{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 493,
  "global_step": 1969,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0005078720162519045,
      "grad_norm": 0.8817383050918579,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.9482,
      "step": 1
    },
    {
      "epoch": 0.0005078720162519045,
      "eval_loss": 1.4919016361236572,
      "eval_runtime": 167.0655,
      "eval_samples_per_second": 5.89,
      "eval_steps_per_second": 1.472,
      "step": 1
    },
    {
      "epoch": 0.001015744032503809,
      "grad_norm": 1.0007613897323608,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.7274,
      "step": 2
    },
    {
      "epoch": 0.0015236160487557136,
      "grad_norm": 0.8447886109352112,
      "learning_rate": 6e-06,
      "loss": 1.8784,
      "step": 3
    },
    {
      "epoch": 0.002031488065007618,
      "grad_norm": 0.9496368765830994,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.709,
      "step": 4
    },
    {
      "epoch": 0.0025393600812595226,
      "grad_norm": 0.9440109729766846,
      "learning_rate": 1e-05,
      "loss": 1.4441,
      "step": 5
    },
    {
      "epoch": 0.0030472320975114273,
      "grad_norm": 0.9887049794197083,
      "learning_rate": 1.2e-05,
      "loss": 1.6845,
      "step": 6
    },
    {
      "epoch": 0.0035551041137633316,
      "grad_norm": 1.1271413564682007,
      "learning_rate": 1.4e-05,
      "loss": 1.6197,
      "step": 7
    },
    {
      "epoch": 0.004062976130015236,
      "grad_norm": 1.0455557107925415,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.8842,
      "step": 8
    },
    {
      "epoch": 0.0045708481462671405,
      "grad_norm": 0.9560151100158691,
      "learning_rate": 1.8e-05,
      "loss": 1.8362,
      "step": 9
    },
    {
      "epoch": 0.005078720162519045,
      "grad_norm": 0.97687166929245,
      "learning_rate": 2e-05,
      "loss": 1.6373,
      "step": 10
    },
    {
      "epoch": 0.00558659217877095,
      "grad_norm": 0.9685382843017578,
      "learning_rate": 1.9999987141189844e-05,
      "loss": 1.6555,
      "step": 11
    },
    {
      "epoch": 0.006094464195022855,
      "grad_norm": 0.8788787722587585,
      "learning_rate": 1.9999948564792438e-05,
      "loss": 1.8128,
      "step": 12
    },
    {
      "epoch": 0.006602336211274758,
      "grad_norm": 0.9693357944488525,
      "learning_rate": 1.9999884270906992e-05,
      "loss": 1.6594,
      "step": 13
    },
    {
      "epoch": 0.007110208227526663,
      "grad_norm": 0.8221359252929688,
      "learning_rate": 1.9999794259698856e-05,
      "loss": 1.7551,
      "step": 14
    },
    {
      "epoch": 0.007618080243778568,
      "grad_norm": 0.9802096486091614,
      "learning_rate": 1.9999678531399517e-05,
      "loss": 1.7172,
      "step": 15
    },
    {
      "epoch": 0.008125952260030472,
      "grad_norm": 0.8529878854751587,
      "learning_rate": 1.99995370863066e-05,
      "loss": 1.6958,
      "step": 16
    },
    {
      "epoch": 0.008633824276282377,
      "grad_norm": 1.0876967906951904,
      "learning_rate": 1.9999369924783872e-05,
      "loss": 1.7709,
      "step": 17
    },
    {
      "epoch": 0.009141696292534281,
      "grad_norm": 1.0921239852905273,
      "learning_rate": 1.9999177047261227e-05,
      "loss": 1.8711,
      "step": 18
    },
    {
      "epoch": 0.009649568308786187,
      "grad_norm": 0.9870465993881226,
      "learning_rate": 1.9998958454234704e-05,
      "loss": 1.5542,
      "step": 19
    },
    {
      "epoch": 0.01015744032503809,
