 0.028130673589162714,
      "grad_norm": 0.3803512752056122,
      "learning_rate": 9.76809113347315e-05,
      "loss": 1.5364,
      "step": 766
    },
    {
      "epoch": 0.028167397706119843,
      "grad_norm": 0.39345091581344604,
      "learning_rate": 9.747012820690543e-05,
      "loss": 1.4482,
      "step": 767
    },
    {
      "epoch": 0.028204121823076977,
      "grad_norm": 0.40236222743988037,
      "learning_rate": 9.725935632579104e-05,
      "loss": 1.3884,
      "step": 768
    },
    {
      "epoch": 0.028240845940034107,
      "grad_norm": 0.41066014766693115,
      "learning_rate": 9.704859662838855e-05,
      "loss": 1.4161,
      "step": 769
    },
    {
      "epoch": 0.02827757005699124,
      "grad_norm": 0.4068271815776825,
      "learning_rate": 9.683785005164411e-05,
      "loss": 1.3632,
      "step": 770
    },
    {
      "epoch": 0.02831429417394837,
      "grad_norm": 0.40461504459381104,
      "learning_rate": 9.662711753244551e-05,
      "loss": 1.5967,
      "step": 771
    },
    {
      "epoch": 0.028351018290905503,
      "grad_norm": 0.4769410490989685,
      "learning_rate": 9.641640000761802e-05,
      "loss": 1.476,
      "step": 772
    },
    {
      "epoch": 0.028387742407862633,
      "grad_norm": 0.4702551066875458,
      "learning_rate": 9.620569841392029e-05,
      "loss": 1.363,
      "step": 773
    },
    {
      "epoch": 0.028424466524819766,
      "grad_norm": 0.5259777903556824,
      "learning_rate": 9.59950136880401e-05,
      "loss": 1.4726,
      "step": 774
    },
    {
      "epoch": 0.028461190641776896,
      "grad_norm": 0.4807552397251129,
      "learning_rate": 9.57843467665903e-05,
      "loss": 1.5359,
      "step": 775
    },
    {
      "epoch": 0.028497914758734026,
      "grad_norm": 0.5497702956199646,
      "learning_rate": 9.557369858610453e-05,
      "loss": 1.273,
      "step": 776
    },
    {
      "epoch": 0.02853463887569116,
      "grad_norm": 0.5623729228973389,
      "learning_rate": 9.53630700830332e-05,
      "loss": 1.1587,
      "step": 777
    },
    {
      "epoch": 0.02857136299264829,
      "grad_norm": 0.5789384245872498,
      "learning_rate": 9.51524621937391e-05,
      "loss": 1.4694,
      "step": 778
    },
    {
      "epoch": 0.028608087109605423,
      "grad_norm": 0.6016849279403687,
      "learning_rate": 9.494187585449358e-05,
      "loss": 1.1646,
      "step": 779
    },
    {
      "epoch": 0.028644811226562553,
      "grad_norm": 0.6032176613807678,
      "learning_rate": 9.473131200147205e-05,
      "loss": 1.3134,
      "step": 780
    },
    {
      "epoch": 0.028681535343519686,
      "grad_norm": 0.7069623470306396,
      "learning_rate": 9.452077157074994e-05,
      "loss": 1.3161,
      "step": 781
    },
    {
      "epoch": 0.028718259460476816,
      "grad_norm": 0.9594960808753967,
      "learning_rate": 9.431025549829862e-05,
      "loss": 1.4082,
      "step": 782
    },
    {
      "epoch": 0.02875498357743395,
      "grad_norm": 0.7513540387153625,
      "learning_rate": 9.409976471998118e-05,
      "loss": 1.2999,
      "step": 783
    },
    {
      "epoch": 0.02879170769439108,
      "grad_norm": 0.7618380784988403,
      "learning_rate": 9.388930017154819e-05,
      "loss": 1.1385,
      "step": 784
    },
    {
      "epoch": 0.028828431811348212,
      "grad_norm": 0.8093544244766235,
      "learning_rate": 9.367886278863366e-05,
      "loss": 1.1498,
      "step": 785
    },
    {
      "epoch": 0.028865155928305342,
      "grad_norm": 0.6843802332878113,
      "learning_rate": 9.346845350675088e-05,
      "loss": 1.086,
      "step": 786
    },
    {
      "epoch": 0.028901880045262476,
      "grad_norm": 0.7241513729095459,
      "learning_rate": 9.325807326128814e-05,
      "loss": 1.4161,
      "step": 787
    },
    {
      "epoch": 0.028938604162219606,
      "grad_norm": 0.944890022277832,
      "learning_rate": 9.304772298750463e-05,
      "loss": 1.1936,
      "step": 788
    },
    {
      "epoch": 0.028975328279176735,
      "grad_norm": 0.8273677825927734,
      "learning_rate": 9.283740362052642e-05,
      "loss": 1.2034,
      "step": 789
    },
    {
      "epoch": 0.02901205239613387,
      "grad_norm": 0.821712851524353,
      "learning_rate": 9.26271160953421e-05,
      "loss": 1.4431,
      "step": 790
    },
    {
      "epoch": 0.029048776513091,
      "grad_norm": 1.0124765634536743,
      "learning_rate": 9.241686134679867e-05,
      "loss": 1.2359,
      "step": 791
    },
    {
      "epoch": 0.029085500630048132,
      "grad_norm": 0.8707107305526733,
      "learning_rate": 9.220664030959749e-05,
      "loss": 1.0281,
      "step": 792
    },
    {
      "epoch": 0.029122224747005262,
      "grad_norm": 0.9230208992958069,
      "learning_rate": 9.199645391828999e-05,
      "loss": 1.1927,
      "step": 793
    },
    {
      "epoch": 0.029158948863962395,
      "grad_norm": 0.9754524230957031,
      "learning_rate": 9.178630310727365e-05,
      "loss": 1.0299,
      "step": 794
    },
    {
      "epoch": 0.029195672980919525,
      "grad_norm": 1.1014398336410522,
      "learning_rate": 9.157618881078772e-05,
      "loss": 0.8489,
      "step": 795
    },
    {
      "epoch": 0.02923239709787666,
      "grad_norm": 0.9878800511360168,
      "learning_rate": 9.136611196290915e-05,
      "loss": 1.0769,
      "step": 796
    },
    {
      "epoch": 0.02926912121483379,
      "grad_norm": 1.0553607940673828,
      "learning_rate": 9.115607349754834e-05,
      "loss": 1.0315,
      "step": 797
    },
    {
      "epoch": 0.02930584533179092,
      "grad_norm": 1.3339306116104126,
      "learning_rate": 9.094607434844523e-05,
      "loss": 1.0299,
      "step": 798
    },
    {
      "epoch": 0.02934256944874805,
      "grad_norm": 1.0148766040802002,
      "learning_rate": 9.07361154491648e-05,
      "loss": 0.7614,
      "step": 799
    },
    {
      "epoch": 0.029379293565705185,
      "grad_norm": 1.7734603881835938,
      "learning_rate": 9.052619773309317e-05,
      "loss": 1.124,
      "step": 800
    },
    {
      "epoch": 0.029416017682662315,
      "grad_norm": 0.180131196975708,
      "learning_rate": 9.031632213343339e-05,
      "loss": 1.7441,
      "step": 801
    },
    {
      "epoch": 0.029452741799619445,
      "grad_norm": 0.1895289570093155,
      "learning_rate": 9.01064895832012e-05,
      "loss": 1.587,
      "step": 802
    },
    {
      "epoch": 0.029489465916576578,
      "grad_norm": 0.20419828593730927,
      "learning_rate": 8.98967010152211e-05,
      "loss": 1.6683,
      "step": 803
    },
    {
      "epoch": 0.029526190033533708,
      "grad_norm": 0.21604055166244507,
      "learning_rate": 8.968695736212193e-05,
      "loss": 1.5141,
      "step": 804
    },
    {
      "epoch": 0.02956291415049084,
      "grad_norm": 0.24004197120666504,
      "learning_rate": 8.947725955633294e-05,
      "loss": 1.5442,
      "step": 805
    },
    {
      "epoch": 0.02959963826744797,
      "grad_norm": 0.2613770067691803,
      "learning_rate": 8.926760853007946e-05,
      "loss": 1.6769,
      "step": 806
    },
    {
      "epoch": 0.029636362384405104,
      "grad_norm": 0.274088978767395,
      "learning_rate": 8.905800521537905e-05,
      "loss": 1.4364,
      "step": 807
    },
    {
      "epoch": 0.029673086501362234,
      "grad_norm": 0.2878991961479187,
      "learning_rate": 8.884845054403699e-05,
      "loss": 1.4072,
      "step": 808
    },
    {
      "epoch": 0.029709810618319368,
      "grad_norm": 0.2797437012195587,
      "learning_rate": 8.863894544764236e-05,
      "loss": 1.5066,
      "step": 809
    },
    {
      "epoch": 0.029746534735276497,
      "grad_norm": 0.31455984711647034,
      "learning_rate": 8.84294908575639e-05,
      "loss": 1.5181,
      "step": 810
    },
    {
      "epoch": 0.02978325885223363,
      "grad_norm": 0.3306396007537842,
      "learning_rate": 8.822008770494572e-05,
      "loss": 1.1563,
      "step": 811
    },
    {
      "epoch": 0.02981998296919076,
      "grad_norm": 0.33453407883644104,
      "learning_rate": 8.801073692070337e-05,
      "loss": 1.5052,
      "step": 812
    },
    {
      "epoch": 0.029856707086147894,
      "grad_norm": 0.4377453327178955,
      "learning_rate": 8.780143943551954e-05,
      "loss": 1.5642,
      "step": 813
    },
    {
      "epoch": 0.029893431203105024,
      "grad_norm": 0.35234156250953674,
      "learning_rate": 8.759219617983999e-05,
      "loss": 1.5959,
      "step": 814
    },
    {
      "epoch": 0.029930155320062157,
      "grad_norm": 0.40431466698646545,
      "learning_rate": 8.738300808386935e-05,
      "loss": 1.3485,
      "step": 815
    },
    {
      "epoch": 0.029966879437019287,
      "grad_norm": 0.39255428314208984,
      "learning_rate": 8.717387607756713e-05,
      "loss": 1.4891,
      "step": 816
    },
    {
      "epoch": 0.030003603553976417,
      "grad_norm": 0.42981892824172974,
      "learning_rate": 8.696480109064342e-05,
      "loss": 1.5643,
      "step": 817
    },
    {
      "epoch": 0.03004032767093355,
      "grad_norm": 0.3940938115119934,
      "learning_rate": 8.675578405255485e-05,
      "loss": 1.4305,
      "step": 818
    },
    {
      "epoch": 0.03007705178789068,
      "grad_norm": 0.38230079412460327,
      "learning_rate": 8.654682589250038e-05,
      "loss": 1.2898,
      "step": 819
    },
    {
      "epoch": 0.030113775904847814,
      "grad_norm": 0.4427676200866699,
      "learning_rate": 8.633792753941733e-05,
      "loss": 1.4564,
      "step": 820
    },
    {
      "epoch": 0.030150500021804943,
      "grad_norm": 0.48113635182380676,
      "learning_rate": 8.612908992197705e-05,
      "loss": 1.3895,
      "step": 821
    },
    {
      "epoch": 0.030187224138762077,
      "grad_norm": 0.45686447620391846,
      "learning_rate": 8.592031396858093e-05,
      "loss": 1.5234,
      "step": 822
    },
    {
      "epoch": 0.030223948255719207,
      "grad_norm": 0.46626755595207214,
      "learning_rate": 8.571160060735624e-05,
      "loss": 1.2721,
      "step": 823
    },
    {
      "epoch": 0.03026067237267634,
      "grad_norm": 0.47862714529037476,
      "learning_rate": 8.550295076615188e-05,
      "loss": 1.3904,
      "step": 824
    },
    {
      "epoch": 0.03029739648963347,
      "grad_norm": 0.5480010509490967,
      "learning_rate": 8.529436537253458e-05,
      "loss": 1.5528,
      "step": 825
    },
    {
      "epoch": 0.030334120606590603,
      "grad_norm": 0.5171056985855103,
      "learning_rate": 8.508584535378439e-05,
      "loss": 1.2387,
      "step": 826
    },
    {
      "epoch": 0.030370844723547733,
      "grad_norm": 0.6371538639068604,
      "learning_rate": 8.487739163689079e-05,
      "loss": 1.3105,
      "step": 827
    },
    {
      "epoch": 0.030407568840504866,
      "grad_norm": 0.580216646194458,
      "learning_rate": 8.466900514854847e-05,
      "loss": 1.3231,
      "step": 828
    },
    {
      "epoch": 0.030444292957461996,
      "grad_norm": 0.6418307423591614,
      "learning_rate": 8.446068681515334e-05,
      "loss": 1.3021,
      "step": 829
    },
    {
      "epoch": 0.030481017074419126,
      "grad_norm": 0.5890262126922607,
      "learning_rate": 8.425243756279824e-05,
      "loss": 1.4005,
      "step": 830
    },
    {
      "epoch": 0.03051774119137626,
      "grad_norm": 0.8933138251304626,
      "learning_rate": 8.404425831726894e-05,
      "loss": 1.1844,
      "step": 831
    },
    {
      "epoch": 0.03055446530833339,
      "grad_norm": 0.6531504988670349,
      "learning_rate": 8.383615000404e-05,
      "loss": 1.1763,
      "step": 832
    },
    {
      "epoch": 0.030591189425290523,
      "grad_norm": 0.7496176958084106,
      "learning_rate": 8.362811354827059e-05,
      "loss": 1.4384,
      "step": 833
    },
    {
      "epoch": 0.030627913542247653,
      "grad_norm": 0.6474002003669739,
      "learning_rate": 8.342014987480047e-05,
      "loss": 1.2784,
      "step": 834
    },
    {
      "epoch": 0.030664637659204786,
      "grad_norm": 0.7260277271270752,
      "learning_rate": 8.321225990814591e-05,
      "loss": 1.333,
      "step": 835
    },
    {
      "epoch": 0.030701361776161916,
      "grad_norm": 0.7249194383621216,
      "learning_rate": 8.300444457249543e-05,
      "loss": 0.8566,
      "step": 836
    },
    {
      "epoch": 0.03073808589311905,
      "grad_norm": 0.6852389574050903,
      "learning_rate": 8.279670479170573e-05,
      "loss": 0.897,
      "step": 837
    },
    {
      "epoch": 0.03077481001007618,
      "grad_norm": 0.735329270362854,
      "learning_rate": 8.258904148929775e-05,
      "loss": 0.9921,
      "step": 838
    },
    {
      "epoch": 0.030811534127033312,
      "grad_norm": 0.9245840907096863,
      "learning_rate": 8.238145558845235e-05,
      "loss": 1.0928,
      "step": 839
    },
    {
      "epoch": 0.030848258243990442,
      "grad_norm": 0.8907131552696228,
      "learning_rate": 8.217394801200631e-05,
      "loss": 1.0017,
      "step": 840
    },
    {
      "epoch": 0.030884982360947576,
      "grad_norm": 1.036307692527771,
      "learning_rate": 8.196651968244826e-05,
      "loss": 1.1678,
      "step": 841
    },
    {
      "epoch": 0.030921706477904706,
      "grad_norm": 1.0132153034210205,
      "learning_rate": 8.175917152191447e-05,
      "loss": 1.1477,
      "step": 842
    },
    {
      "epoch": 0.030958430594861835,
      "grad_norm": 1.0038292407989502,
      "learning_rate": 8.15519044521848e-05,
      "loss": 1.123,
      "step": 843
    },
    {
      "epoch": 0.03099515471181897,
      "grad_norm": 1.0851421356201172,
      "learning_rate": 8.134471939467874e-05,
      "loss": 1.1991,
      "step": 844
    },
    {
      "epoch": 0.0310318788287761,
      "grad_norm": 1.029754638671875,
      "learning_rate": 8.113761727045105e-05,
      "loss": 1.3549,
      "step": 845
    },
    {
      "epoch": 0.031068602945733232,
      "grad_norm": 1.1511602401733398,
      "learning_rate": 8.093059900018792e-05,
      "loss": 1.1348,
      "step": 846
    },
    {
      "epoch": 0.031105327062690362,
      "grad_norm": 1.1457711458206177,
      "learning_rate": 8.072366550420266e-05,
      "loss": 1.2294,
      "step": 847
    },
    {
      "epoch": 0.031142051179647495,
      "grad_norm": 1.1224700212478638,
      "learning_rate": 8.051681770243175e-05,
      "loss": 0.9537,
      "step": 848
    },
    {
      "epoch": 0.031178775296604625,
      "grad_norm": 1.3428674936294556,
      "learning_rate": 8.031005651443073e-05,
      "loss": 1.2428,
      "step": 849
    },
    {
      "epoch": 0.03121549941356176,
      "grad_norm": 1.904398798942566,
      "learning_rate": 8.010338285937006e-05,
      "loss": 1.2804,
      "step": 850
    },
    {
      "epoch": 0.03125222353051889,
      "grad_norm": 0.15849736332893372,
      "learning_rate": 7.989679765603108e-05,
      "loss": 1.5147,
      "step": 851
    },
    {
      "epoch": 0.03128894764747602,
      "grad_norm": 0.1855199635028839,
      "learning_rate": 7.969030182280192e-05,
      "loss": 1.7363,
      "step": 852
    },
    {
      "epoch": 0.031325671764433155,
      "grad_norm": 0.1991463452577591,
      "learning_rate": 7.948389627767343e-05,
      "loss": 1.6589,
      "step": 853
    },
    {
      "epoch": 0.03136239588139028,
      "grad_norm": 0.22372905910015106,
      "learning_rate": 7.927758193823501e-05,
      "loss": 1.6706,
      "step": 854
    },
    {
      "epoch": 0.031399119998347415,
      "grad_norm": 0.21848812699317932,
      "learning_rate": 7.907135972167069e-05,
      "loss": 1.6816,
      "step": 855
    },
    {
      "epoch": 0.03143584411530455,
      "grad_norm": 0.23211084306240082,
      "learning_rate": 7.88652305447549e-05,
      "loss": 1.472,
      "step": 856
    },
    {
      "epoch": 0.031472568232261675,
      "grad_norm": 0.281437486410141,
      "learning_rate": 7.865919532384844e-05,
      "loss": 1.4305,
      "step": 857
    },
    {
      "epoch": 0.03150929234921881,
      "grad_norm": 0.27912604808807373,
      "learning_rate": 7.845325497489449e-05,
      "loss": 1.6768,
      "step": 858
    },
    {
      "epoch": 0.03154601646617594,
      "grad_norm": 0.3138715326786041,
      "learning_rate": 7.82474104134144e-05,
      "loss": 1.5919,
      "step": 859
    },
    {
      "epoch": 0.031582740583133075,
      "grad_norm": 0.29909583926200867,
      "learning_rate": 7.804166255450373e-05,
      "loss": 1.4783,
      "step": 860
    },
    {
      "epoch": 0.0316194647000902,
      "grad_norm": 0.323032408952713,
      "learning_rate": 7.783601231282812e-05,
      "loss": 1.4964,
      "step": 861
    },
    {
      "epoch": 0.031656188817047334,
      "grad_norm": 0.34091702103614807,
      "learning_rate": 7.763046060261932e-05,
      "loss": 1.3875,
      "step": 862
    },
    {
      "epoch": 0.03169291293400447,
      "grad_norm": 0.319950133562088,
      "learning_rate": 7.742500833767094e-05,
      "loss": 1.5038,
      "step": 863
    },
    {
      "epoch": 0.0317296370509616,
      "grad_norm": 0.37528109550476074,
      "learning_rate": 7.721965643133458e-05,
      "loss": 1.3913,
      "step": 864
    },
    {
      "epoch": 0.03176636116791873,
      "grad_norm": 0.36791715025901794,
      "learning_rate": 7.701440579651566e-05,
      "loss": 1.465,
      "step": 865
    },
    {
      "epoch": 0.03180308528487586,
      "grad_norm": 0.4140302240848541,
      "learning_rate": 7.680925734566937e-05,
      "loss": 1.3074,
      "step": 866
    },
    {
      "epoch": 0.031839809401832994,
      "grad_norm": 0.4149440824985504,
      "learning_rate": 7.660421199079669e-05,
      "loss": 1.2866,
      "step": 867
    },
    {
      "epoch": 0.03187653351879012,
      "grad_norm": 0.43856704235076904,
      "learning_rate": 7.639927064344022e-05,
      "loss": 1.6006,
      "step": 868
    },
    {
      "epoch": 0.031913257635747254,
      "grad_norm": 0.39718830585479736,
      "learning_rate": 7.619443421468021e-05,
      "loss": 1.4567,
      "step": 869
    },
    {
      "epoch": 0.03194998175270439,
      "grad_norm": 0.4931643009185791,
      "learning_rate": 7.598970361513051e-05,
      "loss": 1.4044,
      "step": 870
    },
    {
      "epoch": 0.03198670586966152,
      "grad_norm": 0.46928489208221436,
      "learning_rate": 7.578507975493448e-05,
      "loss": 1.5276,
      "step": 871
    },
    {
      "epoch": 0.03202342998661865,
      "grad_norm": 0.48663702607154846,
      "learning_rate": 7.558056354376098e-05,
      "loss": 1.373,
      "step": 872
    },
    {
      "epoch": 0.03206015410357578,
      "grad_norm": 0.5179852247238159,
      "learning_rate": 7.537615589080027e-05,
      "loss": 1.4413,
      "step": 873
    },
    {
      "epoch": 0.032096878220532914,
      "grad_norm": 0.531245231628418,
      "learning_rate": 7.517185770476006e-05,
      "loss": 1.602,
      "step": 874
    },
    {
      "epoch": 0.03213360233749005,
      "grad_norm": 0.5570663809776306,
      "learning_rate": 7.496766989386136e-05,
      "loss": 1.2532,
      "step": 875
    },
    {
      "epoch": 0.03217032645444717,
      "grad_norm": 0.6418626308441162,
      "learning_rate": 7.476359336583454e-05,
      "loss": 1.2906,
      "step": 876
    },
    {
      "epoch": 0.03220705057140431,
      "grad_norm": 0.5781128406524658,
      "learning_rate": 7.455962902791522e-05,
      "loss": 1.2811,
      "step": 877
    },
    {
      "epoch": 0.03224377468836144,
      "grad_norm": 0.5941858887672424,
      "learning_rate": 7.435577778684033e-05,
      "loss": 1.1388,
      "step": 878
    },
    {
      "epoch": 0.03228049880531857,
      "grad_norm": 0.703433632850647,
      "learning_rate": 7.415204054884399e-05,
      "loss": 1.363,
      "step": 879
    },
    {
      "epoch": 0.0323172229222757,
      "grad_norm": 0.6535505056381226,
      "learning_rate": 7.394841821965345e-05,
      "loss": 1.3276,
      "step": 880
    },
    {
      "epoch": 0.03235394703923283,
      "grad_norm": 0.6709651350975037,
      "learning_rate": 7.374491170448525e-05,
      "loss": 1.3351,
      "step": 881
    },
    {
      "epoch": 0.03239067115618997,
      "grad_norm": 0.6895983219146729,
      "learning_rate": 7.3541521908041e-05,
      "loss": 1.119,
      "step": 882
    },
    {
      "epoch": 0.03242739527314709,
      "grad_norm": 0.6529479026794434,
      "learning_rate": 7.33382497345034e-05,
      "loss": 1.1771,
      "step": 883
    },
    {
      "epoch": 0.032464119390104226,
      "grad_norm": 0.8023227453231812,
      "learning_rate": 7.313509608753231e-05,
      "loss": 1.1731,
      "step": 884
    },
    {
      "epoch": 0.03250084350706136,
      "grad_norm": 0.7079395055770874,
      "learning_rate": 7.293206187026061e-05,
      "loss": 1.0021,
      "step": 885
    },
    {
      "epoch": 0.03253756762401849,
      "grad_norm": 0.8687023520469666,
      "learning_rate": 7.27291479852903e-05,
      "loss": 1.3393,
      "step": 886
    },
    {
      "epoch": 0.03257429174097562,
      "grad_norm": 0.7452741861343384,
      "learning_rate": 7.252635533468843e-05,
      "loss": 1.2218,
      "step": 887
    },
    {
      "epoch": 0.03261101585793275,
      "grad_norm": 0.8241674304008484,
      "learning_rate": 7.232368481998309e-05,
      "loss": 1.3897,
      "step": 888
    },
    {
      "epoch": 0.032647739974889886,
      "grad_norm": 0.8692129254341125,
      "learning_rate": 7.212113734215932e-05,
      "loss": 1.295,
      "step": 889
    },
    {
      "epoch": 0.03268446409184702,
      "grad_norm": 0.7841311693191528,
      "learning_rate": 7.191871380165538e-05,
      "loss": 1.3232,
      "step": 890
    },
    {
      "epoch": 0.032721188208804146,
      "grad_norm": 0.8816729784011841,
      "learning_rate": 7.17164150983584e-05,
      "loss": 1.0383,
      "step": 891
    },
    {
      "epoch": 0.03275791232576128,
      "grad_norm": 0.776283323764801,
      "learning_rate": 7.151424213160061e-05,
      "loss": 0.7598,
      "step": 892
    },
    {
      "epoch": 0.03279463644271841,
      "grad_norm": 0.8918838500976562,
      "learning_rate": 7.131219580015521e-05,
      "loss": 1.1468,
      "step": 893
    },
    {
      "epoch": 0.032831360559675546,
      "grad_norm": 1.2128825187683105,
      "learning_rate": 7.11102770022325e-05,
      "loss": 1.0311,
      "step": 894
    },
    {
      "epoch": 0.03286808467663267,
      "grad_norm": 1.158266544342041,
      "learning_rate": 7.090848663547574e-05,
      "loss": 1.2435,
      "step": 895
    },
    {
      "epoch": 0.032904808793589806,
      "grad_norm": 1.0083034038543701,
      "learning_rate": 7.070682559695736e-05,
      "loss": 0.9549,
      "step": 896
    },
    {
      "epoch": 0.03294153291054694,
      "grad_norm": 1.1738016605377197,
      "learning_rate": 7.050529478317476e-05,
      "loss": 1.0923,
      "step": 897
    },
    {
      "epoch": 0.032978257027504065,
      "grad_norm": 0.9966412782669067,
      "learning_rate": 7.03038950900464e-05,
      "loss": 0.8258,
      "step": 898
    },
    {
      "epoch": 0.0330149811444612,
      "grad_norm": 1.2997379302978516,
      "learning_rate": 7.010262741290798e-05,
      "loss": 0.8997,
      "step": 899
    },
    {
      "epoch": 0.03305170526141833,
      "grad_norm": 2.036944627761841,
      "learning_rate": 6.990149264650814e-05,
      "loss": 1.1026,
      "step": 900
    },
    {
      "epoch": 0.033088429378375465,
      "grad_norm": 0.16966252028942108,
      "learning_rate": 6.970049168500474e-05,
      "loss": 1.4564,
      "step": 901
    },
    {
      "epoch": 0.03312515349533259,
      "grad_norm": 0.22211810946464539,
      "learning_rate": 6.94996254219608e-05,
      "loss": 1.6675,
      "step": 902
    },
    {
      "epoch": 0.033161877612289725,
      "grad_norm": 0.2073749154806137,
      "learning_rate": 6.929889475034048e-05,
      "loss": 1.7664,
      "step": 903
    },
    {
      "epoch": 0.03319860172924686,
      "grad_norm": 0.24061159789562225,
      "learning_rate": 6.909830056250527e-05,
      "loss": 1.5661,
      "step": 904
    },
    {
      "epoch": 0.03323532584620399,
      "grad_norm": 0.23003491759300232,
      "learning_rate": 6.889784375020978e-05,
      "loss": 1.3533,
      "step": 905
    },
    {
      "epoch": 0.03327204996316112,
      "grad_norm": 0.2518197000026703,
      "learning_rate": 6.869752520459803e-05,
      "loss": 1.6381,
      "step": 906
    },
    {
      "epoch": 0.03330877408011825,
      "grad_norm": 0.2447218894958496,
      "learning_rate": 6.849734581619918e-05,
      "loss": 1.3859,
      "step": 907
    },
    {
      "epoch": 0.033345498197075385,
      "grad_norm": 0.2801475524902344,
      "learning_rate": 6.829730647492404e-05,
      "loss": 1.5005,
      "step": 908
    },
    {
      "epoch": 0.03338222231403251,
      "grad_norm": 0.293434202671051,
      "learning_rate": 6.80974080700606e-05,
      "loss": 1.4905,
      "step": 909
    },
    {
      "epoch": 0.033418946430989645,
      "grad_norm": 0.3092429041862488,
      "learning_rate": 6.789765149027039e-05,
      "loss": 1.4553,
      "step": 910
    },
    {
      "epoch": 0.03345567054794678,
      "grad_norm": 0.3090967535972595,
      "learning_rate": 6.769803762358443e-05,
      "loss": 1.3481,
      "step": 911
    },
    {
      "epoch": 0.03349239466490391,
      "grad_norm": 0.3487565815448761,
      "learning_rate": 6.749856735739928e-05,
      "loss": 1.4567,
      "step": 912
    },
    {
      "epoch": 0.03352911878186104,
      "grad_norm": 0.3138130307197571,
      "learning_rate": 6.729924157847323e-05,
      "loss": 1.6106,
      "step": 913
    },
    {
      "epoch": 0.03356584289881817,
      "grad_norm": 0.3384358882904053,
      "learning_rate": 6.710006117292209e-05,
      "loss": 1.2935,
      "step": 914
    },
    {
      "epoch": 0.033602567015775305,
      "grad_norm": 0.3508756756782532,
      "learning_rate": 6.690102702621548e-05,
      "loss": 1.6955,
      "step": 915
    },
    {
      "epoch": 0.03363929113273244,
      "grad_norm": 0.35203954577445984,
      "learning_rate": 6.670214002317278e-05,
      "loss": 1.2355,
      "step": 916
    },
    {
      "epoch": 0.033676015249689564,
      "grad_norm": 0.4230559766292572,
      "learning_rate": 6.650340104795932e-05,
      "loss": 1.466,
      "step": 917
    },
    {
      "epoch": 0.0337127393666467,
      "grad_norm": 0.4085327386856079,
      "learning_rate": 6.630481098408228e-05,
      "loss": 1.5455,
      "step": 918
    },
    {
      "epoch": 0.03374946348360383,
      "grad_norm": 0.4054946005344391,
      "learning_rate": 6.610637071438686e-05,
      "loss": 1.4008,
      "step": 919
    },
    {
      "epoch": 0.033786187600560964,
      "grad_norm": 0.39696818590164185,
      "learning_rate": 6.590808112105232e-05,
      "loss": 1.4549,
      "step": 920
    },
    {
      "epoch": 0.03382291171751809,
      "grad_norm": 0.47683775424957275,
      "learning_rate": 6.570994308558812e-05,
      "loss": 1.49,
      "step": 921
    },
    {
      "epoch": 0.033859635834475224,
      "grad_norm": 0.4491752088069916,
      "learning_rate": 6.551195748882997e-05,
      "loss": 1.3789,
      "step": 922
    },
    {
      "epoch": 0.03389635995143236,
      "grad_norm": 0.49369654059410095,
      "learning_rate": 6.531412521093586e-05,
      "loss": 1.534,
      "step": 923
    },
    {
      "epoch": 0.033933084068389484,
      "grad_norm": 0.44287335872650146,
      "learning_rate": 6.51164471313822e-05,
      "loss": 1.2561,
      "step": 924
    },
    {
      "epoch": 0.03396980818534662,
      "grad_norm": 0.5025992393493652,
      "learning_rate": 6.491892412895995e-05,
      "loss": 1.3094,
      "step": 925
    },
    {
      "epoch": 0.03400653230230375,
      "grad_norm": 0.5359137058258057,
      "learning_rate": 6.472155708177052e-05,
      "loss": 1.3072,
      "step": 926
    },
    {
      "epoch": 0.034043256419260884,
      "grad_norm": 0.5345491766929626,
      "learning_rate": 6.452434686722224e-05,
      "loss": 1.1052,
      "step": 927
    },
    {
      "epoch": 0.03407998053621801,
      "grad_norm": 0.6920292973518372,
      "learning_rate": 6.432729436202604e-05,
      "loss": 1.5481,
      "step": 928
    },
    {
      "epoch": 0.034116704653175144,
      "grad_norm": 0.6048431992530823,
      "learning_rate": 6.41304004421918e-05,
      "loss": 1.3901,
      "step": 929
    },
    {
      "epoch": 0.03415342877013228,
      "grad_norm": 0.6300055980682373,
      "learning_rate": 6.393366598302446e-05,
      "loss": 1.225,
      "step": 930
    },
    {
      "epoch": 0.03419015288708941,
      "grad_norm": 0.6064823269844055,
      "learning_rate": 6.373709185911998e-05,
      "loss": 1.5015,
      "step": 931
    },
    {
      "epoch": 0.03422687700404654,
      "grad_norm": 0.6306756734848022,
      "learning_rate": 6.354067894436155e-05,
      "loss": 1.1927,
      "step": 932
    },
    {
      "epoch": 0.03426360112100367,
      "grad_norm": 0.6955969333648682,
      "learning_rate": 6.334442811191576e-05,
      "loss": 1.1206,
      "step": 933
    },
    {
      "epoch": 0.0343003252379608,
      "grad_norm": 0.6617854237556458,
      "learning_rate": 6.314834023422858e-05,
      "loss": 1.1482,
      "step": 934
    },
    {
      "epoch": 0.03433704935491793,
      "grad_norm": 0.8001710772514343,
      "learning_rate": 6.295241618302156e-05,
      "loss": 1.1623,
      "step": 935
    },
    {
      "epoch": 0.03437377347187506,
      "grad_norm": 0.7562992572784424,
      "learning_rate": 6.275665682928803e-05,
      "loss": 1.2412,
      "step": 936
    },
    {
      "epoch": 0.034410497588832197,
      "grad_norm": 0.6859000325202942,
      "learning_rate": 6.256106304328905e-05,
      "loss": 1.0531,
      "step": 937
    },
    {
      "epoch": 0.03444722170578933,
      "grad_norm": 0.9365538954734802,
      "learning_rate": 6.23656356945497e-05,
      "loss": 0.9439,
      "step": 938
    },
    {
      "epoch": 0.034483945822746456,
      "grad_norm": 0.8586938977241516,
      "learning_rate": 6.21703756518551e-05,
      "loss": 1.1161,
      "step": 939
    },
    {
      "epoch": 0.03452066993970359,
      "grad_norm": 0.8558727502822876,
      "learning_rate": 6.197528378324665e-05,
      "loss": 1.219,
      "step": 940
    },
    {
      "epoch": 0.03455739405666072,
      "grad_norm": 0.8667901754379272,
      "learning_rate": 6.17803609560181e-05,
      "loss": 1.2651,
      "step": 941
    },
    {
      "epoch": 0.034594118173617856,
      "grad_norm": 0.9176797866821289,
      "learning_rate": 6.158560803671168e-05,
      "loss": 0.9935,
      "step": 942
    },
    {
      "epoch": 0.03463084229057498,
      "grad_norm": 0.8343192934989929,
      "learning_rate": 6.139102589111435e-05,
      "loss": 1.2628,
      "step": 943
    },
    {
      "epoch": 0.034667566407532116,
      "grad_norm": 0.8579095602035522,
      "learning_rate": 6.119661538425381e-05,
      "loss": 0.6792,
      "step": 944
    },
    {
      "epoch": 0.03470429052448925,
      "grad_norm": 1.0400512218475342,
      "learning_rate": 6.100237738039484e-05,
      "loss": 1.2351,
      "step": 945
    },
    {
      "epoch": 0.03474101464144638,
      "grad_norm": 1.0240145921707153,
      "learning_rate": 6.0808312743035236e-05,
      "loss": 1.075,
      "step": 946
    },
    {
      "epoch": 0.03477773875840351,
      "grad_norm": 1.0353449583053589,
      "learning_rate": 6.061442233490211e-05,
      "loss": 1.2257,
      "step": 947
    },
    {
      "epoch": 0.03481446287536064,
      "grad_norm": 1.0081970691680908,
      "learning_rate": 6.042070701794806e-05,
      "loss": 0.7641,
      "step": 948
    },
    {
      "epoch": 0.034851186992317776,
      "grad_norm": 1.148250699043274,
      "learning_rate": 6.0227167653347305e-05,
      "loss": 1.0207,
      "step": 949
    },
    {
      "epoch": 0.0348879111092749,
      "grad_norm": 1.6070648431777954,
      "learning_rate": 6.0033805101491794e-05,
      "loss": 0.9477,
      "step": 950
    },
    {
      "epoch": 0.034924635226232036,
      "grad_norm": 0.1640123575925827,
      "learning_rate": 5.98406202219875e-05,