      "grad_norm": 0.8051080703735352,
      "learning_rate": 1.999871414626647e-05,
      "loss": 1.8549,
      "step": 20
    },
    {
      "epoch": 0.010665312341289994,
      "grad_norm": 0.9436357021331787,
      "learning_rate": 1.999844412398483e-05,
      "loss": 1.8625,
      "step": 21
    },
    {
      "epoch": 0.0111731843575419,
      "grad_norm": 0.870090901851654,
      "learning_rate": 1.9998148388084213e-05,
      "loss": 1.8584,
      "step": 22
    },
    {
      "epoch": 0.011681056373793804,
      "grad_norm": 0.89214026927948,
      "learning_rate": 1.9997826939325184e-05,
      "loss": 1.7535,
      "step": 23
    },
    {
      "epoch": 0.01218892839004571,
      "grad_norm": 1.119433879852295,
      "learning_rate": 1.9997479778534432e-05,
      "loss": 1.6597,
      "step": 24
    },
    {
      "epoch": 0.012696800406297613,
      "grad_norm": 0.7907304167747498,
      "learning_rate": 1.9997106906604773e-05,
      "loss": 1.8413,
      "step": 25
    },
    {
      "epoch": 0.013204672422549517,
      "grad_norm": 1.2375823259353638,
      "learning_rate": 1.9996708324495144e-05,
      "loss": 1.1807,
      "step": 26
    },
    {
      "epoch": 0.013712544438801422,
      "grad_norm": 1.1361809968948364,
      "learning_rate": 1.99962840332306e-05,
      "loss": 1.6539,
      "step": 27
    },
    {
      "epoch": 0.014220416455053326,
      "grad_norm": 0.967383861541748,
      "learning_rate": 1.9995834033902323e-05,
      "loss": 1.7859,
      "step": 28
    },
    {
      "epoch": 0.014728288471305232,
      "grad_norm": 0.797429621219635,
      "learning_rate": 1.99953583276676e-05,
      "loss": 1.5485,
      "step": 29
    },
    {
      "epoch": 0.015236160487557136,
      "grad_norm": 1.13390052318573,
      "learning_rate": 1.999485691574984e-05,
      "loss": 1.5869,
      "step": 30
    },
    {
      "epoch": 0.01574403250380904,
      "grad_norm": 0.9043781161308289,
      "learning_rate": 1.9994329799438547e-05,
      "loss": 1.8953,
      "step": 31
    },
    {
      "epoch": 0.016251904520060943,
      "grad_norm": 0.9030641317367554,
      "learning_rate": 1.9993776980089344e-05,
      "loss": 1.6764,
      "step": 32
    },
    {
      "epoch": 0.01675977653631285,
      "grad_norm": 0.8708584308624268,
      "learning_rate": 1.9993198459123948e-05,
      "loss": 1.763,
      "step": 33
    },
    {
      "epoch": 0.017267648552564754,
      "grad_norm": 0.8852054476737976,
      "learning_rate": 1.999259423803018e-05,
      "loss": 1.7905,
      "step": 34
    },
    {
      "epoch": 0.017775520568816656,
      "grad_norm": 1.2012213468551636,
      "learning_rate": 1.9991964318361954e-05,
      "loss": 1.6094,
      "step": 35
    },
    {
      "epoch": 0.018283392585068562,
      "grad_norm": 0.8909621834754944,
      "learning_rate": 1.999130870173927e-05,
      "loss": 1.5457,
      "step": 36
    },
    {
      "epoch": 0.018791264601320468,
      "grad_norm": 0.9592351317405701,
      "learning_rate": 1.9990627389848217e-05,
      "loss": 1.7552,
      "step": 37
    },
    {
      "epoch": 0.019299136617572373,
      "grad_norm": 0.8920623660087585,
      "learning_rate": 1.9989920384440968e-05,
      "loss": 1.5726,
      "step": 38
    },
    {
      "epoch": 0.019807008633824275,
      "grad_norm": 0.9171638488769531,
      "learning_rate": 1.9989187687335777e-05,
      "loss": 1.5285,
      "step": 39
    },
    {
      "epoch": 0.02031488065007618,
      "grad_norm": 1.3621869087219238,