      "loss": 1.5857,
      "step": 951
    },
    {
      "epoch": 0.03496135934318917,
      "grad_norm": 0.1892043799161911,
      "learning_rate": 5.964761387365052e-05,
      "loss": 1.7321,
      "step": 952
    },
    {
      "epoch": 0.0349980834601463,
      "grad_norm": 0.19833044707775116,
      "learning_rate": 5.9454786914503255e-05,
      "loss": 1.5357,
      "step": 953
    },
    {
      "epoch": 0.03503480757710343,
      "grad_norm": 0.2202310562133789,
      "learning_rate": 5.926214020177074e-05,
      "loss": 1.6671,
      "step": 954
    },
    {
      "epoch": 0.03507153169406056,
      "grad_norm": 0.22386738657951355,
      "learning_rate": 5.9069674591876534e-05,
      "loss": 1.3449,
      "step": 955
    },
    {
      "epoch": 0.035108255811017695,
      "grad_norm": 0.2465665191411972,
      "learning_rate": 5.887739094043923e-05,
      "loss": 1.5598,
      "step": 956
    },
    {
      "epoch": 0.03514497992797483,
      "grad_norm": 0.24554035067558289,
      "learning_rate": 5.868529010226845e-05,
      "loss": 1.5363,
      "step": 957
    },
    {
      "epoch": 0.035181704044931955,
      "grad_norm": 0.25635093450546265,
      "learning_rate": 5.849337293136112e-05,
      "loss": 1.5247,
      "step": 958
    },
    {
      "epoch": 0.03521842816188909,
      "grad_norm": 0.2762475311756134,
      "learning_rate": 5.830164028089766e-05,
      "loss": 1.403,
      "step": 959
    },
    {
      "epoch": 0.03525515227884622,
      "grad_norm": 0.2849908769130707,
      "learning_rate": 5.811009300323818e-05,
      "loss": 1.1713,
      "step": 960
    },
    {
      "epoch": 0.03529187639580335,
      "grad_norm": 0.30814555287361145,
      "learning_rate": 5.791873194991872e-05,
      "loss": 1.4004,
      "step": 961
    },
    {
      "epoch": 0.03532860051276048,
      "grad_norm": 0.3120114207267761,
      "learning_rate": 5.7727557971647427e-05,
      "loss": 1.5252,
      "step": 962
    },
    {
      "epoch": 0.035365324629717615,
      "grad_norm": 0.31676068902015686,
      "learning_rate": 5.7536571918300864e-05,
      "loss": 1.4518,
      "step": 963
    },
    {
      "epoch": 0.03540204874667475,
      "grad_norm": 0.3428522050380707,
      "learning_rate": 5.734577463892008e-05,
      "loss": 1.2854,
      "step": 964
    },
    {
      "epoch": 0.035438772863631875,
      "grad_norm": 0.3547038435935974,
      "learning_rate": 5.7155166981706956e-05,
      "loss": 1.4033,
      "step": 965
    },
    {
      "epoch": 0.03547549698058901,
      "grad_norm": 0.3325125277042389,
      "learning_rate": 5.6964749794020354e-05,
      "loss": 1.2583,
      "step": 966
    },
    {
      "epoch": 0.03551222109754614,
      "grad_norm": 0.3902745842933655,
      "learning_rate": 5.6774523922372394e-05,
      "loss": 1.446,
      "step": 967
    },
    {
      "epoch": 0.035548945214503275,
      "grad_norm": 0.3968910872936249,
      "learning_rate": 5.6584490212424804e-05,
      "loss": 1.4061,
      "step": 968
    },
    {
      "epoch": 0.0355856693314604,
      "grad_norm": 0.41748133301734924,
      "learning_rate": 5.639464950898491e-05,
      "loss": 1.2681,
      "step": 969
    },
    {
      "epoch": 0.035622393448417534,
      "grad_norm": 0.47420281171798706,
      "learning_rate": 5.620500265600206e-05,
      "loss": 1.4535,
      "step": 970
    },
    {
      "epoch": 0.03565911756537467,
      "grad_norm": 0.4299311339855194,
      "learning_rate": 5.601555049656382e-05,
      "loss": 1.4835,
      "step": 971
    },
    {
      "epoch": 0.0356958416823318,
      "grad_norm": 0.433916836977005,
      "learning_rate": 5.58262938728922e-05,
      "loss": 1.2611,
      "step": 972
    },
    {
      "epoch": 0.03573256579928893,
      "grad_norm": 0.5343939065933228,
      "learning_rate": 5.563723362634008e-05,
      "loss": 1.7834,
      "step": 973
    },
    {
      "epoch": 0.03576928991624606,
      "grad_norm": 0.506473183631897,
      "learning_rate": 5.544837059738719e-05,
      "loss": 1.3761,
      "step": 974
    },
    {
      "epoch": 0.035806014033203194,
      "grad_norm": 0.5955891013145447,
      "learning_rate": 5.525970562563656e-05,
      "loss": 1.4012,
      "step": 975
    },
    {
      "epoch": 0.03584273815016032,
      "grad_norm": 0.6160305738449097,
      "learning_rate": 5.507123954981073e-05,
      "loss": 1.421,
      "step": 976
    },
    {
      "epoch": 0.035879462267117454,
      "grad_norm": 0.6444937586784363,
      "learning_rate": 5.488297320774807e-05,
      "loss": 1.5373,
      "step": 977
    },
    {
      "epoch": 0.03591618638407459,
      "grad_norm": 0.6434566974639893,
      "learning_rate": 5.4694907436399e-05,
      "loss": 1.3468,
      "step": 978
    },
    {
      "epoch": 0.03595291050103172,
      "grad_norm": 0.6509642004966736,
      "learning_rate": 5.4507043071822284e-05,
      "loss": 1.2497,
      "step": 979
    },
    {
      "epoch": 0.03598963461798885,
      "grad_norm": 0.6542109251022339,
      "learning_rate": 5.431938094918132e-05,
      "loss": 1.0581,
      "step": 980
    },
    {
      "epoch": 0.03602635873494598,
      "grad_norm": 0.7372477054595947,
      "learning_rate": 5.41319219027404e-05,
      "loss": 1.1396,
      "step": 981
    },
    {
      "epoch": 0.036063082851903114,
      "grad_norm": 0.7016234993934631,
      "learning_rate": 5.394466676586114e-05,
      "loss": 1.4144,
      "step": 982
    },
    {
      "epoch": 0.03609980696886025,
      "grad_norm": 0.735110342502594,
      "learning_rate": 5.375761637099854e-05,
      "loss": 1.0295,
      "step": 983
    },
    {
      "epoch": 0.036136531085817374,
      "grad_norm": 0.794787585735321,
      "learning_rate": 5.357077154969742e-05,
      "loss": 1.3801,
      "step": 984
    },
    {
      "epoch": 0.03617325520277451,
      "grad_norm": 0.7793825268745422,
      "learning_rate": 5.3384133132588784e-05,
      "loss": 1.2792,
      "step": 985
    },
    {
      "epoch": 0.03620997931973164,
      "grad_norm": 0.9367470741271973,
      "learning_rate": 5.3197701949386e-05,
      "loss": 0.998,
      "step": 986
    },
    {
      "epoch": 0.03624670343668877,
      "grad_norm": 0.8130968809127808,
      "learning_rate": 5.301147882888116e-05,
      "loss": 1.3447,
      "step": 987
    },
    {
      "epoch": 0.0362834275536459,
      "grad_norm": 0.8137392997741699,
      "learning_rate": 5.28254645989414e-05,
      "loss": 1.0821,
      "step": 988
    },
    {
      "epoch": 0.03632015167060303,
      "grad_norm": 0.8557196259498596,
      "learning_rate": 5.2639660086505226e-05,
      "loss": 1.3469,
      "step": 989
    },
    {
      "epoch": 0.03635687578756017,
      "grad_norm": 0.8427239060401917,
      "learning_rate": 5.2454066117578815e-05,
      "loss": 1.1861,
      "step": 990
    },
    {
      "epoch": 0.03639359990451729,
      "grad_norm": 0.9311249256134033,
      "learning_rate": 5.226868351723244e-05,
      "loss": 1.2783,
      "step": 991
    },
    {
      "epoch": 0.036430324021474426,
      "grad_norm": 1.0148221254348755,
      "learning_rate": 5.2083513109596616e-05,
      "loss": 1.0463,
      "step": 992
    },
    {
      "epoch": 0.03646704813843156,
      "grad_norm": 0.8479374647140503,
      "learning_rate": 5.189855571785859e-05,
      "loss": 1.1934,
      "step": 993
    },
    {
      "epoch": 0.03650377225538869,
      "grad_norm": 1.0229803323745728,
      "learning_rate": 5.171381216425863e-05,
      "loss": 1.04,
      "step": 994
    },
    {
      "epoch": 0.03654049637234582,
      "grad_norm": 1.1014984846115112,
      "learning_rate": 5.152928327008635e-05,
      "loss": 0.7685,
      "step": 995
    },
    {
      "epoch": 0.03657722048930295,
      "grad_norm": 0.9733578562736511,
      "learning_rate": 5.134496985567714e-05,
      "loss": 0.9987,
      "step": 996
    },
    {
      "epoch": 0.036613944606260086,
      "grad_norm": 1.2694830894470215,
      "learning_rate": 5.116087274040837e-05,
      "loss": 1.0653,
      "step": 997
    },
    {
      "epoch": 0.03665066872321722,
      "grad_norm": 1.247896671295166,
      "learning_rate": 5.0976992742695925e-05,
      "loss": 1.1303,
      "step": 998
    },
    {
      "epoch": 0.036687392840174346,
      "grad_norm": 1.300336480140686,
      "learning_rate": 5.07933306799904e-05,
      "loss": 1.0243,
      "step": 999
    },
    {
      "epoch": 0.03672411695713148,
      "grad_norm": 1.7647722959518433,
      "learning_rate": 5.060988736877366e-05,
      "loss": 1.2175,
      "step": 1000
    },
    {
      "epoch": 0.03676084107408861,
      "grad_norm": 0.15511269867420197,
      "learning_rate": 5.042666362455498e-05,
      "loss": 1.377,
      "step": 1001
    },
    {
      "epoch": 0.03679756519104574,
      "grad_norm": 0.1891482025384903,
      "learning_rate": 5.024366026186755e-05,
      "loss": 1.6238,
      "step": 1002
    },
    {
      "epoch": 0.03683428930800287,
      "grad_norm": 0.2193506360054016,
      "learning_rate": 5.006087809426496e-05,
      "loss": 1.6859,
      "step": 1003
    },
    {
      "epoch": 0.036871013424960006,
      "grad_norm": 0.23381005227565765,
      "learning_rate": 4.987831793431731e-05,
      "loss": 1.741,
      "step": 1004
    },
    {
      "epoch": 0.03690773754191714,
      "grad_norm": 0.22946472465991974,
      "learning_rate": 4.9695980593607817e-05,
      "loss": 1.4081,
      "step": 1005
    },
    {
      "epoch": 0.036944461658874266,
      "grad_norm": 0.254079669713974,
      "learning_rate": 4.9513866882729146e-05,
      "loss": 1.5159,
      "step": 1006
    },
    {
      "epoch": 0.0369811857758314,
      "grad_norm": 0.2652171850204468,
      "learning_rate": 4.9331977611279777e-05,
      "loss": 1.4236,
      "step": 1007
    },
    {
      "epoch": 0.03701790989278853,
      "grad_norm": 0.27512261271476746,
      "learning_rate": 4.9150313587860433e-05,
      "loss": 1.5148,
      "step": 1008
    },
    {
      "epoch": 0.037054634009745666,
      "grad_norm": 0.2959429621696472,
      "learning_rate": 4.896887562007054e-05,
      "loss": 1.373,
      "step": 1009
    },
    {
      "epoch": 0.03709135812670279,
      "grad_norm": 0.308264821767807,
      "learning_rate": 4.8787664514504504e-05,
      "loss": 1.2208,
      "step": 1010
    },
    {
      "epoch": 0.037128082243659925,
      "grad_norm": 0.3090200126171112,
      "learning_rate": 4.860668107674823e-05,
      "loss": 1.5556,
      "step": 1011
    },
    {
      "epoch": 0.03716480636061706,
      "grad_norm": 0.32343941926956177,
      "learning_rate": 4.8425926111375506e-05,
      "loss": 1.3249,
      "step": 1012
    },
    {
      "epoch": 0.037201530477574185,
      "grad_norm": 0.3529990613460541,
      "learning_rate": 4.824540042194443e-05,
      "loss": 1.5299,
      "step": 1013
    },
    {
      "epoch": 0.03723825459453132,
      "grad_norm": 0.3330044448375702,
      "learning_rate": 4.8065104810993856e-05,
      "loss": 1.3015,
      "step": 1014
    },
    {
      "epoch": 0.03727497871148845,
      "grad_norm": 0.3531375229358673,
      "learning_rate": 4.788504008003978e-05,
      "loss": 1.56,
      "step": 1015
    },
    {
      "epoch": 0.037311702828445585,
      "grad_norm": 0.3901897966861725,
      "learning_rate": 4.770520702957182e-05,
      "loss": 1.4645,
      "step": 1016
    },
    {
      "epoch": 0.03734842694540271,
      "grad_norm": 0.43339279294013977,
      "learning_rate": 4.752560645904962e-05,
      "loss": 1.3754,
      "step": 1017
    },
    {
      "epoch": 0.037385151062359845,
      "grad_norm": 0.40874043107032776,
      "learning_rate": 4.734623916689941e-05,
      "loss": 1.4339,
      "step": 1018
    },
    {
      "epoch": 0.03742187517931698,
      "grad_norm": 0.42034071683883667,
      "learning_rate": 4.716710595051022e-05,
      "loss": 1.3944,
      "step": 1019
    },
    {
      "epoch": 0.03745859929627411,
      "grad_norm": 0.4147108197212219,
      "learning_rate": 4.698820760623064e-05,
      "loss": 1.4243,
      "step": 1020
    },
    {
      "epoch": 0.03749532341323124,
      "grad_norm": 0.44932615756988525,
      "learning_rate": 4.6809544929365004e-05,
      "loss": 1.3545,
      "step": 1021
    },
    {
      "epoch": 0.03753204753018837,
      "grad_norm": 0.466249942779541,
      "learning_rate": 4.663111871417e-05,
      "loss": 1.5458,
      "step": 1022
    },
    {
      "epoch": 0.037568771647145505,
      "grad_norm": 0.5104914307594299,
      "learning_rate": 4.645292975385111e-05,
      "loss": 1.2742,
      "step": 1023
    },
    {
      "epoch": 0.03760549576410264,
      "grad_norm": 0.4866437017917633,
      "learning_rate": 4.627497884055912e-05,
      "loss": 1.3478,
      "step": 1024
    },
    {
      "epoch": 0.037642219881059764,
      "grad_norm": 0.6096803545951843,
      "learning_rate": 4.609726676538652e-05,
      "loss": 1.4148,
      "step": 1025
    },
    {
      "epoch": 0.0376789439980169,
      "grad_norm": 0.5660794377326965,
      "learning_rate": 4.591979431836402e-05,
      "loss": 1.3655,
      "step": 1026
    },
    {
      "epoch": 0.03771566811497403,
      "grad_norm": 0.5538148880004883,
      "learning_rate": 4.574256228845706e-05,
      "loss": 1.4539,
      "step": 1027
    },
    {
      "epoch": 0.03775239223193116,
      "grad_norm": 0.5961002111434937,
      "learning_rate": 4.5565571463562365e-05,
      "loss": 1.3115,
      "step": 1028
    },
    {
      "epoch": 0.03778911634888829,
      "grad_norm": 0.6135901212692261,
      "learning_rate": 4.5388822630504256e-05,
      "loss": 1.4322,
      "step": 1029
    },
    {
      "epoch": 0.037825840465845424,
      "grad_norm": 0.6604675650596619,
      "learning_rate": 4.521231657503132e-05,
      "loss": 1.3261,
      "step": 1030
    },
    {
      "epoch": 0.03786256458280256,
      "grad_norm": 0.630408525466919,
      "learning_rate": 4.503605408181286e-05,
      "loss": 1.2626,
      "step": 1031
    },
    {
      "epoch": 0.037899288699759684,
      "grad_norm": 0.7597583532333374,
      "learning_rate": 4.486003593443537e-05,
      "loss": 1.2651,
      "step": 1032
    },
    {
      "epoch": 0.03793601281671682,
      "grad_norm": 0.65804123878479,
      "learning_rate": 4.468426291539914e-05,
      "loss": 1.1503,
      "step": 1033
    },
    {
      "epoch": 0.03797273693367395,
      "grad_norm": 0.6689855456352234,
      "learning_rate": 4.4508735806114654e-05,
      "loss": 1.0524,
      "step": 1034
    },
    {
      "epoch": 0.038009461050631084,
      "grad_norm": 0.7008857131004333,
      "learning_rate": 4.433345538689929e-05,
      "loss": 1.0103,
      "step": 1035
    },
    {
      "epoch": 0.03804618516758821,
      "grad_norm": 0.8248462080955505,
      "learning_rate": 4.415842243697369e-05,
      "loss": 1.0913,
      "step": 1036
    },
    {
      "epoch": 0.038082909284545344,
      "grad_norm": 0.781531572341919,
      "learning_rate": 4.39836377344583e-05,
      "loss": 1.0677,
      "step": 1037
    },
    {
      "epoch": 0.03811963340150248,
      "grad_norm": 0.8204673528671265,
      "learning_rate": 4.380910205637012e-05,
      "loss": 1.0837,
      "step": 1038
    },
    {
      "epoch": 0.038156357518459603,
      "grad_norm": 1.0311377048492432,
      "learning_rate": 4.363481617861893e-05,
      "loss": 1.1286,
      "step": 1039
    },
    {
      "epoch": 0.03819308163541674,
      "grad_norm": 0.9294654130935669,
      "learning_rate": 4.346078087600412e-05,
      "loss": 1.0124,
      "step": 1040
    },
    {
      "epoch": 0.03822980575237387,
      "grad_norm": 0.9555936455726624,
      "learning_rate": 4.3286996922211034e-05,
      "loss": 1.142,
      "step": 1041
    },
    {
      "epoch": 0.038266529869331004,
      "grad_norm": 0.9653811454772949,
      "learning_rate": 4.311346508980772e-05,
      "loss": 1.1534,
      "step": 1042
    },
    {
      "epoch": 0.03830325398628813,
      "grad_norm": 0.9094187617301941,
      "learning_rate": 4.2940186150241365e-05,
      "loss": 1.2184,
      "step": 1043
    },
    {
      "epoch": 0.03833997810324526,