      "learning_rate": 1.998842930041696e-05,
      "loss": 1.4605,
      "step": 40
    },
    {
      "epoch": 0.020822752666328086,
      "grad_norm": 1.1085385084152222,
      "learning_rate": 1.9987645225634915e-05,
      "loss": 1.6428,
      "step": 41
    },
    {
      "epoch": 0.02133062468257999,
      "grad_norm": 1.2515841722488403,
      "learning_rate": 1.9986835465006088e-05,
      "loss": 1.3788,
      "step": 42
    },
    {
      "epoch": 0.021838496698831894,
      "grad_norm": 1.1154283285140991,
      "learning_rate": 1.9986000020613e-05,
      "loss": 1.8016,
      "step": 43
    },
    {
      "epoch": 0.0223463687150838,
      "grad_norm": 1.0101381540298462,
      "learning_rate": 1.9985138894604205e-05,
      "loss": 1.712,
      "step": 44
    },
    {
      "epoch": 0.0228542407313357,
      "grad_norm": 1.146767020225525,
      "learning_rate": 1.9984252089194317e-05,
      "loss": 1.7227,
      "step": 45
    },
    {
      "epoch": 0.023362112747587607,
      "grad_norm": 0.4284367263317108,
      "learning_rate": 1.9983339606663993e-05,
      "loss": 0.2803,
      "step": 46
    },
    {
      "epoch": 0.023869984763839513,
      "grad_norm": 1.1581532955169678,
      "learning_rate": 1.998240144935992e-05,
      "loss": 1.7686,
      "step": 47
    },
    {
      "epoch": 0.02437785678009142,
      "grad_norm": 1.028751015663147,
      "learning_rate": 1.9981437619694813e-05,
      "loss": 1.4811,
      "step": 48
    },
    {
      "epoch": 0.02488572879634332,
      "grad_norm": 1.0959101915359497,
      "learning_rate": 1.998044812014741e-05,
      "loss": 1.516,
      "step": 49
    },
    {
      "epoch": 0.025393600812595226,
      "grad_norm": 1.1301296949386597,
      "learning_rate": 1.997943295326247e-05,
      "loss": 1.5932,
      "step": 50
    },
    {
      "epoch": 0.02590147282884713,
      "grad_norm": 1.0573782920837402,
      "learning_rate": 1.9978392121650767e-05,
      "loss": 1.6835,
      "step": 51
    },
    {
      "epoch": 0.026409344845099034,
      "grad_norm": 1.027848720550537,
      "learning_rate": 1.9977325627989066e-05,
      "loss": 1.6119,
      "step": 52
    },
    {
      "epoch": 0.02691721686135094,
      "grad_norm": 1.2942439317703247,
      "learning_rate": 1.9976233475020136e-05,
      "loss": 1.685,
      "step": 53
    },
    {
      "epoch": 0.027425088877602845,
      "grad_norm": 1.1388881206512451,
      "learning_rate": 1.9975115665552732e-05,
      "loss": 1.5779,
      "step": 54
    },
    {
      "epoch": 0.027932960893854747,
      "grad_norm": 1.0829545259475708,
      "learning_rate": 1.99739722024616e-05,
      "loss": 1.3557,
      "step": 55
    },
    {
      "epoch": 0.028440832910106652,
      "grad_norm": 1.2315815687179565,
      "learning_rate": 1.997280308868745e-05,
      "loss": 1.2091,
      "step": 56
    },
    {
      "epoch": 0.028948704926358558,
      "grad_norm": 1.185844898223877,
      "learning_rate": 1.9971608327236968e-05,
      "loss": 1.5351,
      "step": 57
    },
    {
      "epoch": 0.029456576942610464,
      "grad_norm": 0.9325307011604309,
      "learning_rate": 1.9970387921182793e-05,
      "loss": 1.6289,
      "step": 58
    },
    {
      "epoch": 0.029964448958862366,
      "grad_norm": 1.5263203382492065,
      "learning_rate": 1.996914187366352e-05,
      "loss": 1.4187,
      "step": 59
    },
    {
      "epoch": 0.03047232097511427,
      "grad_norm": 1.1244592666625977,
      "learning_rate": 1.9967870187883685e-05,
      "loss": 1.5894,
      "step": 60