      "grad_norm": 0.996573269367218,
      "learning_rate": 4.27671608738349e-05,
      "loss": 1.1947,
      "step": 1044
    },
    {
      "epoch": 0.0383767022202024,
      "grad_norm": 0.8436797261238098,
      "learning_rate": 4.2594390029783534e-05,
      "loss": 0.6473,
      "step": 1045
    },
    {
      "epoch": 0.03841342633715953,
      "grad_norm": 1.1252999305725098,
      "learning_rate": 4.242187438615153e-05,
      "loss": 1.2264,
      "step": 1046
    },
    {
      "epoch": 0.038450150454116656,
      "grad_norm": 1.0413941144943237,
      "learning_rate": 4.224961470986849e-05,
      "loss": 1.0179,
      "step": 1047
    },
    {
      "epoch": 0.03848687457107379,
      "grad_norm": 1.1413769721984863,
      "learning_rate": 4.207761176672614e-05,
      "loss": 0.8282,
      "step": 1048
    },
    {
      "epoch": 0.03852359868803092,
      "grad_norm": 1.6900075674057007,
      "learning_rate": 4.190586632137491e-05,
      "loss": 1.1933,
      "step": 1049
    },
    {
      "epoch": 0.038560322804988056,
      "grad_norm": 1.814725399017334,
      "learning_rate": 4.173437913732048e-05,
      "loss": 1.3267,
      "step": 1050
    },
    {
      "epoch": 0.03859704692194518,
      "grad_norm": 0.17399807274341583,
      "learning_rate": 4.156315097692037e-05,
      "loss": 1.5736,
      "step": 1051
    },
    {
      "epoch": 0.038633771038902316,
      "grad_norm": 0.20164960622787476,
      "learning_rate": 4.139218260138074e-05,
      "loss": 1.7043,
      "step": 1052
    },
    {
      "epoch": 0.03867049515585945,
      "grad_norm": 0.20605356991291046,
      "learning_rate": 4.12214747707527e-05,
      "loss": 1.6609,
      "step": 1053
    },
    {
      "epoch": 0.038707219272816576,
      "grad_norm": 0.21172446012496948,
      "learning_rate": 4.1051028243929125e-05,
      "loss": 1.4481,
      "step": 1054
    },
    {
      "epoch": 0.03874394338977371,
      "grad_norm": 0.23490869998931885,
      "learning_rate": 4.088084377864135e-05,
      "loss": 1.6206,
      "step": 1055
    },
    {
      "epoch": 0.03878066750673084,
      "grad_norm": 0.2606179714202881,
      "learning_rate": 4.07109221314556e-05,
      "loss": 1.4051,
      "step": 1056
    },
    {
      "epoch": 0.038817391623687976,
      "grad_norm": 0.2648545205593109,
      "learning_rate": 4.054126405776971e-05,
      "loss": 1.3835,
      "step": 1057
    },
    {
      "epoch": 0.0388541157406451,
      "grad_norm": 0.2682105004787445,
      "learning_rate": 4.037187031180985e-05,
      "loss": 1.5435,
      "step": 1058
    },
    {
      "epoch": 0.038890839857602236,
      "grad_norm": 0.2959931790828705,
      "learning_rate": 4.020274164662707e-05,
      "loss": 1.4154,
      "step": 1059
    },
    {
      "epoch": 0.03892756397455937,
      "grad_norm": 0.28602543473243713,
      "learning_rate": 4.003387881409397e-05,
      "loss": 1.3952,
      "step": 1060
    },
    {
      "epoch": 0.0389642880915165,
      "grad_norm": 0.32159423828125,
      "learning_rate": 3.986528256490141e-05,
      "loss": 1.4261,
      "step": 1061
    },
    {
      "epoch": 0.03900101220847363,
      "grad_norm": 0.31909215450286865,
      "learning_rate": 3.969695364855511e-05,
      "loss": 1.6201,
      "step": 1062
    },
    {
      "epoch": 0.03903773632543076,
      "grad_norm": 0.3381721079349518,
      "learning_rate": 3.952889281337235e-05,
      "loss": 1.4999,
      "step": 1063
    },
    {
      "epoch": 0.039074460442387896,
      "grad_norm": 0.34948664903640747,
      "learning_rate": 3.93611008064786e-05,
      "loss": 1.306,
      "step": 1064
    },
    {
      "epoch": 0.03911118455934502,
      "grad_norm": 0.34947502613067627,
      "learning_rate": 3.9193578373804364e-05,
      "loss": 1.3543,
      "step": 1065
    },
    {
      "epoch": 0.039147908676302155,
      "grad_norm": 0.3818924129009247,
      "learning_rate": 3.90263262600816e-05,
      "loss": 1.419,
      "step": 1066
    },
    {
      "epoch": 0.03918463279325929,
      "grad_norm": 0.4030739665031433,
      "learning_rate": 3.88593452088406e-05,
      "loss": 1.3393,
      "step": 1067
    },
    {
      "epoch": 0.03922135691021642,
      "grad_norm": 0.42556071281433105,
      "learning_rate": 3.869263596240661e-05,
      "loss": 1.5446,
      "step": 1068
    },
    {
      "epoch": 0.03925808102717355,
      "grad_norm": 0.45733964443206787,
      "learning_rate": 3.8526199261896544e-05,
      "loss": 1.5824,
      "step": 1069
    },
    {
      "epoch": 0.03929480514413068,
      "grad_norm": 0.4844672381877899,
      "learning_rate": 3.836003584721577e-05,
      "loss": 1.6315,
      "step": 1070
    },
    {
      "epoch": 0.039331529261087815,
      "grad_norm": 0.4872642159461975,
      "learning_rate": 3.8194146457054655e-05,
      "loss": 1.4024,
      "step": 1071
    },
    {
      "epoch": 0.03936825337804495,
      "grad_norm": 0.5006334185600281,
      "learning_rate": 3.802853182888543e-05,
      "loss": 1.3321,
      "step": 1072
    },
    {
      "epoch": 0.039404977495002075,
      "grad_norm": 0.5102384686470032,
      "learning_rate": 3.786319269895877e-05,
      "loss": 1.4152,
      "step": 1073
    },
    {
      "epoch": 0.03944170161195921,
      "grad_norm": 0.553783655166626,
      "learning_rate": 3.769812980230074e-05,
      "loss": 1.4391,
      "step": 1074
    },
    {
      "epoch": 0.03947842572891634,
      "grad_norm": 0.546916663646698,
      "learning_rate": 3.7533343872709294e-05,
      "loss": 1.2688,
      "step": 1075
    },
    {
      "epoch": 0.039515149845873475,
      "grad_norm": 0.5837032198905945,
      "learning_rate": 3.736883564275112e-05,
      "loss": 1.4011,
      "step": 1076
    },
    {
      "epoch": 0.0395518739628306,
      "grad_norm": 0.6270307302474976,
      "learning_rate": 3.7204605843758386e-05,
      "loss": 1.5403,
      "step": 1077
    },
    {
      "epoch": 0.039588598079787735,
      "grad_norm": 0.5871410369873047,
      "learning_rate": 3.704065520582549e-05,
      "loss": 1.3669,
      "step": 1078
    },
    {
      "epoch": 0.03962532219674487,
      "grad_norm": 0.6564159989356995,
      "learning_rate": 3.6876984457805786e-05,
      "loss": 1.1926,
      "step": 1079
    },
    {
      "epoch": 0.039662046313701994,
      "grad_norm": 0.7062531113624573,
      "learning_rate": 3.671359432730834e-05,
      "loss": 1.2361,
      "step": 1080
    },
    {
      "epoch": 0.03969877043065913,
      "grad_norm": 0.7568163871765137,
      "learning_rate": 3.655048554069478e-05,
      "loss": 1.288,
      "step": 1081
    },
    {
      "epoch": 0.03973549454761626,
      "grad_norm": 0.8362852931022644,
      "learning_rate": 3.638765882307589e-05,
      "loss": 1.1232,
      "step": 1082
    },
    {
      "epoch": 0.039772218664573394,
      "grad_norm": 0.8132412433624268,
      "learning_rate": 3.6225114898308634e-05,
      "loss": 1.0621,
      "step": 1083
    },
    {
      "epoch": 0.03980894278153052,
      "grad_norm": 0.886319637298584,
      "learning_rate": 3.6062854488992714e-05,
      "loss": 1.2307,
      "step": 1084
    },
    {
      "epoch": 0.039845666898487654,
      "grad_norm": 0.7949456572532654,
      "learning_rate": 3.5900878316467454e-05,
      "loss": 1.2483,
      "step": 1085
    },
    {
      "epoch": 0.03988239101544479,
      "grad_norm": 0.6839745044708252,
      "learning_rate": 3.573918710080857e-05,
      "loss": 0.7561,
      "step": 1086
    },
    {
      "epoch": 0.03991911513240192,
      "grad_norm": 0.7769864201545715,
      "learning_rate": 3.5577781560825066e-05,
      "loss": 1.2021,
      "step": 1087
    },
    {
      "epoch": 0.03995583924935905,
      "grad_norm": 0.8365476727485657,
      "learning_rate": 3.541666241405588e-05,
      "loss": 1.1692,
      "step": 1088
    },
    {
      "epoch": 0.03999256336631618,
      "grad_norm": 0.8464269638061523,
      "learning_rate": 3.5255830376766764e-05,
      "loss": 0.907,
      "step": 1089
    },
    {
      "epoch": 0.040029287483273314,
      "grad_norm": 0.8388364911079407,
      "learning_rate": 3.509528616394716e-05,
      "loss": 1.3167,
      "step": 1090
    },
    {
      "epoch": 0.04006601160023045,
      "grad_norm": 0.9036702513694763,
      "learning_rate": 3.4935030489306883e-05,
      "loss": 0.9787,
      "step": 1091
    },
    {
      "epoch": 0.040102735717187574,
      "grad_norm": 0.9291563034057617,
      "learning_rate": 3.4775064065273165e-05,
      "loss": 1.181,
      "step": 1092
    },
    {
      "epoch": 0.04013945983414471,
      "grad_norm": 0.8095523118972778,
      "learning_rate": 3.4615387602987236e-05,
      "loss": 0.942,
      "step": 1093
    },
    {
      "epoch": 0.04017618395110184,
      "grad_norm": 1.0449271202087402,
      "learning_rate": 3.445600181230134e-05,
      "loss": 0.9857,
      "step": 1094
    },
    {
      "epoch": 0.04021290806805897,
      "grad_norm": 0.9336936473846436,
      "learning_rate": 3.429690740177549e-05,
      "loss": 0.5451,
      "step": 1095
    },
    {
      "epoch": 0.0402496321850161,
      "grad_norm": 1.27911376953125,
      "learning_rate": 3.413810507867436e-05,
      "loss": 1.1538,
      "step": 1096
    },
    {
      "epoch": 0.040286356301973233,
      "grad_norm": 1.1625890731811523,
      "learning_rate": 3.397959554896415e-05,
      "loss": 0.9012,
      "step": 1097
    },
    {
      "epoch": 0.04032308041893037,
      "grad_norm": 1.5193313360214233,
      "learning_rate": 3.3821379517309405e-05,
      "loss": 1.3543,
      "step": 1098
    },
    {
      "epoch": 0.04035980453588749,
      "grad_norm": 1.3823412656784058,
      "learning_rate": 3.3663457687069924e-05,
      "loss": 0.938,
      "step": 1099
    },
    {
      "epoch": 0.04039652865284463,
      "grad_norm": 1.5481663942337036,
      "learning_rate": 3.350583076029754e-05,
      "loss": 1.33,
      "step": 1100
    },
    {
      "epoch": 0.04043325276980176,
      "grad_norm": 0.16551245748996735,
      "learning_rate": 3.334849943773323e-05,
      "loss": 1.4462,
      "step": 1101
    },
    {
      "epoch": 0.04046997688675889,
      "grad_norm": 0.18680691719055176,
      "learning_rate": 3.319146441880371e-05,
      "loss": 1.7897,
      "step": 1102
    },
    {
      "epoch": 0.04050670100371602,
      "grad_norm": 0.20133081078529358,
      "learning_rate": 3.3034726401618444e-05,
      "loss": 1.4895,
      "step": 1103
    },
    {
      "epoch": 0.04054342512067315,
      "grad_norm": 0.22049792110919952,
      "learning_rate": 3.28782860829667e-05,
      "loss": 1.6375,
      "step": 1104
    },
    {
      "epoch": 0.040580149237630286,
      "grad_norm": 0.22869889438152313,
      "learning_rate": 3.272214415831418e-05,
      "loss": 1.629,
      "step": 1105
    },
    {
      "epoch": 0.04061687335458741,
      "grad_norm": 0.23674330115318298,
      "learning_rate": 3.2566301321800085e-05,
      "loss": 1.3961,
      "step": 1106
    },
    {
      "epoch": 0.040653597471544546,
      "grad_norm": 0.2582617402076721,
      "learning_rate": 3.241075826623401e-05,
      "loss": 1.4909,
      "step": 1107
    },
    {
      "epoch": 0.04069032158850168,
      "grad_norm": 0.24834193289279938,
      "learning_rate": 3.225551568309284e-05,
      "loss": 1.4574,
      "step": 1108
    },
    {
      "epoch": 0.04072704570545881,
      "grad_norm": 0.2793501019477844,
      "learning_rate": 3.210057426251773e-05,
      "loss": 1.2859,
      "step": 1109
    },
    {
      "epoch": 0.04076376982241594,
      "grad_norm": 0.29668989777565,
      "learning_rate": 3.1945934693310896e-05,
      "loss": 1.413,
      "step": 1110
    },
    {
      "epoch": 0.04080049393937307,
      "grad_norm": 0.3079325556755066,
      "learning_rate": 3.179159766293282e-05,
      "loss": 1.4458,
      "step": 1111
    },
    {
      "epoch": 0.040837218056330206,
      "grad_norm": 0.2956605553627014,
      "learning_rate": 3.163756385749889e-05,
      "loss": 1.3668,
      "step": 1112
    },
    {
      "epoch": 0.04087394217328734,
      "grad_norm": 0.3393270671367645,
      "learning_rate": 3.148383396177653e-05,
      "loss": 1.4577,
      "step": 1113
    },
    {
      "epoch": 0.040910666290244466,
      "grad_norm": 0.3519705533981323,
      "learning_rate": 3.133040865918213e-05,
      "loss": 1.2252,
      "step": 1114
    },
    {
      "epoch": 0.0409473904072016,
      "grad_norm": 0.32640624046325684,
      "learning_rate": 3.117728863177796e-05,
      "loss": 1.4825,
      "step": 1115
    },
    {
      "epoch": 0.04098411452415873,
      "grad_norm": 0.3808278441429138,
      "learning_rate": 3.102447456026919e-05,
      "loss": 1.5264,
      "step": 1116
    },
    {
      "epoch": 0.041020838641115866,
      "grad_norm": 0.35617879033088684,
      "learning_rate": 3.0871967124000834e-05,
      "loss": 1.1416,
      "step": 1117
    },
    {
      "epoch": 0.04105756275807299,
      "grad_norm": 0.3930893838405609,
      "learning_rate": 3.0719767000954714e-05,
      "loss": 1.3325,
      "step": 1118
    },
    {
      "epoch": 0.041094286875030125,
      "grad_norm": 0.4028378129005432,
      "learning_rate": 3.056787486774656e-05,
      "loss": 1.3808,
      "step": 1119
    },
    {
      "epoch": 0.04113101099198726,
      "grad_norm": 0.4413599669933319,
      "learning_rate": 3.041629139962283e-05,
      "loss": 1.3991,
      "step": 1120
    },
    {
      "epoch": 0.041167735108944385,
      "grad_norm": 0.44256123900413513,
      "learning_rate": 3.0265017270457775e-05,
      "loss": 1.3232,
      "step": 1121
    },
    {
      "epoch": 0.04120445922590152,
      "grad_norm": 0.48134544491767883,
      "learning_rate": 3.0114053152750556e-05,
      "loss": 1.4457,
      "step": 1122
    },
    {
      "epoch": 0.04124118334285865,
      "grad_norm": 0.474954754114151,
      "learning_rate": 2.9963399717622077e-05,
      "loss": 1.5721,
      "step": 1123
    },
    {
      "epoch": 0.041277907459815785,
      "grad_norm": 0.533463180065155,
      "learning_rate": 2.98130576348121e-05,
      "loss": 1.3214,
      "step": 1124
    },
    {
      "epoch": 0.04131463157677291,
      "grad_norm": 0.5395529270172119,
      "learning_rate": 2.966302757267625e-05,
      "loss": 1.0051,
      "step": 1125
    },
    {
      "epoch": 0.04131463157677291,
      "eval_loss": 1.262094497680664,
      "eval_runtime": 1108.7807,
      "eval_samples_per_second": 41.363,
      "eval_steps_per_second": 20.681,
      "step": 1125
    },
    {
      "epoch": 0.041351355693730045,
      "grad_norm": 0.5946307182312012,
      "learning_rate": 2.9513310198183065e-05,
      "loss": 1.3899,
      "step": 1126
    },
    {
      "epoch": 0.04138807981068718,
      "grad_norm": 0.5627098679542542,
      "learning_rate": 2.936390617691097e-05,
      "loss": 1.3988,
      "step": 1127
    },
    {
      "epoch": 0.04142480392764431,
      "grad_norm": 0.6260983943939209,
      "learning_rate": 2.9214816173045356e-05,
      "loss": 1.2403,
      "step": 1128
    },
    {
      "epoch": 0.04146152804460144,
      "grad_norm": 0.6079756021499634,
      "learning_rate": 2.906604084937572e-05,
      "loss": 1.4305,
      "step": 1129
    },
    {
      "epoch": 0.04149825216155857,
      "grad_norm": 0.9089528322219849,
      "learning_rate": 2.8917580867292526e-05,
      "loss": 1.2688,
      "step": 1130
    },
    {
      "epoch": 0.041534976278515705,
      "grad_norm": 0.6574273109436035,
      "learning_rate": 2.8769436886784408e-05,
      "loss": 1.2145,
      "step": 1131
    },
    {
      "epoch": 0.04157170039547283,
      "grad_norm": 0.6633460521697998,
      "learning_rate": 2.862160956643517e-05,
      "loss": 1.2916,
      "step": 1132
    },
    {
      "epoch": 0.041608424512429965,
      "grad_norm": 0.8151691555976868,
      "learning_rate": 2.847409956342092e-05,
      "loss": 0.8015,
      "step": 1133
    },
    {
      "epoch": 0.0416451486293871,
      "grad_norm": 0.7017518281936646,
      "learning_rate": 2.8326907533507074e-05,
      "loss": 0.8071,
      "step": 1134
    },
    {
      "epoch": 0.04168187274634423,
      "grad_norm": 0.7146211862564087,
      "learning_rate": 2.8180034131045464e-05,
      "loss": 1.1742,
      "step": 1135
    },
    {
      "epoch": 0.04171859686330136,
      "grad_norm": 0.8099738955497742,
      "learning_rate": 2.8033480008971546e-05,
      "loss": 1.0008,
      "step": 1136
    },
    {
      "epoch": 0.04175532098025849,
      "grad_norm": 0.7735601663589478,
      "learning_rate": 2.7887245818801277e-05,
      "loss": 1.422,
      "step": 1137
    },
    {
      "epoch": 0.041792045097215624,
      "grad_norm": 0.6984381675720215,
      "learning_rate": 2.7741332210628345e-05,
      "loss": 1.2289,
      "step": 1138
    },
    {
      "epoch": 0.04182876921417276,
      "grad_norm": 0.9226275682449341,
      "learning_rate": 2.759573983312138e-05,
      "loss": 1.3146,
      "step": 1139
    },
    {
      "epoch": 0.041865493331129884,
      "grad_norm": 0.7991454601287842,
      "learning_rate": 2.7450469333520855e-05,
      "loss": 1.0188,
      "step": 1140
    },
    {
      "epoch": 0.04190221744808702,
      "grad_norm": 0.9062628746032715,
      "learning_rate": 2.730552135763632e-05,
      "loss": 0.841,
      "step": 1141
    },
    {
      "epoch": 0.04193894156504415,
      "grad_norm": 0.9036419987678528,
      "learning_rate": 2.7160896549843562e-05,
      "loss": 1.0628,
      "step": 1142
    },
    {
      "epoch": 0.041975665682001284,
      "grad_norm": 0.9827883243560791,
      "learning_rate": 2.701659555308169e-05,
      "loss": 1.094,
      "step": 1143
    },
    {
      "epoch": 0.04201238979895841,
      "grad_norm": 0.8482446074485779,
      "learning_rate": 2.6872619008850274e-05,
      "loss": 0.8267,
      "step": 1144
    },
    {
      "epoch": 0.042049113915915544,
      "grad_norm": 0.8491653800010681,
      "learning_rate": 2.672896755720654e-05,
      "loss": 0.905,
      "step": 1145
    },
    {
      "epoch": 0.04208583803287268,
      "grad_norm": 1.0981497764587402,
      "learning_rate": 2.6585641836762433e-05,
      "loss": 0.9423,
      "step": 1146
    },
    {
      "epoch": 0.042122562149829804,
      "grad_norm": 1.1127378940582275,
      "learning_rate": 2.6442642484681944e-05,
      "loss": 0.9595,
      "step": 1147
    },
    {
      "epoch": 0.04215928626678694,
      "grad_norm": 1.6657700538635254,
      "learning_rate": 2.6299970136678077e-05,
      "loss": 1.1075,
      "step": 1148
    },
    {
      "epoch": 0.04219601038374407,
      "grad_norm": 1.9846173524856567,
      "learning_rate": 2.6157625427010156e-05,
      "loss": 1.0853,
      "step": 1149
    },
    {
      "epoch": 0.042232734500701204,
      "grad_norm": 1.7222540378570557,
      "learning_rate": 2.6015608988480955e-05,
      "loss": 1.5832,
      "step": 1150
    },
    {
      "epoch": 0.04226945861765833,
      "grad_norm": 0.15658363699913025,
      "learning_rate": 2.5873921452433915e-05,
      "loss": 1.5645,
      "step": 1151
    },
    {
      "epoch": 0.04230618273461546,
      "grad_norm": 0.18964284658432007,
      "learning_rate": 2.57325634487503e-05,
      "loss": 1.6559,
      "step": 1152
    },
    {
      "epoch": 0.0423429068515726,
      "grad_norm": 0.21651172637939453,
      "learning_rate": 2.5591535605846383e-05,
      "loss": 1.6518,
      "step": 1153
    },
    {
      "epoch": 0.04237963096852973,
      "grad_norm": 0.20955222845077515,
      "learning_rate": 2.5450838550670808e-05,
      "loss": 1.6138,
      "step": 1154
    },
    {
      "epoch": 0.042416355085486857,
      "grad_norm": 0.2332032173871994,
      "learning_rate": 2.5310472908701555e-05,
      "loss": 1.3939,
      "step": 1155
    },
    {
      "epoch": 0.04245307920244399,
      "grad_norm": 0.23824694752693176,
      "learning_rate": 2.5170439303943294e-05,
      "loss": 1.5844,
      "step": 1156
    },
    {
      "epoch": 0.04248980331940112,
      "grad_norm": 0.2618229389190674,
      "learning_rate": 2.503073835892471e-05,
      "loss": 1.3773,
      "step": 1157
    },
    {
      "epoch": 0.04252652743635825,
      "grad_norm": 0.268458753824234,
      "learning_rate": 2.4891370694695517e-05,
      "loss": 1.6335,
      "step": 1158
    },
    {
      "epoch": 0.04256325155331538,
      "grad_norm": 0.2882594168186188,
      "learning_rate": 2.4752336930823837e-05,
      "loss": 1.5284,
      "step": 1159
    },
    {
      "epoch": 0.042599975670272516,
      "grad_norm": 0.2790129482746124,
      "learning_rate": 2.4613637685393432e-05,
      "loss": 1.5137,
      "step": 1160
    },
    {
      "epoch": 0.04263669978722965,
      "grad_norm": 0.29624471068382263,
      "learning_rate": 2.4475273575000936e-05,
      "loss": 1.4307,
      "step": 1161
    },
    {
      "epoch": 0.042673423904186776,
      "grad_norm": 0.3192678391933441,
      "learning_rate": 2.4337245214753103e-05,
      "loss": 1.43,
      "step": 1162
    },
    {
      "epoch": 0.04271014802114391,
      "grad_norm": 0.31192949414253235,
      "learning_rate": 2.4199553218264093e-05,
      "loss": 1.225,
      "step": 1163
    },
    {
      "epoch": 0.04274687213810104,
      "grad_norm": 0.3523472845554352,
      "learning_rate": 2.4062198197652752e-05,
      "loss": 1.4192,
      "step": 1164
    },
    {
      "epoch": 0.042783596255058176,
      "grad_norm": 0.3479517102241516,
      "learning_rate": 2.3925180763539844e-05,
      "loss": 1.5284,
      "step": 1165
    },
    {
      "epoch": 0.0428203203720153,
      "grad_norm": 0.3700026869773865,
      "learning_rate": 2.3788501525045438e-05,
      "loss": 1.3087,
      "step": 1166
    },
    {
      "epoch": 0.042857044488972436,
      "grad_norm": 0.3925465941429138,
      "learning_rate": 2.3652161089786086e-05,
      "loss": 1.5137,
      "step": 1167
    },
    {
      "epoch": 0.04289376860592957,
      "grad_norm": 0.3899960219860077,
      "learning_rate": 2.351616006387214e-05,
      "loss": 1.3156,
      "step": 1168
    },
    {
      "epoch": 0.0429304927228867,
      "grad_norm": 0.37500688433647156,
      "learning_rate": 2.3380499051905137e-05,
      "loss": 1.3841,
      "step": 1169
    },
    {
      "epoch": 0.04296721683984383,
      "grad_norm": 0.4503501355648041,
      "learning_rate": 2.324517865697501e-05,
      "loss": 1.3522,
      "step": 1170
    },
    {
      "epoch": 0.04300394095680096,
      "grad_norm": 0.437712699174881,
      "learning_rate": 2.3110199480657525e-05,
      "loss": 1.396,
      "step": 1171
    },
    {
      "epoch": 0.043040665073758096,
      "grad_norm": 0.4187954366207123,
      "learning_rate": 2.2975562123011495e-05,
      "loss": 1.1387,
      "step": 1172
    },
    {
      "epoch": 0.04307738919071522,
      "grad_norm": 0.47373610734939575,
      "learning_rate": 2.2841267182576143e-05,
      "loss": 1.2818,
      "step": 1173
    },
    {
      "epoch": 0.043114113307672355,
      "grad_norm": 0.5407326817512512,
      "learning_rate": 2.2707315256368433e-05,
      "loss": 1.4052,
      "step": 1174
    },
    {
      "epoch": 0.04315083742462949,
      "grad_norm": 0.5349066257476807,
      "learning_rate": 2.2573706939880555e-05,
      "loss": 1.2237,
      "step": 1175
    },
    {
      "epoch": 0.04318756154158662,
      "grad_norm": 0.521129310131073,
      "learning_rate": 2.2440442827077045e-05,
      "loss": 1.5984,
      "step": 1176
    },
    {
      "epoch": 0.04322428565854375,
      "grad_norm": 0.6487154960632324,
      "learning_rate": 2.230752351039228e-05,
      "loss": 1.3261,
      "step": 1177
    },
    {
      "epoch": 0.04326100977550088,
      "grad_norm": 0.5437444448471069,
      "learning_rate": 2.2174949580727832e-05,
      "loss": 1.3541,
      "step": 1178
    },
    {
      "epoch": 0.043297733892458015,
      "grad_norm": 0.6246458292007446,
      "learning_rate": 2.2042721627449846e-05,
      "loss": 1.293,
      "step": 1179
    },
    {
      "epoch": 0.04333445800941515,
      "grad_norm": 0.5926059484481812,
      "learning_rate": 2.1910840238386398e-05,
      "loss": 1.1301,
      "step": 1180
    },
    {
      "epoch": 0.043371182126372275,
      "grad_norm": 0.7066393494606018,
      "learning_rate": 2.1779305999824884e-05,
      "loss": 1.4093,
      "step": 1181
    },
    {
      "epoch": 0.04340790624332941,
      "grad_norm": 0.6763113141059875,
      "learning_rate": 2.164811949650942e-05,
      "loss": 1.1033,
      "step": 1182
    },
    {
      "epoch": 0.04344463036028654,
      "grad_norm": 0.6880610585212708,
      "learning_rate": 2.1517281311638217e-05,
      "loss": 1.3459,
      "step": 1183
    },
    {
      "epoch": 0.04348135447724367,
      "grad_norm": 0.7546311020851135,
      "learning_rate": 2.1386792026861103e-05,
      "loss": 0.938,
      "step": 1184
    },
    {
      "epoch": 0.0435180785942008,
      "grad_norm": 0.7434496283531189,
      "learning_rate": 2.125665222227675e-05,
      "loss": 1.1434,
      "step": 1185
    },
    {
      "epoch": 0.043554802711157935,
      "grad_norm": 0.801835834980011,
      "learning_rate": 2.112686247643024e-05,
      "loss": 1.4937,
      "step": 1186
    },
    {
      "epoch": 0.04359152682811507,
      "grad_norm": 0.8252962231636047,
      "learning_rate": 2.09974233663104e-05,
      "loss": 1.1632,
      "step": 1187
    },
    {
      "epoch": 0.043628250945072194,
      "grad_norm": 0.8671730160713196,
      "learning_rate": 2.0868335467347366e-05,
      "loss": 1.0688,
      "step": 1188
    },
    {
      "epoch": 0.04366497506202933,
      "grad_norm": 0.8788164854049683,
      "learning_rate": 2.073959935340988e-05,
      "loss": 1.2219,
      "step": 1189
    },
    {
      "epoch": 0.04370169917898646,
      "grad_norm": 0.8732954263687134,
      "learning_rate": 2.06112155968028e-05,
      "loss": 1.3156,
      "step": 1190
    },
    {
      "epoch": 0.043738423295943595,
      "grad_norm": 0.9020258784294128,
      "learning_rate": 2.0483184768264596e-05,
      "loss": 1.0873,
      "step": 1191
    },
    {
      "epoch": 0.04377514741290072,
      "grad_norm": 0.9553011059761047,
      "learning_rate": 2.035550743696468e-05,
      "loss": 1.2354,
      "step": 1192
    },
    {
      "epoch": 0.043811871529857854,
      "grad_norm": 0.876915454864502,
      "learning_rate": 2.022818417050113e-05,
      "loss": 0.7298,
      "step": 1193
    },
    {
      "epoch": 0.04384859564681499,
      "grad_norm": 0.8597220778465271,
      "learning_rate": 2.0101215534897855e-05,
      "loss": 0.9797,
      "step": 1194
    },
    {
      "epoch": 0.04388531976377212,
      "grad_norm": 1.0591257810592651,
      "learning_rate": 1.99746020946023e-05,
      "loss": 1.308,
      "step": 1195
    },
    {
      "epoch": 0.04392204388072925,
      "grad_norm": 1.11239492893219,
      "learning_rate": 1.9848344412482854e-05,
      "loss": 1.0363,
      "step": 1196
    },
    {
      "epoch": 0.04395876799768638,
      "grad_norm": 1.0520296096801758,
      "learning_rate": 1.9722443049826344e-05,
      "loss": 1.2541,
      "step": 1197
    },
    {
      "epoch": 0.043995492114643514,
      "grad_norm": 1.212968111038208,
      "learning_rate": 1.9596898566335576e-05,
      "loss": 0.9089,
      "step": 1198
    },
    {
      "epoch": 0.04403221623160064,
      "grad_norm": 1.3280837535858154,
      "learning_rate": 1.9471711520126824e-05,
      "loss": 1.1311,
      "step": 1199
    },
    {
      "epoch": 0.044068940348557774,
      "grad_norm": 1.8388601541519165,
      "learning_rate": 1.9346882467727325e-05,
      "loss": 1.1585,
      "step": 1200
    },
    {
      "epoch": 0.04410566446551491,
      "grad_norm": 0.15597866475582123,
      "learning_rate": 1.9222411964072884e-05,
      "loss": 1.6227,
      "step": 1201
    },
    {
      "epoch": 0.04414238858247204,
      "grad_norm": 0.18923719227313995,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 1.7345,
      "step": 1202
    },
    {
      "epoch": 0.04417911269942917,
      "grad_norm": 0.19939024746418,
      "learning_rate": 1.8974548814769944e-05,
      "loss": 1.7024,
      "step": 1203
    },
    {
      "epoch": 0.0442158368163863,
      "grad_norm": 0.19955849647521973,
      "learning_rate": 1.8851157271013442e-05,
      "loss": 1.6801,
      "step": 1204
    },
    {
      "epoch": 0.044252560933343434,
      "grad_norm": 0.21925851702690125,
      "learning_rate": 1.872812647978095e-05,
      "loss": 1.4425,
      "step": 1205
    },
    {
      "epoch": 0.04428928505030057,
      "grad_norm": 0.2462303340435028,
      "learning_rate": 1.8605456988014015e-05,
      "loss": 1.4863,
      "step": 1206
    },
    {
      "epoch": 0.04432600916725769,
      "grad_norm": 0.26405367255210876,
      "learning_rate": 1.8483149341047923e-05,
      "loss": 1.6516,
      "step": 1207
    },
    {
      "epoch": 0.04436273328421483,
      "grad_norm": 0.2757001221179962,
      "learning_rate": 1.8361204082609352e-05,
      "loss": 1.5282,
      "step": 1208
    },
    {
      "epoch": 0.04439945740117196,
      "grad_norm": 0.3055432438850403,
      "learning_rate": 1.8239621754813995e-05,
      "loss": 1.5236,
      "step": 1209
    },
    {
      "epoch": 0.044436181518129086,
      "grad_norm": 0.31371381878852844,
      "learning_rate": 1.811840289816409e-05,
      "loss": 1.5558,
      "step": 1210
    },
    {
      "epoch": 0.04447290563508622,
      "grad_norm": 0.2873939573764801,
      "learning_rate": 1.799754805154603e-05,
      "loss": 1.6137,
      "step": 1211
    },
    {
      "epoch": 0.04450962975204335,
      "grad_norm": 0.30505770444869995,
      "learning_rate": 1.787705775222802e-05,
      "loss": 1.3023,
      "step": 1212
    },
    {
      "epoch": 0.044546353869000486,
      "grad_norm": 0.3023243248462677,
      "learning_rate": 1.775693253585763e-05,
      "loss": 1.5203,
      "step": 1213
    },
    {
      "epoch": 0.04458307798595761,
      "grad_norm": 0.3116466701030731,
      "learning_rate": 1.763717293645939e-05,
      "loss": 1.3516,
      "step": 1214
    },
    {
      "epoch": 0.044619802102914746,
      "grad_norm": 0.35461676120758057,
      "learning_rate": 1.7517779486432495e-05,
      "loss": 1.5348,
      "step": 1215
    },
    {
      "epoch": 0.04465652621987188,
      "grad_norm": 0.3483802378177643,
      "learning_rate": 1.7398752716548395e-05,
      "loss": 1.4687,
      "step": 1216
    },
    {
      "epoch": 0.04469325033682901,
      "grad_norm": 0.3820308446884155,
      "learning_rate": 1.728009315594843e-05,
      "loss": 1.4626,
      "step": 1217
    },
    {
      "epoch": 0.04472997445378614,
      "grad_norm": 0.40812668204307556,
      "learning_rate": 1.716180133214149e-05,
      "loss": 1.3626,
      "step": 1218
    },
    {
      "epoch": 0.04476669857074327,
      "grad_norm": 0.40593403577804565,
      "learning_rate": 1.704387777100165e-05,
      "loss": 1.5823,
      "step": 1219
    },
    {
      "epoch": 0.044803422687700406,
      "grad_norm": 0.424576997756958,
      "learning_rate": 1.6926322996765897e-05,
      "loss": 1.4499,
      "step": 1220
    },
    {
      "epoch": 0.04484014680465754,
      "grad_norm": 0.4259289801120758,
      "learning_rate": 1.6809137532031704e-05,
      "loss": 1.3686,
      "step": 1221
    },
    {
      "epoch": 0.044876870921614666,
      "grad_norm": 0.4739649295806885,
      "learning_rate": 1.6692321897754758e-05,
      "loss": 1.443,
      "step": 1222
    },
    {
      "epoch": 0.0449135950385718,
      "grad_norm": 0.46144747734069824,
      "learning_rate": 1.65758766132467e-05,
      "loss": 1.3949,
      "step": 1223
    },
    {
      "epoch": 0.04495031915552893,
      "grad_norm": 0.5201800465583801,
      "learning_rate": 1.6459802196172668e-05,
      "loss": 1.3713,
      "step": 1224
    },
    {
      "epoch": 0.04498704327248606,
      "grad_norm": 0.4891027510166168,
      "learning_rate": 1.634409916254914e-05,
      "loss": 1.3956,
      "step": 1225
    },
    {
      "epoch": 0.04502376738944319,
      "grad_norm": 0.5617945790290833,
      "learning_rate": 1.622876802674158e-05,
      "loss": 1.2424,
      "step": 1226
    },
    {
      "epoch": 0.045060491506400326,