    },
    {
      "epoch": 0.030980192991366177,
      "grad_norm": 1.3506087064743042,
      "learning_rate": 1.9966572867113764e-05,
      "loss": 1.5138,
      "step": 61
    },
    {
      "epoch": 0.03148806500761808,
      "grad_norm": 1.2022825479507446,
      "learning_rate": 1.996524991469016e-05,
      "loss": 1.6979,
      "step": 62
    },
    {
      "epoch": 0.031995937023869984,
      "grad_norm": 1.4655373096466064,
      "learning_rate": 1.9963901334015184e-05,
      "loss": 1.5341,
      "step": 63
    },
    {
      "epoch": 0.03250380904012189,
      "grad_norm": 1.375972867012024,
      "learning_rate": 1.996252712855707e-05,
      "loss": 1.2517,
      "step": 64
    },
    {
      "epoch": 0.033011681056373796,
      "grad_norm": 1.1926246881484985,
      "learning_rate": 1.9961127301849947e-05,
      "loss": 1.7264,
      "step": 65
    },
    {
      "epoch": 0.0335195530726257,
      "grad_norm": 1.2697643041610718,
      "learning_rate": 1.9959701857493835e-05,
      "loss": 1.6141,
      "step": 66
    },
    {
      "epoch": 0.0340274250888776,
      "grad_norm": 1.122188925743103,
      "learning_rate": 1.995825079915464e-05,
      "loss": 1.5998,
      "step": 67
    },
    {
      "epoch": 0.03453529710512951,
      "grad_norm": 1.2801930904388428,
      "learning_rate": 1.9956774130564132e-05,
      "loss": 1.6505,
      "step": 68
    },
    {
      "epoch": 0.03504316912138141,
      "grad_norm": 1.1839287281036377,
      "learning_rate": 1.995527185551996e-05,
      "loss": 1.6169,
      "step": 69
    },
    {
      "epoch": 0.03555104113763331,
      "grad_norm": 1.2095578908920288,
      "learning_rate": 1.995374397788561e-05,
      "loss": 1.754,
      "step": 70
    },
    {
      "epoch": 0.03605891315388522,
      "grad_norm": 1.2569252252578735,
      "learning_rate": 1.9952190501590425e-05,
      "loss": 1.5258,
      "step": 71
    },
    {
      "epoch": 0.036566785170137124,
      "grad_norm": 1.2527538537979126,
      "learning_rate": 1.9950611430629577e-05,
      "loss": 1.7069,
      "step": 72
    },
    {
      "epoch": 0.03707465718638903,
      "grad_norm": 1.2936222553253174,
      "learning_rate": 1.9949006769064057e-05,
      "loss": 1.447,
      "step": 73
    },
    {
      "epoch": 0.037582529202640935,
      "grad_norm": 1.5045548677444458,
      "learning_rate": 1.9947376521020675e-05,
      "loss": 1.5354,
      "step": 74
    },
    {
      "epoch": 0.03809040121889284,
      "grad_norm": 1.201399564743042,
      "learning_rate": 1.994572069069204e-05,
      "loss": 1.583,
      "step": 75
    },
    {
      "epoch": 0.038598273235144746,
      "grad_norm": 1.2044777870178223,
      "learning_rate": 1.9944039282336553e-05,
      "loss": 1.4365,
      "step": 76
    },
    {
      "epoch": 0.03910614525139665,
      "grad_norm": 1.3132214546203613,
      "learning_rate": 1.99423323002784e-05,
      "loss": 1.6793,
      "step": 77
    },
    {
      "epoch": 0.03961401726764855,
      "grad_norm": 1.4409575462341309,
      "learning_rate": 1.994059974890753e-05,
      "loss": 1.6365,
      "step": 78
    },
    {
      "epoch": 0.04012188928390046,
      "grad_norm": 1.273102879524231,
      "learning_rate": 1.9938841632679647e-05,
      "loss": 1.6006,
      "step": 79
    },
    {
      "epoch": 0.04062976130015236,
      "grad_norm": 1.4772100448608398,
      "learning_rate": 1.9937057956116217e-05,
      "loss": 1.5227,
      "step": 80
    },
    {
      "epoch": 0.041137633316404264,