      "grad_norm": 0.5572589039802551,
      "learning_rate": 1.6113809301462125e-05,
      "loss": 1.2522,
      "step": 1227
    },
    {
      "epoch": 0.04509721562335746,
      "grad_norm": 0.5615737438201904,
      "learning_rate": 1.599922349776738e-05,
      "loss": 1.3497,
      "step": 1228
    },
    {
      "epoch": 0.045133939740314585,
      "grad_norm": 0.5860127210617065,
      "learning_rate": 1.5885011125056047e-05,
      "loss": 1.0807,
      "step": 1229
    },
    {
      "epoch": 0.04517066385727172,
      "grad_norm": 0.560443103313446,
      "learning_rate": 1.5771172691066794e-05,
      "loss": 1.3994,
      "step": 1230
    },
    {
      "epoch": 0.04520738797422885,
      "grad_norm": 0.6717036366462708,
      "learning_rate": 1.565770870187585e-05,
      "loss": 1.3859,
      "step": 1231
    },
    {
      "epoch": 0.045244112091185985,
      "grad_norm": 0.6881125569343567,
      "learning_rate": 1.5544619661894864e-05,
      "loss": 1.0095,
      "step": 1232
    },
    {
      "epoch": 0.04528083620814311,
      "grad_norm": 0.6939812898635864,
      "learning_rate": 1.543190607386861e-05,
      "loss": 1.1463,
      "step": 1233
    },
    {
      "epoch": 0.045317560325100245,
      "grad_norm": 0.6921367049217224,
      "learning_rate": 1.5319568438872745e-05,
      "loss": 0.9047,
      "step": 1234
    },
    {
      "epoch": 0.04535428444205738,
      "grad_norm": 0.799044668674469,
      "learning_rate": 1.520760725631164e-05,
      "loss": 1.2291,
      "step": 1235
    },
    {
      "epoch": 0.045391008559014505,
      "grad_norm": 0.7895994782447815,
      "learning_rate": 1.5096023023916094e-05,
      "loss": 1.0209,
      "step": 1236
    },
    {
      "epoch": 0.04542773267597164,
      "grad_norm": 0.8128568530082703,
      "learning_rate": 1.498481623774115e-05,
      "loss": 0.8527,
      "step": 1237
    },
    {
      "epoch": 0.04546445679292877,
      "grad_norm": 0.7516756057739258,
      "learning_rate": 1.4873987392163947e-05,
      "loss": 1.0517,
      "step": 1238
    },
    {
      "epoch": 0.045501180909885905,
      "grad_norm": 0.756646990776062,
      "learning_rate": 1.4763536979881354e-05,
      "loss": 1.1319,
      "step": 1239
    },
    {
      "epoch": 0.04553790502684303,
      "grad_norm": 0.9550017714500427,
      "learning_rate": 1.4653465491908003e-05,
      "loss": 1.3255,
      "step": 1240
    },
    {
      "epoch": 0.045574629143800165,
      "grad_norm": 0.9971140027046204,
      "learning_rate": 1.4543773417573925e-05,
      "loss": 0.8763,
      "step": 1241
    },
    {
      "epoch": 0.0456113532607573,
      "grad_norm": 0.9035000205039978,
      "learning_rate": 1.4434461244522458e-05,
      "loss": 0.9841,
      "step": 1242
    },
    {
      "epoch": 0.04564807737771443,
      "grad_norm": 1.002290964126587,
      "learning_rate": 1.4325529458708065e-05,
      "loss": 1.138,
      "step": 1243
    },
    {
      "epoch": 0.04568480149467156,
      "grad_norm": 1.0005804300308228,
      "learning_rate": 1.4216978544394177e-05,
      "loss": 0.96,
      "step": 1244
    },
    {
      "epoch": 0.04572152561162869,
      "grad_norm": 1.0253926515579224,
      "learning_rate": 1.4108808984151023e-05,
      "loss": 0.9183,
      "step": 1245
    },
    {
      "epoch": 0.045758249728585824,
      "grad_norm": 0.9568840265274048,
      "learning_rate": 1.4001021258853509e-05,
      "loss": 0.9683,
      "step": 1246
    },
    {
      "epoch": 0.04579497384554296,
      "grad_norm": 1.2253047227859497,
      "learning_rate": 1.3893615847679065e-05,
      "loss": 1.1265,
      "step": 1247
    },
    {
      "epoch": 0.045831697962500084,
      "grad_norm": 1.1104260683059692,
      "learning_rate": 1.3786593228105494e-05,
      "loss": 0.6034,
      "step": 1248
    },
    {
      "epoch": 0.04586842207945722,
      "grad_norm": 1.4997528791427612,
      "learning_rate": 1.3679953875908957e-05,
      "loss": 1.208,
      "step": 1249
    },
    {
      "epoch": 0.04590514619641435,
      "grad_norm": 1.8989046812057495,
      "learning_rate": 1.3573698265161683e-05,
      "loss": 1.4831,
      "step": 1250
    },
    {
      "epoch": 0.04594187031337148,
      "grad_norm": 0.14705991744995117,
      "learning_rate": 1.3467826868229994e-05,
      "loss": 1.2447,
      "step": 1251
    },
    {
      "epoch": 0.04597859443032861,
      "grad_norm": 0.18212033808231354,
      "learning_rate": 1.3362340155772146e-05,
      "loss": 1.7289,
      "step": 1252
    },
    {
      "epoch": 0.046015318547285744,
      "grad_norm": 0.2010517716407776,
      "learning_rate": 1.3257238596736266e-05,
      "loss": 1.6981,
      "step": 1253
    },
    {
      "epoch": 0.04605204266424288,
      "grad_norm": 0.2143598347902298,
      "learning_rate": 1.3152522658358245e-05,
      "loss": 1.5936,
      "step": 1254
    },
    {
      "epoch": 0.046088766781200004,
      "grad_norm": 0.21208621561527252,
      "learning_rate": 1.3048192806159721e-05,
      "loss": 1.3786,
      "step": 1255
    },
    {
      "epoch": 0.04612549089815714,
      "grad_norm": 0.2341011017560959,
      "learning_rate": 1.2944249503945894e-05,
      "loss": 1.4061,
      "step": 1256
    },
    {
      "epoch": 0.04616221501511427,
      "grad_norm": 0.24541302025318146,
      "learning_rate": 1.2840693213803545e-05,
      "loss": 1.6176,
      "step": 1257
    },
    {
      "epoch": 0.046198939132071404,
      "grad_norm": 0.26293060183525085,
      "learning_rate": 1.2737524396099032e-05,
      "loss": 1.6425,
      "step": 1258
    },
    {
      "epoch": 0.04623566324902853,
      "grad_norm": 0.26363781094551086,
      "learning_rate": 1.2634743509476088e-05,
      "loss": 1.4985,
      "step": 1259
    },
    {
      "epoch": 0.046272387365985664,
      "grad_norm": 0.2901013195514679,
      "learning_rate": 1.2532351010853916e-05,
      "loss": 1.4564,
      "step": 1260
    },
    {
      "epoch": 0.0463091114829428,
      "grad_norm": 0.27847039699554443,
      "learning_rate": 1.243034735542512e-05,
      "loss": 1.3042,
      "step": 1261
    },
    {
      "epoch": 0.04634583559989993,
      "grad_norm": 0.29209014773368835,
      "learning_rate": 1.2328732996653669e-05,
      "loss": 1.5635,
      "step": 1262
    },
    {
      "epoch": 0.04638255971685706,
      "grad_norm": 0.3432471752166748,
      "learning_rate": 1.2227508386272878e-05,
      "loss": 1.4215,
      "step": 1263
    },
    {
      "epoch": 0.04641928383381419,
      "grad_norm": 0.3231754004955292,
      "learning_rate": 1.212667397428342e-05,
      "loss": 1.5293,
      "step": 1264
    },
    {
      "epoch": 0.04645600795077132,
      "grad_norm": 0.3533654510974884,
      "learning_rate": 1.2026230208951306e-05,
      "loss": 1.3258,
      "step": 1265
    },
    {
      "epoch": 0.04649273206772845,
      "grad_norm": 0.379334956407547,
      "learning_rate": 1.1926177536805905e-05,
      "loss": 1.707,
      "step": 1266
    },
    {
      "epoch": 0.04652945618468558,
      "grad_norm": 0.38155263662338257,
      "learning_rate": 1.1826516402637989e-05,
      "loss": 1.1782,
      "step": 1267
    },
    {
      "epoch": 0.046566180301642716,
      "grad_norm": 0.47478413581848145,
      "learning_rate": 1.1727247249497685e-05,
      "loss": 1.2685,
      "step": 1268
    },
    {
      "epoch": 0.04660290441859985,
      "grad_norm": 0.39825162291526794,
      "learning_rate": 1.1628370518692533e-05,
      "loss": 1.4241,
      "step": 1269
    },
    {
      "epoch": 0.046639628535556976,
      "grad_norm": 0.4363865852355957,
      "learning_rate": 1.152988664978556e-05,
      "loss": 1.2339,
      "step": 1270
    },
    {
      "epoch": 0.04667635265251411,
      "grad_norm": 0.47400519251823425,
      "learning_rate": 1.1431796080593283e-05,
      "loss": 1.4923,
      "step": 1271
    },
    {
      "epoch": 0.04671307676947124,
      "grad_norm": 0.49352601170539856,
      "learning_rate": 1.1334099247183783e-05,
      "loss": 1.5727,
      "step": 1272
    },
    {
      "epoch": 0.046749800886428376,
      "grad_norm": 0.47106578946113586,
      "learning_rate": 1.1236796583874787e-05,
      "loss": 1.1573,
      "step": 1273
    },
    {
      "epoch": 0.0467865250033855,
      "grad_norm": 0.5897020101547241,
      "learning_rate": 1.1139888523231678e-05,
      "loss": 1.5761,
      "step": 1274
    },
    {
      "epoch": 0.046823249120342636,
      "grad_norm": 0.569318950176239,
      "learning_rate": 1.1043375496065611e-05,
      "loss": 1.4013,
      "step": 1275
    },
    {
      "epoch": 0.04685997323729977,
      "grad_norm": 0.5247970819473267,
      "learning_rate": 1.0947257931431642e-05,
      "loss": 1.469,
      "step": 1276
    },
    {
      "epoch": 0.046896697354256896,
      "grad_norm": 0.5704692602157593,
      "learning_rate": 1.0851536256626705e-05,
      "loss": 1.1892,
      "step": 1277
    },
    {
      "epoch": 0.04693342147121403,
      "grad_norm": 0.6168922185897827,
      "learning_rate": 1.0756210897187812e-05,
      "loss": 1.3421,
      "step": 1278
    },
    {
      "epoch": 0.04697014558817116,
      "grad_norm": 0.5921631455421448,
      "learning_rate": 1.0661282276890127e-05,
      "loss": 1.177,
      "step": 1279
    },
    {
      "epoch": 0.047006869705128296,
      "grad_norm": 0.6388888955116272,
      "learning_rate": 1.0566750817745074e-05,
      "loss": 1.2236,
      "step": 1280
    },
    {
      "epoch": 0.04704359382208542,
      "grad_norm": 0.6330857276916504,
      "learning_rate": 1.0472616939998492e-05,
      "loss": 1.2269,
      "step": 1281
    },
    {
      "epoch": 0.047080317939042556,
      "grad_norm": 0.638800323009491,
      "learning_rate": 1.0378881062128731e-05,
      "loss": 1.0489,
      "step": 1282
    },
    {
      "epoch": 0.04711704205599969,
      "grad_norm": 0.6920540928840637,
      "learning_rate": 1.0285543600844804e-05,
      "loss": 1.1016,
      "step": 1283
    },
    {
      "epoch": 0.04715376617295682,
      "grad_norm": 0.8549882173538208,
      "learning_rate": 1.019260497108453e-05,
      "loss": 1.0688,
      "step": 1284
    },
    {
      "epoch": 0.04719049028991395,
      "grad_norm": 0.7586686611175537,
      "learning_rate": 1.010006558601274e-05,
      "loss": 1.0094,
      "step": 1285
    },
    {
      "epoch": 0.04722721440687108,
      "grad_norm": 0.6976699233055115,
      "learning_rate": 1.000792585701934e-05,
      "loss": 0.9283,
      "step": 1286
    },
    {
      "epoch": 0.047263938523828215,
      "grad_norm": 0.7644249796867371,
      "learning_rate": 9.91618619371757e-06,
      "loss": 1.0598,
      "step": 1287
    },
    {
      "epoch": 0.04730066264078535,
      "grad_norm": 0.7376473546028137,
      "learning_rate": 9.82484700394215e-06,
      "loss": 1.0796,
      "step": 1288
    },
    {
      "epoch": 0.047337386757742475,
      "grad_norm": 0.8817504644393921,
      "learning_rate": 9.73390869374743e-06,
      "loss": 1.0137,
      "step": 1289
    },
    {
      "epoch": 0.04737411087469961,
      "grad_norm": 0.9132905006408691,
      "learning_rate": 9.643371667405698e-06,
      "loss": 1.1351,
      "step": 1290
    },
    {
      "epoch": 0.04741083499165674,
      "grad_norm": 0.9441564679145813,
      "learning_rate": 9.553236327405246e-06,
      "loss": 1.2052,
      "step": 1291
    },
    {
      "epoch": 0.04744755910861387,
      "grad_norm": 0.8101649284362793,
      "learning_rate": 9.463503074448677e-06,
      "loss": 0.9795,
      "step": 1292
    },
    {
      "epoch": 0.047484283225571,
      "grad_norm": 0.9158912301063538,
      "learning_rate": 9.374172307451068e-06,
      "loss": 0.9678,
      "step": 1293
    },
    {
      "epoch": 0.047521007342528135,
      "grad_norm": 1.2156285047531128,
      "learning_rate": 9.285244423538197e-06,
      "loss": 1.2999,
      "step": 1294
    },
    {
      "epoch": 0.04755773145948527,
      "grad_norm": 1.0012671947479248,
      "learning_rate": 9.196719818044886e-06,
      "loss": 1.2677,
      "step": 1295
    },
    {
      "epoch": 0.047594455576442395,
      "grad_norm": 1.1155622005462646,
      "learning_rate": 9.108598884513053e-06,
      "loss": 1.2022,
      "step": 1296
    },
    {
      "epoch": 0.04763117969339953,
      "grad_norm": 1.0548162460327148,
      "learning_rate": 9.020882014690136e-06,
      "loss": 0.8501,
      "step": 1297
    },
    {
      "epoch": 0.04766790381035666,
      "grad_norm": 1.1760525703430176,
      "learning_rate": 8.933569598527247e-06,
      "loss": 1.1983,
      "step": 1298
    },
    {
      "epoch": 0.047704627927313795,
      "grad_norm": 1.2093044519424438,
      "learning_rate": 8.846662024177477e-06,
      "loss": 1.2051,
      "step": 1299
    },
    {
      "epoch": 0.04774135204427092,
      "grad_norm": 1.9937169551849365,
      "learning_rate": 8.760159677994172e-06,
      "loss": 1.1526,
      "step": 1300
    },
    {
      "epoch": 0.047778076161228054,
      "grad_norm": 0.16868595778942108,
      "learning_rate": 8.674062944529216e-06,
      "loss": 1.7883,
      "step": 1301
    },
    {
      "epoch": 0.04781480027818519,
      "grad_norm": 0.19784905016422272,
      "learning_rate": 8.588372206531292e-06,
      "loss": 1.7872,
      "step": 1302
    },
    {
      "epoch": 0.047851524395142314,
      "grad_norm": 0.20712849497795105,
      "learning_rate": 8.503087844944213e-06,
      "loss": 1.6312,
      "step": 1303
    },
    {
      "epoch": 0.04788824851209945,
      "grad_norm": 0.22899803519248962,
      "learning_rate": 8.418210238905256e-06,
      "loss": 1.4456,
      "step": 1304
    },
    {
      "epoch": 0.04792497262905658,
      "grad_norm": 0.2344983071088791,
      "learning_rate": 8.333739765743398e-06,
      "loss": 1.7223,
      "step": 1305
    },
    {
      "epoch": 0.047961696746013714,
      "grad_norm": 0.2784774601459503,
      "learning_rate": 8.249676800977658e-06,
      "loss": 1.6699,
      "step": 1306
    },
    {
      "epoch": 0.04799842086297084,
      "grad_norm": 0.253754585981369,
      "learning_rate": 8.16602171831553e-06,
      "loss": 1.6038,
      "step": 1307
    },
    {
      "epoch": 0.048035144979927974,
      "grad_norm": 0.26395776867866516,
      "learning_rate": 8.082774889651168e-06,
      "loss": 1.5056,
      "step": 1308
    },
    {
      "epoch": 0.04807186909688511,
      "grad_norm": 0.3018849194049835,
      "learning_rate": 7.999936685063835e-06,
      "loss": 1.6492,
      "step": 1309
    },
    {
      "epoch": 0.04810859321384224,
      "grad_norm": 0.2880493402481079,
      "learning_rate": 7.91750747281621e-06,
      "loss": 1.3165,
      "step": 1310
    },
    {
      "epoch": 0.04814531733079937,
      "grad_norm": 0.3200315237045288,
      "learning_rate": 7.835487619352811e-06,
      "loss": 1.4915,
      "step": 1311
    },
    {
      "epoch": 0.0481820414477565,
      "grad_norm": 0.33737853169441223,
      "learning_rate": 7.753877489298244e-06,
      "loss": 1.5183,
      "step": 1312
    },
    {
      "epoch": 0.048218765564713634,
      "grad_norm": 0.32933565974235535,
      "learning_rate": 7.67267744545579e-06,
      "loss": 1.384,
      "step": 1313
    },
    {
      "epoch": 0.04825548968167077,
      "grad_norm": 0.3540426194667816,
      "learning_rate": 7.591887848805545e-06,
      "loss": 1.3545,
      "step": 1314
    },
    {
      "epoch": 0.048292213798627893,
      "grad_norm": 0.3980407118797302,
      "learning_rate": 7.5115090585029966e-06,
      "loss": 1.472,
      "step": 1315
    },
    {
      "epoch": 0.04832893791558503,
      "grad_norm": 0.35630708932876587,
      "learning_rate": 7.431541431877342e-06,
      "loss": 1.4491,
      "step": 1316
    },
    {
      "epoch": 0.04836566203254216,
      "grad_norm": 0.3926009237766266,
      "learning_rate": 7.351985324429933e-06,
      "loss": 1.3691,
      "step": 1317
    },
    {
      "epoch": 0.04840238614949929,
      "grad_norm": 0.3919435739517212,
      "learning_rate": 7.272841089832694e-06,
      "loss": 1.3676,
      "step": 1318
    },
    {
      "epoch": 0.04843911026645642,