      "grad_norm": 1.306842565536499,
      "learning_rate": 1.993524872380443e-05,
      "loss": 1.491,
      "step": 81
    },
    {
      "epoch": 0.04164550533265617,
      "grad_norm": 1.2774075269699097,
      "learning_rate": 1.993341394039719e-05,
      "loss": 1.7854,
      "step": 82
    },
    {
      "epoch": 0.042153377348908075,
      "grad_norm": 1.4964087009429932,
      "learning_rate": 1.993155361061314e-05,
      "loss": 1.6788,
      "step": 83
    },
    {
      "epoch": 0.04266124936515998,
      "grad_norm": 1.237638235092163,
      "learning_rate": 1.9929667739236597e-05,
      "loss": 1.8024,
      "step": 84
    },
    {
      "epoch": 0.043169121381411886,
      "grad_norm": 1.7545373439788818,
      "learning_rate": 1.9927756331117575e-05,
      "loss": 1.1102,
      "step": 85
    },
    {
      "epoch": 0.04367699339766379,
      "grad_norm": 1.4063007831573486,
      "learning_rate": 1.992581939117176e-05,
      "loss": 1.2884,
      "step": 86
    },
    {
      "epoch": 0.04418486541391569,
      "grad_norm": 1.5220825672149658,
      "learning_rate": 1.9923856924380496e-05,
      "loss": 1.4794,
      "step": 87
    },
    {
      "epoch": 0.0446927374301676,
      "grad_norm": 1.3065992593765259,
      "learning_rate": 1.9921868935790788e-05,
      "loss": 1.6815,
      "step": 88
    },
    {
      "epoch": 0.0452006094464195,
      "grad_norm": 1.5807737112045288,
      "learning_rate": 1.9919855430515265e-05,
      "loss": 1.3399,
      "step": 89
    },
    {
      "epoch": 0.0457084814626714,
      "grad_norm": 1.4048736095428467,
      "learning_rate": 1.9917816413732184e-05,
      "loss": 1.7414,
      "step": 90
    },
    {
      "epoch": 0.04621635347892331,
      "grad_norm": 1.6283764839172363,
      "learning_rate": 1.9915751890685416e-05,
      "loss": 1.5164,
      "step": 91
    },
    {
      "epoch": 0.046724225495175215,
      "grad_norm": 2.085670232772827,
      "learning_rate": 1.9913661866684415e-05,
      "loss": 1.4067,
      "step": 92
    },
    {
      "epoch": 0.047232097511427124,
      "grad_norm": 1.2962095737457275,
      "learning_rate": 1.9911546347104232e-05,
      "loss": 1.3445,
      "step": 93
    },
    {
      "epoch": 0.047739969527679026,
      "grad_norm": 1.3311445713043213,
      "learning_rate": 1.9909405337385474e-05,
      "loss": 1.6662,
      "step": 94
    },
    {
      "epoch": 0.04824784154393093,
      "grad_norm": 1.7060577869415283,
      "learning_rate": 1.9907238843034317e-05,
      "loss": 1.5718,
      "step": 95
    },
    {
      "epoch": 0.04875571356018284,
      "grad_norm": 1.2818933725357056,
      "learning_rate": 1.990504686962246e-05,
      "loss": 1.6056,
      "step": 96
    },
    {
      "epoch": 0.04926358557643474,
      "grad_norm": 1.5512213706970215,
      "learning_rate": 1.990282942278714e-05,
      "loss": 1.6164,
      "step": 97
    },
    {
      "epoch": 0.04977145759268664,
      "grad_norm": 1.6890562772750854,
      "learning_rate": 1.9900586508231107e-05,
      "loss": 1.4841,
      "step": 98
    },
    {
      "epoch": 0.05027932960893855,
      "grad_norm": 1.4586176872253418,
      "learning_rate": 1.9898318131722597e-05,
      "loss": 1.5043,
      "step": 99
    },
    {
      "epoch": 0.05078720162519045,
      "grad_norm": 1.9947463274002075,
      "learning_rate": 1.9896024299095336e-05,
      "loss": 1.11,
      "step": 100
    },
    {
      "epoch": 0.051295073641442354,
      "grad_norm": 1.809464454650879,
      "learning_rate": 1.989370501624852e-05,
      "loss": 1.4424,
      "step": 101
    },
    {
      "epoch": 0.05180294565769426,
      "grad_norm": 1.4731370210647583,
      "learning_rate": 1.9891360289146787e-05,
      "loss": 1.4353,
      "step": 102
    },
    {
      "epoch": 0.052310817673946165,
      "grad_norm": 1.6137124300003052,
      "learning_rate": 1.9888990123820223e-05,
      "loss": 1.4225,
      "step": 103
    },
    {
      "epoch": 0.05281868969019807,
      "grad_norm": 1.5897164344787598,
      "learning_rate": 1.9886594526364323e-05,
      "loss": 1.5393,
      "step": 104
    },
    {
      "epoch": 0.053326561706449976,
      "grad_norm": 1.5969938039779663,
      "learning_rate": 1.988417350294e-05,
      "loss": 1.6938,
      "step": 105
    },
    {
      "epoch": 0.05383443372270188,
      "grad_norm": 1.7524147033691406,
      "learning_rate": 1.988172705977354e-05,
      "loss": 1.4503,
      "step": 106
    },
    {
      "epoch": 0.05434230573895378,
      "grad_norm": 1.5974395275115967,
      "learning_rate": 1.9879255203156626e-05,
      "loss": 1.6654,
      "step": 107
    },
    {
      "epoch": 0.05485017775520569,
      "grad_norm": 1.6734551191329956,
      "learning_rate": 1.9876757939446273e-05,
      "loss": 1.7293,
      "step": 108
    },
    {
      "epoch": 0.05535804977145759,
      "grad_norm": 1.6386557817459106,
      "learning_rate": 1.9874235275064855e-05,
      "loss": 1.6286,
      "step": 109
    },
    {
      "epoch": 0.055865921787709494,
      "grad_norm": 1.3887147903442383,
      "learning_rate": 1.9871687216500065e-05,
      "loss": 1.5914,
      "step": 110
    },
    {
      "epoch": 0.0563737938039614,
      "grad_norm": 1.4445892572402954,
      "learning_rate": 1.9869113770304903e-05,
      "loss": 1.8198,
      "step": 111
    },
    {
      "epoch": 0.056881665820213305,
      "grad_norm": 1.891767978668213,
      "learning_rate": 1.986651494309766e-05,
      "loss": 1.5657,
      "step": 112
    },
    {
      "epoch": 0.057389537836465214,
      "grad_norm": 1.4021755456924438,
      "learning_rate": 1.9863890741561896e-05,
      "loss": 1.5877,
      "step": 113
    },
    {
      "epoch": 0.057897409852717116,
      "grad_norm": 1.3860752582550049,
      "learning_rate": 1.9861241172446443e-05,
      "loss": 1.5665,
      "step": 114
    },
    {
      "epoch": 0.05840528186896902,
      "grad_norm": 1.810243844985962,
      "learning_rate": 1.985856624256535e-05,
      "loss": 1.6717,
      "step": 115
    },
    {
      "epoch": 0.05891315388522093,
      "grad_norm": 1.4407758712768555,
      "learning_rate": 1.985586595879791e-05,
      "loss": 1.6235,
      "step": 116
    },
    {
      "epoch": 0.05942102590147283,
      "grad_norm": 1.3299332857131958,
      "learning_rate": 1.9853140328088608e-05,
      "loss": 1.4443,
      "step": 117
    },
    {
      "epoch": 0.05992889791772473,
      "grad_norm": 1.984378457069397,
      "learning_rate": 1.9850389357447116e-05,
      "loss": 1.3835,
      "step": 118
    },
    {
      "epoch": 0.06043676993397664,
      "grad_norm": 1.7486463785171509,
      "learning_rate": 1.9847613053948272e-05,
      "loss": 1.6556,
      "step": 119
    },
    {
      "epoch": 0.06094464195022854,
      "grad_norm": 1.381593108177185,
      "learning_rate": 1.9844811424732078e-05,
      "loss": 1.6322,
      "step": 120
    },
    {
      "epoch": 0.061452513966480445,
      "grad_norm": 1.4813175201416016,
      "learning_rate": 1.9841984477003647e-05,
      "loss": 1.3805,
      "step": 121