      "grad_norm": 0.4175911247730255,
      "learning_rate": 7.194109079926514e-06,
      "loss": 1.3604,
      "step": 1319
    },
    {
      "epoch": 0.04847583438341355,
      "grad_norm": 0.40168043971061707,
      "learning_rate": 7.115789644719728e-06,
      "loss": 1.2789,
      "step": 1320
    },
    {
      "epoch": 0.04851255850037069,
      "grad_norm": 0.4483112394809723,
      "learning_rate": 7.037883132386547e-06,
      "loss": 1.5524,
      "step": 1321
    },
    {
      "epoch": 0.04854928261732781,
      "grad_norm": 0.4645056128501892,
      "learning_rate": 6.960389889265517e-06,
      "loss": 1.3911,
      "step": 1322
    },
    {
      "epoch": 0.048586006734284946,
      "grad_norm": 0.5309298038482666,
      "learning_rate": 6.883310259857944e-06,
      "loss": 1.3882,
      "step": 1323
    },
    {
      "epoch": 0.04862273085124208,
      "grad_norm": 0.5240669250488281,
      "learning_rate": 6.806644586826383e-06,
      "loss": 1.1695,
      "step": 1324
    },
    {
      "epoch": 0.04865945496819921,
      "grad_norm": 0.6165319085121155,
      "learning_rate": 6.730393210993147e-06,
      "loss": 1.4322,
      "step": 1325
    },
    {
      "epoch": 0.04869617908515634,
      "grad_norm": 0.5332204103469849,
      "learning_rate": 6.654556471338746e-06,
      "loss": 1.194,
      "step": 1326
    },
    {
      "epoch": 0.04873290320211347,
      "grad_norm": 0.6382244229316711,
      "learning_rate": 6.579134705000412e-06,
      "loss": 1.3024,
      "step": 1327
    },
    {
      "epoch": 0.048769627319070606,
      "grad_norm": 0.5919054746627808,
      "learning_rate": 6.504128247270546e-06,
      "loss": 1.1951,
      "step": 1328
    },
    {
      "epoch": 0.04880635143602773,
      "grad_norm": 0.6471994519233704,
      "learning_rate": 6.429537431595312e-06,
      "loss": 1.4823,
      "step": 1329
    },
    {
      "epoch": 0.048843075552984866,
      "grad_norm": 0.6599531769752502,
      "learning_rate": 6.355362589573077e-06,
      "loss": 1.1875,
      "step": 1330
    },
    {
      "epoch": 0.048879799669942,
      "grad_norm": 0.8023070096969604,
      "learning_rate": 6.2816040509530165e-06,
      "loss": 1.0226,
      "step": 1331
    },
    {
      "epoch": 0.04891652378689913,
      "grad_norm": 0.6406183242797852,
      "learning_rate": 6.2082621436335475e-06,
      "loss": 1.1755,
      "step": 1332
    },
    {
      "epoch": 0.04895324790385626,
      "grad_norm": 0.6497223377227783,
      "learning_rate": 6.135337193660962e-06,
      "loss": 0.9662,
      "step": 1333
    },
    {
      "epoch": 0.04898997202081339,
      "grad_norm": 0.6811114549636841,
      "learning_rate": 6.062829525227909e-06,
      "loss": 1.105,
      "step": 1334
    },
    {
      "epoch": 0.049026696137770526,
      "grad_norm": 0.835659384727478,
      "learning_rate": 5.990739460672024e-06,
      "loss": 1.2335,
      "step": 1335
    },
    {
      "epoch": 0.04906342025472766,
      "grad_norm": 0.8356134295463562,
      "learning_rate": 5.9190673204744255e-06,
      "loss": 0.8365,
      "step": 1336
    },
    {
      "epoch": 0.049100144371684785,
      "grad_norm": 0.6929663419723511,
      "learning_rate": 5.84781342325833e-06,
      "loss": 1.1542,
      "step": 1337
    },
    {
      "epoch": 0.04913686848864192,
      "grad_norm": 0.7417263388633728,
      "learning_rate": 5.77697808578761e-06,
      "loss": 1.0084,
      "step": 1338
    },
    {
      "epoch": 0.04917359260559905,
      "grad_norm": 0.8116679787635803,
      "learning_rate": 5.706561622965467e-06,
      "loss": 1.2596,
      "step": 1339
    },
    {
      "epoch": 0.049210316722556185,
      "grad_norm": 0.8768166303634644,
      "learning_rate": 5.636564347832907e-06,
      "loss": 0.9987,
      "step": 1340
    },
    {
      "epoch": 0.04924704083951331,
      "grad_norm": 0.8618502020835876,
      "learning_rate": 5.566986571567401e-06,
      "loss": 1.2061,
      "step": 1341
    },
    {
      "epoch": 0.049283764956470445,
      "grad_norm": 0.7894880175590515,
      "learning_rate": 5.497828603481569e-06,
      "loss": 0.9364,
      "step": 1342
    },
    {
      "epoch": 0.04932048907342758,
      "grad_norm": 1.2407780885696411,
      "learning_rate": 5.429090751021704e-06,
      "loss": 1.0144,
      "step": 1343
    },
    {
      "epoch": 0.049357213190384705,
      "grad_norm": 0.9257331490516663,
      "learning_rate": 5.3607733197664436e-06,
      "loss": 1.1454,
      "step": 1344
    },
    {
      "epoch": 0.04939393730734184,
      "grad_norm": 1.0792808532714844,
      "learning_rate": 5.2928766134254345e-06,
      "loss": 1.0379,
      "step": 1345
    },
    {
      "epoch": 0.04943066142429897,
      "grad_norm": 1.0704092979431152,
      "learning_rate": 5.225400933837954e-06,
      "loss": 0.9817,
      "step": 1346
    },
    {
      "epoch": 0.049467385541256105,
      "grad_norm": 1.1258933544158936,
      "learning_rate": 5.158346580971573e-06,
      "loss": 1.3388,
      "step": 1347
    },
    {
      "epoch": 0.04950410965821323,
      "grad_norm": 1.3436347246170044,
      "learning_rate": 5.091713852920854e-06,
      "loss": 1.0902,
      "step": 1348
    },
    {
      "epoch": 0.049540833775170365,
      "grad_norm": 1.3909409046173096,
      "learning_rate": 5.025503045905933e-06,
      "loss": 1.2435,
      "step": 1349
    },
    {
      "epoch": 0.0495775578921275,
      "grad_norm": 1.4868143796920776,
      "learning_rate": 4.959714454271369e-06,
      "loss": 1.2709,
      "step": 1350
    },
    {
      "epoch": 0.04961428200908463,
      "grad_norm": 0.17249305546283722,
      "learning_rate": 4.8943483704846475e-06,
      "loss": 1.7711,
      "step": 1351
    },
    {
      "epoch": 0.04965100612604176,
      "grad_norm": 0.19064277410507202,
      "learning_rate": 4.829405085134997e-06,
      "loss": 1.6214,
      "step": 1352
    },
    {
      "epoch": 0.04968773024299889,
      "grad_norm": 0.21073834598064423,
      "learning_rate": 4.764884886932086e-06,
      "loss": 1.5914,
      "step": 1353
    },
    {
      "epoch": 0.049724454359956025,
      "grad_norm": 0.2495916187763214,
      "learning_rate": 4.700788062704687e-06,
      "loss": 1.5433,
      "step": 1354
    },
    {
      "epoch": 0.04976117847691315,
      "grad_norm": 0.23190660774707794,
      "learning_rate": 4.6371148973994525e-06,
      "loss": 1.6387,
      "step": 1355
    },
    {
      "epoch": 0.049797902593870284,
      "grad_norm": 0.2605883479118347,
      "learning_rate": 4.573865674079625e-06,
      "loss": 1.511,
      "step": 1356
    },
    {
      "epoch": 0.04983462671082742,
      "grad_norm": 0.27561816573143005,
      "learning_rate": 4.511040673923828e-06,
      "loss": 1.5322,
      "step": 1357
    },
    {
      "epoch": 0.04987135082778455,
      "grad_norm": 0.2927427589893341,
      "learning_rate": 4.448640176224694e-06,
      "loss": 1.4982,
      "step": 1358
    },
    {
      "epoch": 0.04990807494474168,
      "grad_norm": 0.3009487986564636,
      "learning_rate": 4.386664458387779e-06,
      "loss": 1.4877,
      "step": 1359
    },
    {
      "epoch": 0.04994479906169881,
      "grad_norm": 0.29899242520332336,
      "learning_rate": 4.325113795930203e-06,
      "loss": 1.5148,
      "step": 1360
    },
    {
      "epoch": 0.049981523178655944,
      "grad_norm": 0.3272773325443268,
      "learning_rate": 4.263988462479484e-06,
      "loss": 1.6519,
      "step": 1361
    },
    {
      "epoch": 0.05001824729561308,
      "grad_norm": 0.3480778634548187,
      "learning_rate": 4.203288729772326e-06,
      "loss": 1.379,
      "step": 1362
    },
    {
      "epoch": 0.050054971412570204,
      "grad_norm": 0.3345130980014801,
      "learning_rate": 4.143014867653383e-06,
      "loss": 1.5614,
      "step": 1363
    },
    {
      "epoch": 0.05009169552952734,
      "grad_norm": 0.3547692596912384,
      "learning_rate": 4.083167144074073e-06,
      "loss": 1.3768,
      "step": 1364
    },
    {
      "epoch": 0.05012841964648447,
      "grad_norm": 0.3831464648246765,
      "learning_rate": 4.023745825091407e-06,
      "loss": 1.5219,
      "step": 1365
    },
    {
      "epoch": 0.050165143763441604,
      "grad_norm": 0.37881243228912354,
      "learning_rate": 3.964751174866765e-06,
      "loss": 1.4037,
      "step": 1366
    },
    {
      "epoch": 0.05020186788039873,
      "grad_norm": 0.3848232328891754,
      "learning_rate": 3.906183455664725e-06,
      "loss": 1.6502,
      "step": 1367
    },
    {
      "epoch": 0.050238591997355864,
      "grad_norm": 0.4264500141143799,
      "learning_rate": 3.84804292785198e-06,
      "loss": 1.5974,
      "step": 1368
    },
    {
      "epoch": 0.050275316114313,
      "grad_norm": 0.42792972922325134,
      "learning_rate": 3.7903298498960572e-06,
      "loss": 1.3323,
      "step": 1369
    },
    {
      "epoch": 0.05031204023127012,
      "grad_norm": 0.4355097711086273,
      "learning_rate": 3.7330444783642338e-06,
      "loss": 1.3661,
      "step": 1370
    },
    {
      "epoch": 0.05034876434822726,
      "grad_norm": 0.43402203917503357,
      "learning_rate": 3.676187067922421e-06,
      "loss": 1.2927,
      "step": 1371
    },
    {
      "epoch": 0.05038548846518439,
      "grad_norm": 0.4872083067893982,
      "learning_rate": 3.619757871333973e-06,
      "loss": 1.3547,
      "step": 1372
    },
    {
      "epoch": 0.05042221258214152,
      "grad_norm": 0.47426435351371765,
      "learning_rate": 3.563757139458579e-06,
      "loss": 1.2083,
      "step": 1373
    },
    {
      "epoch": 0.05045893669909865,
      "grad_norm": 0.4910436272621155,
      "learning_rate": 3.5081851212512175e-06,
      "loss": 1.2093,
      "step": 1374
    },
    {
      "epoch": 0.05049566081605578,
      "grad_norm": 0.5242326259613037,
      "learning_rate": 3.4530420637609363e-06,
      "loss": 1.4832,
      "step": 1375
    },
    {
      "epoch": 0.05053238493301292,
      "grad_norm": 0.577261745929718,
      "learning_rate": 3.3983282121298086e-06,
      "loss": 1.0825,
      "step": 1376
    },
    {
      "epoch": 0.05056910904997005,
      "grad_norm": 0.5669834017753601,
      "learning_rate": 3.3440438095919126e-06,
      "loss": 1.0555,
      "step": 1377
    },
    {
      "epoch": 0.050605833166927176,
      "grad_norm": 0.5862236022949219,
      "learning_rate": 3.290189097472096e-06,
      "loss": 1.501,
      "step": 1378
    },
    {
      "epoch": 0.05064255728388431,
      "grad_norm": 0.6771510243415833,
      "learning_rate": 3.236764315185037e-06,
      "loss": 1.2267,
      "step": 1379
    },
    {
      "epoch": 0.05067928140084144,
      "grad_norm": 0.6482548713684082,
      "learning_rate": 3.1837697002341293e-06,
      "loss": 1.2639,
      "step": 1380
    },
    {
      "epoch": 0.05071600551779857,
      "grad_norm": 0.6974939107894897,
      "learning_rate": 3.131205488210409e-06,
      "loss": 1.1329,
      "step": 1381
    },
    {
      "epoch": 0.0507527296347557,
      "grad_norm": 0.7010265588760376,
      "learning_rate": 3.0790719127915646e-06,
      "loss": 0.9958,
      "step": 1382
    },
    {
      "epoch": 0.050789453751712836,
      "grad_norm": 0.8448973298072815,
      "learning_rate": 3.0273692057408265e-06,
      "loss": 1.0403,
      "step": 1383
    },
    {
      "epoch": 0.05082617786866997,
      "grad_norm": 0.6972507238388062,
      "learning_rate": 2.976097596905969e-06,
      "loss": 1.2725,
      "step": 1384
    },
    {
      "epoch": 0.050862901985627096,
      "grad_norm": 0.7378231287002563,
      "learning_rate": 2.9252573142183326e-06,
      "loss": 1.0899,
      "step": 1385
    },
    {
      "epoch": 0.05089962610258423,
      "grad_norm": 0.7057532072067261,
      "learning_rate": 2.874848583691714e-06,
      "loss": 1.1301,
      "step": 1386
    },
    {
      "epoch": 0.05093635021954136,
      "grad_norm": 0.9979751110076904,
      "learning_rate": 2.8248716294214774e-06,
      "loss": 1.1633,
      "step": 1387
    },
    {
      "epoch": 0.050973074336498496,
      "grad_norm": 0.8682811856269836,
      "learning_rate": 2.7753266735834338e-06,
      "loss": 1.2863,
      "step": 1388
    },
    {
      "epoch": 0.05100979845345562,
      "grad_norm": 0.8106860518455505,
      "learning_rate": 2.7262139364329643e-06,
      "loss": 1.0458,
      "step": 1389
    },
    {
      "epoch": 0.051046522570412756,
      "grad_norm": 0.9181169271469116,
      "learning_rate": 2.677533636303964e-06,
      "loss": 1.287,
      "step": 1390
    },
    {
      "epoch": 0.05108324668736989,
      "grad_norm": 0.8462876081466675,
      "learning_rate": 2.6292859896079213e-06,
      "loss": 0.9462,
      "step": 1391
    },
    {
      "epoch": 0.05111997080432702,
      "grad_norm": 0.795131266117096,
      "learning_rate": 2.581471210832931e-06,
      "loss": 0.8932,
      "step": 1392
    },
    {
      "epoch": 0.05115669492128415,
      "grad_norm": 0.9683177471160889,
      "learning_rate": 2.5340895125427364e-06,
      "loss": 1.1519,
      "step": 1393
    },
    {
      "epoch": 0.05119341903824128,
      "grad_norm": 1.0324594974517822,
      "learning_rate": 2.4871411053757898e-06,
      "loss": 1.1512,
      "step": 1394
    },
    {
      "epoch": 0.051230143155198415,
      "grad_norm": 1.1333949565887451,
      "learning_rate": 2.440626198044327e-06,
      "loss": 1.2427,
      "step": 1395
    },
    {
      "epoch": 0.05126686727215554,
      "grad_norm": 1.0914883613586426,
      "learning_rate": 2.394544997333437e-06,
      "loss": 0.9285,
      "step": 1396
    },
    {
      "epoch": 0.051303591389112675,
      "grad_norm": 1.1409419775009155,
      "learning_rate": 2.3488977081001394e-06,
      "loss": 1.0743,
      "step": 1397
    },
    {
      "epoch": 0.05134031550606981,
      "grad_norm": 0.9835062623023987,
      "learning_rate": 2.3036845332724543e-06,
      "loss": 0.8515,
      "step": 1398
    },
    {
      "epoch": 0.05137703962302694,
      "grad_norm": 1.059680700302124,
      "learning_rate": 2.2589056738485324e-06,
      "loss": 0.757,
      "step": 1399
    },
    {
      "epoch": 0.05141376373998407,
      "grad_norm": 1.9644958972930908,
      "learning_rate": 2.2145613288957478e-06,
      "loss": 1.8049,
      "step": 1400
    },
    {
      "epoch": 0.0514504878569412,
      "grad_norm": 0.16022421419620514,
      "learning_rate": 2.170651695549786e-06,
      "loss": 1.5506,
      "step": 1401
    },
    {
      "epoch": 0.051487211973898335,
      "grad_norm": 0.19960249960422516,
      "learning_rate": 2.1271769690138332e-06,
      "loss": 1.7443,
      "step": 1402
    },
    {
      "epoch": 0.05152393609085547,
      "grad_norm": 0.2070237547159195,
      "learning_rate": 2.084137342557646e-06,
      "loss": 1.653,
      "step": 1403
    },
    {
      "epoch": 0.051560660207812595,
      "grad_norm": 0.2135138362646103,
      "learning_rate": 2.0415330075166937e-06,
      "loss": 1.5366,
      "step": 1404
    },
    {
      "epoch": 0.05159738432476973,
      "grad_norm": 0.23227788507938385,
      "learning_rate": 1.9993641532913833e-06,
      "loss": 1.3766,
      "step": 1405
    },
    {
      "epoch": 0.05163410844172686,
      "grad_norm": 0.2690650224685669,
      "learning_rate": 1.9576309673461357e-06,
      "loss": 1.5288,
      "step": 1406
    },
    {
      "epoch": 0.05167083255868399,
      "grad_norm": 0.27190253138542175,
      "learning_rate": 1.916333635208556e-06,
      "loss": 1.465,
      "step": 1407
    },
    {
      "epoch": 0.05170755667564112,
      "grad_norm": 0.2702263593673706,
      "learning_rate": 1.8754723404686425e-06,
      "loss": 1.7212,
      "step": 1408
    },
    {
      "epoch": 0.051744280792598255,
      "grad_norm": 0.2781308591365814,
      "learning_rate": 1.8350472647780116e-06,
      "loss": 1.3864,
      "step": 1409
    },
    {
      "epoch": 0.05178100490955539,
      "grad_norm": 0.2912030518054962,
      "learning_rate": 1.7950585878489856e-06,
      "loss": 1.4936,
      "step": 1410
    },
    {
      "epoch": 0.051817729026512514,
      "grad_norm": 0.31920066475868225,