    },
    {
      "epoch": 0.061960385982732354,
      "grad_norm": 1.7013801336288452,
      "learning_rate": 1.9839132218033225e-05,
      "loss": 1.4291,
      "step": 122
    },
    {
      "epoch": 0.062468257998984256,
      "grad_norm": 1.7752875089645386,
      "learning_rate": 1.9836254655156135e-05,
      "loss": 1.4994,
      "step": 123
    },
    {
      "epoch": 0.06297613001523616,
      "grad_norm": 1.5406975746154785,
      "learning_rate": 1.9833351795772788e-05,
      "loss": 1.6347,
      "step": 124
    },
    {
      "epoch": 0.06348400203148806,
      "grad_norm": 2.0284907817840576,
      "learning_rate": 1.9830423647348652e-05,
      "loss": 1.4502,
      "step": 125
    },
    {
      "epoch": 0.06399187404773997,
      "grad_norm": 1.535901665687561,
      "learning_rate": 1.982747021741422e-05,
      "loss": 1.5141,
      "step": 126
    },
    {
      "epoch": 0.06449974606399188,
      "grad_norm": 1.566285252571106,
      "learning_rate": 1.9824491513565016e-05,
      "loss": 1.2966,
      "step": 127
    },
    {
      "epoch": 0.06500761808024377,
      "grad_norm": 1.9784098863601685,
      "learning_rate": 1.9821487543461554e-05,
      "loss": 1.5775,
      "step": 128
    },
    {
      "epoch": 0.06551549009649568,
      "grad_norm": 1.4289242029190063,
      "learning_rate": 1.981845831482933e-05,
      "loss": 1.4593,
      "step": 129
    },
    {
      "epoch": 0.06602336211274759,
      "grad_norm": 1.6669222116470337,
      "learning_rate": 1.9815403835458803e-05,
      "loss": 1.691,
      "step": 130
    },
    {
      "epoch": 0.06653123412899949,
      "grad_norm": 1.4953184127807617,
      "learning_rate": 1.9812324113205366e-05,
      "loss": 1.5924,
      "step": 131
    },
    {
      "epoch": 0.0670391061452514,
      "grad_norm": 2.38492488861084,
      "learning_rate": 1.9809219155989335e-05,
      "loss": 1.5344,
      "step": 132
    },
    {
      "epoch": 0.0675469781615033,
      "grad_norm": 2.0180890560150146,
      "learning_rate": 1.9806088971795914e-05,
      "loss": 1.5879,
      "step": 133
    },
    {
      "epoch": 0.0680548501777552,
      "grad_norm": 1.7884663343429565,
      "learning_rate": 1.9802933568675193e-05,
      "loss": 1.4413,
      "step": 134
    },
    {
      "epoch": 0.06856272219400711,
      "grad_norm": 1.3341087102890015,
      "learning_rate": 1.9799752954742123e-05,
      "loss": 1.329,
      "step": 135
    },
    {
      "epoch": 0.06907059421025902,
      "grad_norm": 2.230578660964966,
      "learning_rate": 1.979654713817648e-05,
      "loss": 1.4909,
      "step": 136
    },
    {
      "epoch": 0.06957846622651091,
      "grad_norm": 1.7144185304641724,
      "learning_rate": 1.9793316127222864e-05,
      "loss": 1.2556,
      "step": 137
    },
    {
      "epoch": 0.07008633824276282,
      "grad_norm": 1.8278745412826538,
      "learning_rate": 1.979005993019067e-05,
      "loss": 1.4866,
      "step": 138
    },
    {
      "epoch": 0.07059421025901473,
      "grad_norm": 1.563427448272705,
      "learning_rate": 1.978677855545406e-05,
      "loss": 1.5526,
      "step": 139
    },
    {
      "epoch": 0.07110208227526663,
      "grad_norm": 1.5219600200653076,
      "learning_rate": 1.9783472011451944e-05,
      "loss": 1.6047,
      "step": 140
    },
    {
      "epoch": 0.07160995429151853,
      "grad_norm": 1.6114145517349243,
      "learning_rate": 1.978014030668797e-05,
      "loss": 1.6691,
      "step": 141
    },
    {
      "epoch": 0.07211782630777044,