      "learning_rate": 1.7555064874538397e-06,
      "loss": 1.3942,
      "step": 1411
    },
    {
      "epoch": 0.05185445314346965,
      "grad_norm": 0.3033767640590668,
      "learning_rate": 1.7163911394240672e-06,
      "loss": 1.5894,
      "step": 1412
    },
    {
      "epoch": 0.05189117726042678,
      "grad_norm": 0.36123421788215637,
      "learning_rate": 1.6777127176495043e-06,
      "loss": 1.4027,
      "step": 1413
    },
    {
      "epoch": 0.051927901377383914,
      "grad_norm": 0.3683607280254364,
      "learning_rate": 1.6394713940776296e-06,
      "loss": 1.4363,
      "step": 1414
    },
    {
      "epoch": 0.05196462549434104,
      "grad_norm": 0.3586466312408447,
      "learning_rate": 1.6016673387127646e-06,
      "loss": 1.3447,
      "step": 1415
    },
    {
      "epoch": 0.052001349611298174,
      "grad_norm": 0.4015662968158722,
      "learning_rate": 1.5643007196153302e-06,
      "loss": 1.4737,
      "step": 1416
    },
    {
      "epoch": 0.05203807372825531,
      "grad_norm": 0.41332030296325684,
      "learning_rate": 1.5273717029010925e-06,
      "loss": 1.2422,
      "step": 1417
    },
    {
      "epoch": 0.05207479784521244,
      "grad_norm": 0.43748196959495544,
      "learning_rate": 1.4908804527404286e-06,
      "loss": 1.4962,
      "step": 1418
    },
    {
      "epoch": 0.05211152196216957,
      "grad_norm": 0.4502474367618561,
      "learning_rate": 1.4548271313575835e-06,
      "loss": 1.341,
      "step": 1419
    },
    {
      "epoch": 0.0521482460791267,
      "grad_norm": 0.4579099714756012,
      "learning_rate": 1.4192118990299707e-06,
      "loss": 1.4483,
      "step": 1420
    },
    {
      "epoch": 0.052184970196083834,
      "grad_norm": 0.48423585295677185,
      "learning_rate": 1.3840349140874619e-06,
      "loss": 1.1926,
      "step": 1421
    },
    {
      "epoch": 0.05222169431304096,
      "grad_norm": 0.5109123587608337,
      "learning_rate": 1.3492963329116537e-06,
      "loss": 1.4,
      "step": 1422
    },
    {
      "epoch": 0.052258418429998094,
      "grad_norm": 0.5163117051124573,
      "learning_rate": 1.3149963099352014e-06,
      "loss": 1.4024,
      "step": 1423
    },
    {
      "epoch": 0.05229514254695523,
      "grad_norm": 0.5232509970664978,
      "learning_rate": 1.2811349976411202e-06,
      "loss": 1.2364,
      "step": 1424
    },
    {
      "epoch": 0.05233186666391236,
      "grad_norm": 0.5789957642555237,
      "learning_rate": 1.2477125465620853e-06,
      "loss": 1.3655,
      "step": 1425
    },
    {
      "epoch": 0.05236859078086949,
      "grad_norm": 0.5959565043449402,
      "learning_rate": 1.2147291052798216e-06,
      "loss": 1.3821,
      "step": 1426
    },
    {
      "epoch": 0.05240531489782662,
      "grad_norm": 0.6262691020965576,
      "learning_rate": 1.1821848204243814e-06,
      "loss": 1.1136,
      "step": 1427
    },
    {
      "epoch": 0.05244203901478375,
      "grad_norm": 0.6518732905387878,
      "learning_rate": 1.1500798366735233e-06,
      "loss": 1.5501,
      "step": 1428
    },
    {
      "epoch": 0.05247876313174089,
      "grad_norm": 0.6539964079856873,
      "learning_rate": 1.1184142967520794e-06,
      "loss": 1.3061,
      "step": 1429
    },
    {
      "epoch": 0.05251548724869801,
      "grad_norm": 0.7202858328819275,
      "learning_rate": 1.0871883414312777e-06,
      "loss": 1.0438,
      "step": 1430
    },
    {
      "epoch": 0.052552211365655147,
      "grad_norm": 0.6566448211669922,
      "learning_rate": 1.0564021095281652e-06,
      "loss": 1.2892,
      "step": 1431
    },
    {
      "epoch": 0.05258893548261228,
      "grad_norm": 0.6669291853904724,
      "learning_rate": 1.0260557379049519e-06,
      "loss": 1.1298,
      "step": 1432
    },
    {
      "epoch": 0.05262565959956941,
      "grad_norm": 0.7811323404312134,
      "learning_rate": 9.96149361468457e-07,
      "loss": 1.1922,
      "step": 1433
    },
    {
      "epoch": 0.05266238371652654,
      "grad_norm": 0.8510951399803162,
      "learning_rate": 9.66683113169431e-07,
      "loss": 1.3827,
      "step": 1434
    },
    {
      "epoch": 0.05269910783348367,
      "grad_norm": 0.7888028025627136,
      "learning_rate": 9.376571240020227e-07,
      "loss": 1.1548,
      "step": 1435
    },
    {
      "epoch": 0.052735831950440806,
      "grad_norm": 0.8457626104354858,
      "learning_rate": 9.090715230031688e-07,
      "loss": 1.2501,
      "step": 1436
    },
    {
      "epoch": 0.05277255606739793,
      "grad_norm": 0.8713982701301575,
      "learning_rate": 8.809264372520609e-07,
      "loss": 0.8258,
      "step": 1437
    },
    {
      "epoch": 0.052809280184355066,
      "grad_norm": 0.7833504676818848,
      "learning_rate": 8.532219918695128e-07,
      "loss": 0.9928,
      "step": 1438
    },
    {
      "epoch": 0.0528460043013122,
      "grad_norm": 0.9501321911811829,
      "learning_rate": 8.259583100174606e-07,
      "loss": 1.2495,
      "step": 1439
    },
    {
      "epoch": 0.05288272841826933,
      "grad_norm": 0.8797469139099121,
      "learning_rate": 7.991355128984079e-07,
      "loss": 1.2464,
      "step": 1440
    },
    {
      "epoch": 0.05291945253522646,
      "grad_norm": 0.8270118236541748,
      "learning_rate": 7.727537197548707e-07,
      "loss": 1.1221,
      "step": 1441
    },
    {
      "epoch": 0.05295617665218359,
      "grad_norm": 0.8835216760635376,
      "learning_rate": 7.468130478688218e-07,
      "loss": 1.1955,
      "step": 1442
    },
    {
      "epoch": 0.052992900769140726,
      "grad_norm": 0.9682307839393616,
      "learning_rate": 7.213136125612586e-07,
      "loss": 1.369,
      "step": 1443
    },
    {
      "epoch": 0.05302962488609786,
      "grad_norm": 0.8620365858078003,
      "learning_rate": 6.962555271915805e-07,
      "loss": 0.9356,
      "step": 1444
    },
    {
      "epoch": 0.053066349003054986,
      "grad_norm": 1.0178951025009155,
      "learning_rate": 6.716389031571568e-07,
      "loss": 0.9237,
      "step": 1445
    },
    {
      "epoch": 0.05310307312001212,
      "grad_norm": 1.129179835319519,
      "learning_rate": 6.474638498928265e-07,
      "loss": 1.2062,
      "step": 1446
    },
    {
      "epoch": 0.05313979723696925,
      "grad_norm": 1.0531798601150513,
      "learning_rate": 6.237304748703543e-07,
      "loss": 0.8946,
      "step": 1447
    },
    {
      "epoch": 0.05317652135392638,
      "grad_norm": 1.2256871461868286,
      "learning_rate": 6.004388835980423e-07,
      "loss": 0.9899,
      "step": 1448
    },
    {
      "epoch": 0.05321324547088351,
      "grad_norm": 1.4514309167861938,
      "learning_rate": 5.77589179620186e-07,
      "loss": 1.1099,
      "step": 1449
    },
    {
      "epoch": 0.053249969587840645,
      "grad_norm": 2.0371334552764893,
      "learning_rate": 5.55181464516652e-07,
      "loss": 1.1249,
      "step": 1450
    },
    {
      "epoch": 0.05328669370479778,
      "grad_norm": 0.15685400366783142,
      "learning_rate": 5.332158379024122e-07,
      "loss": 1.5735,
      "step": 1451
    },
    {
      "epoch": 0.053323417821754905,
      "grad_norm": 0.1831342875957489,
      "learning_rate": 5.116923974270993e-07,
      "loss": 1.4026,
      "step": 1452
    },
    {
      "epoch": 0.05336014193871204,
      "grad_norm": 0.1996254175901413,
      "learning_rate": 4.906112387745965e-07,
      "loss": 1.4875,
      "step": 1453
    },
    {
      "epoch": 0.05339686605566917,
      "grad_norm": 0.21851493418216705,
      "learning_rate": 4.6997245566257064e-07,
      "loss": 1.4926,
      "step": 1454
    },
    {
      "epoch": 0.053433590172626305,
      "grad_norm": 0.2196923792362213,
      "learning_rate": 4.497761398421063e-07,
      "loss": 1.5037,
      "step": 1455
    },
    {
      "epoch": 0.05347031428958343,
      "grad_norm": 0.22560672461986542,
      "learning_rate": 4.3002238109723927e-07,
      "loss": 1.4558,
      "step": 1456
    },
    {
      "epoch": 0.053507038406540565,
      "grad_norm": 0.2492704838514328,
      "learning_rate": 4.107112672446123e-07,
      "loss": 1.5459,
      "step": 1457
    },
    {
      "epoch": 0.0535437625234977,
      "grad_norm": 0.24907377362251282,
      "learning_rate": 3.9184288413306456e-07,
      "loss": 1.3573,
      "step": 1458
    },
    {
      "epoch": 0.05358048664045483,
      "grad_norm": 0.28745633363723755,
      "learning_rate": 3.734173156432208e-07,
      "loss": 1.4093,
      "step": 1459
    },
    {
      "epoch": 0.05361721075741196,
      "grad_norm": 0.27202585339546204,
      "learning_rate": 3.554346436871581e-07,
      "loss": 1.2246,
      "step": 1460
    },
    {
      "epoch": 0.05365393487436909,
      "grad_norm": 0.30589747428894043,
      "learning_rate": 3.3789494820803957e-07,
      "loss": 1.435,
      "step": 1461
    },
    {
      "epoch": 0.053690658991326225,
      "grad_norm": 0.31751465797424316,
      "learning_rate": 3.2079830717972606e-07,
      "loss": 1.4132,
      "step": 1462
    },
    {
      "epoch": 0.05372738310828335,
      "grad_norm": 0.3328952193260193,
      "learning_rate": 3.041447966064648e-07,
      "loss": 1.4576,
      "step": 1463
    },
    {
      "epoch": 0.053764107225240484,
      "grad_norm": 0.3328303098678589,
      "learning_rate": 2.8793449052254563e-07,
      "loss": 1.4568,
      "step": 1464
    },
    {
      "epoch": 0.05380083134219762,
      "grad_norm": 0.35016849637031555,
      "learning_rate": 2.721674609919345e-07,
      "loss": 1.4055,
      "step": 1465
    },
    {
      "epoch": 0.05383755545915475,
      "grad_norm": 0.3538975715637207,
      "learning_rate": 2.568437781080069e-07,
      "loss": 1.4922,
      "step": 1466
    },
    {
      "epoch": 0.05387427957611188,
      "grad_norm": 0.37794607877731323,
      "learning_rate": 2.4196350999320384e-07,
      "loss": 1.497,
      "step": 1467
    },
    {
      "epoch": 0.05391100369306901,
      "grad_norm": 0.3674625754356384,
      "learning_rate": 2.275267227987321e-07,
      "loss": 1.4164,
      "step": 1468
    },
    {
      "epoch": 0.053947727810026144,
      "grad_norm": 0.404354065656662,
      "learning_rate": 2.135334807042866e-07,
      "loss": 1.3693,
      "step": 1469
    },
    {
      "epoch": 0.05398445192698328,
      "grad_norm": 0.43882104754447937,
      "learning_rate": 1.9998384591773944e-07,
      "loss": 1.2486,
      "step": 1470
    },
    {
      "epoch": 0.054021176043940404,
      "grad_norm": 0.4645138084888458,
      "learning_rate": 1.8687787867489592e-07,
      "loss": 1.598,
      "step": 1471
    },
    {
      "epoch": 0.05405790016089754,
      "grad_norm": 0.46232399344444275,
      "learning_rate": 1.7421563723919454e-07,
      "loss": 1.4767,
      "step": 1472
    },
    {
      "epoch": 0.05409462427785467,
      "grad_norm": 0.5317553877830505,
      "learning_rate": 1.6199717790145174e-07,
      "loss": 1.3926,
      "step": 1473
    },
    {
      "epoch": 0.0541313483948118,
      "grad_norm": 0.4804433286190033,
      "learning_rate": 1.5022255497962879e-07,
      "loss": 1.4368,
      "step": 1474
    },
    {
      "epoch": 0.05416807251176893,
      "grad_norm": 0.5335580706596375,
      "learning_rate": 1.3889182081860962e-07,
      "loss": 1.1784,
      "step": 1475
    },
    {
      "epoch": 0.054204796628726064,
      "grad_norm": 0.5776420831680298,
      "learning_rate": 1.2800502578991235e-07,
      "loss": 1.2014,
      "step": 1476
    },
    {
      "epoch": 0.0542415207456832,
      "grad_norm": 0.5317526459693909,
      "learning_rate": 1.1756221829148928e-07,
      "loss": 1.2449,
      "step": 1477
    },
    {
      "epoch": 0.054278244862640324,
      "grad_norm": 0.6151018142700195,
      "learning_rate": 1.0756344474753821e-07,
      "loss": 1.1658,
      "step": 1478
    },
    {
      "epoch": 0.05431496897959746,
      "grad_norm": 0.633737325668335,
      "learning_rate": 9.800874960826933e-08,
      "loss": 1.4655,
      "step": 1479
    },
    {
      "epoch": 0.05435169309655459,
      "grad_norm": 0.6437058448791504,
      "learning_rate": 8.889817534969425e-08,
      "loss": 1.3235,
      "step": 1480
    },
    {
      "epoch": 0.054388417213511724,
      "grad_norm": 0.6941370368003845,
      "learning_rate": 8.023176247348163e-08,
      "loss": 1.2014,
      "step": 1481
    },
    {
      "epoch": 0.05442514133046885,
      "grad_norm": 0.652090847492218,
      "learning_rate": 7.200954950673522e-08,
      "loss": 1.1276,
      "step": 1482
    },
    {
      "epoch": 0.05446186544742598,
      "grad_norm": 0.7048991322517395,
      "learning_rate": 6.423157300184946e-08,
      "loss": 0.9756,
      "step": 1483
    },
    {
      "epoch": 0.05449858956438312,
      "grad_norm": 0.759270429611206,
      "learning_rate": 5.6897867536331864e-08,
      "loss": 1.4236,
      "step": 1484
    },
    {
      "epoch": 0.05453531368134025,
      "grad_norm": 0.8099513649940491,
      "learning_rate": 5.000846571264761e-08,
      "loss": 1.0612,
      "step": 1485
    },
    {
      "epoch": 0.054572037798297376,
      "grad_norm": 0.8122737407684326,
      "learning_rate": 4.35633981580974e-08,
      "loss": 1.1602,
      "step": 1486
    },
    {
      "epoch": 0.05460876191525451,
      "grad_norm": 0.8399105668067932,
      "learning_rate": 3.756269352462871e-08,
      "loss": 0.9412,
      "step": 1487
    },
    {
      "epoch": 0.05464548603221164,
      "grad_norm": 0.9260868430137634,
      "learning_rate": 3.20063784888025e-08,
      "loss": 1.2207,
      "step": 1488
    },
    {
      "epoch": 0.05468221014916877,
      "grad_norm": 0.8493340611457825,
      "learning_rate": 2.6894477751548964e-08,
      "loss": 1.2726,
      "step": 1489
    },
    {
      "epoch": 0.0547189342661259,
      "grad_norm": 0.8446462750434875,
      "learning_rate": 2.222701403818972e-08,
      "loss": 0.8549,
      "step": 1490
    },
    {
      "epoch": 0.054755658383083036,
      "grad_norm": 0.8777040243148804,
      "learning_rate": 1.8004008098226887e-08,
      "loss": 1.1841,
      "step": 1491
    },
    {
      "epoch": 0.05479238250004017,
      "grad_norm": 1.0345079898834229,
      "learning_rate": 1.4225478705309769e-08,
      "loss": 1.4095,
      "step": 1492
    },
    {
      "epoch": 0.054829106616997296,
      "grad_norm": 1.047092318534851,
      "learning_rate": 1.0891442657134932e-08,
      "loss": 1.3107,
      "step": 1493
    },
    {
      "epoch": 0.05486583073395443,
      "grad_norm": 1.1943391561508179,
      "learning_rate": 8.001914775401798e-09,
      "loss": 1.1147,
      "step": 1494
    },
    {
      "epoch": 0.05490255485091156,
      "grad_norm": 0.9803537726402283,
      "learning_rate": 5.5569079056794206e-09,
      "loss": 0.7455,
      "step": 1495
    },
    {
      "epoch": 0.054939278967868696,
      "grad_norm": 0.9217520356178284,
      "learning_rate": 3.5564329174064824e-09,
      "loss": 1.0322,
      "step": 1496
    },
    {
      "epoch": 0.05497600308482582,
      "grad_norm": 1.409850001335144,
      "learning_rate": 2.0004987038246824e-09,
      "loss": 0.9365,
      "step": 1497
    },
    {
      "epoch": 0.055012727201782956,
      "grad_norm": 1.0592516660690308,
      "learning_rate": 8.891121819565306e-10,
      "loss": 1.1755,
      "step": 1498
    },
    {
      "epoch": 0.05504945131874009,
      "grad_norm": 1.381886601448059,
      "learning_rate": 2.2227829252763344e-10,
      "loss": 1.3446,
      "step": 1499
    },
    {
      "epoch": 0.055086175435697216,
      "grad_norm": 1.5955051183700562,
      "learning_rate": 0.0,
      "loss": 1.1789,
      "step": 1500
    },
    {
      "epoch": 0.055086175435697216,
      "eval_loss": 1.2574902772903442,
      "eval_runtime": 1106.9066,
      "eval_samples_per_second": 41.433,
      "eval_steps_per_second": 20.716,
      "step": 1500
    }
  ],
  "logging_steps": 1,
  "max_steps": 1500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 375,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.8930384443303526e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}