      "grad_norm": 2.0263187885284424,
      "learning_rate": 1.9776783449730494e-05,
      "loss": 1.5557,
      "step": 142
    },
    {
      "epoch": 0.07262569832402235,
      "grad_norm": 1.488948106765747,
      "learning_rate": 1.9773401449212545e-05,
      "loss": 1.6008,
      "step": 143
    },
    {
      "epoch": 0.07313357034027425,
      "grad_norm": 1.7363940477371216,
      "learning_rate": 1.976999431383183e-05,
      "loss": 1.5921,
      "step": 144
    },
    {
      "epoch": 0.07364144235652616,
      "grad_norm": 1.658381700515747,
      "learning_rate": 1.9766562052350682e-05,
      "loss": 1.7221,
      "step": 145
    },
    {
      "epoch": 0.07414931437277807,
      "grad_norm": 1.9282320737838745,
      "learning_rate": 1.9763104673596075e-05,
      "loss": 1.4233,
      "step": 146
    },
    {
      "epoch": 0.07465718638902996,
      "grad_norm": 2.092620372772217,
      "learning_rate": 1.9759622186459548e-05,
      "loss": 1.3735,
      "step": 147
    },
    {
      "epoch": 0.07516505840528187,
      "grad_norm": 1.7551285028457642,
      "learning_rate": 1.975611459989724e-05,
      "loss": 1.4086,
      "step": 148
    },
    {
      "epoch": 0.07567293042153378,
      "grad_norm": 1.4977078437805176,
      "learning_rate": 1.9752581922929833e-05,
      "loss": 1.5749,
      "step": 149
    },
    {
      "epoch": 0.07618080243778567,
      "grad_norm": 1.57418692111969,
      "learning_rate": 1.974902416464252e-05,
      "loss": 1.4769,
      "step": 150
    },
    {
      "epoch": 0.07668867445403758,
      "grad_norm": 1.671919584274292,
      "learning_rate": 1.9745441334185012e-05,
      "loss": 1.5281,
      "step": 151
    },
    {
      "epoch": 0.07719654647028949,
      "grad_norm": 1.7778983116149902,
      "learning_rate": 1.9741833440771495e-05,
      "loss": 1.6097,
      "step": 152
    },
    {
      "epoch": 0.07770441848654139,
      "grad_norm": 1.651626706123352,
      "learning_rate": 1.973820049368062e-05,
      "loss": 1.4676,
      "step": 153
    },
    {
      "epoch": 0.0782122905027933,
      "grad_norm": 1.7673014402389526,
      "learning_rate": 1.9734542502255457e-05,
      "loss": 1.4988,
      "step": 154
    },
    {
      "epoch": 0.0787201625190452,
      "grad_norm": 2.4674925804138184,
      "learning_rate": 1.973085947590349e-05,
      "loss": 1.4203,
      "step": 155
    },
    {
      "epoch": 0.0792280345352971,
      "grad_norm": 1.5197545289993286,
      "learning_rate": 1.9727151424096585e-05,
      "loss": 1.6221,
      "step": 156
    },
    {
      "epoch": 0.07973590655154901,
      "grad_norm": 2.3471405506134033,
      "learning_rate": 1.972341835637097e-05,
      "loss": 1.3436,
      "step": 157
    },
    {
      "epoch": 0.08024377856780092,
      "grad_norm": 2.414863348007202,
      "learning_rate": 1.9719660282327205e-05,
      "loss": 1.5306,
      "step": 158
    },
    {
      "epoch": 0.08075165058405281,
      "grad_norm": 2.0253164768218994,
      "learning_rate": 1.9715877211630168e-05,
      "loss": 1.1846,
      "step": 159
    },
    {
      "epoch": 0.08125952260030472,
      "grad_norm": 2.0572662353515625,
      "learning_rate": 1.971206915400901e-05,
      "loss": 1.7391,
      "step": 160
    },
    {
      "epoch": 0.08176739461655663,
      "grad_norm": 1.8957135677337646,
      "learning_rate": 1.9708236119257154e-05,
      "loss": 1.3965,
      "step": 161
    },
    {
      "epoch": 0.08227526663280853,
      "grad_norm": 1.7855340242385864,
      "learning_rate": 1.9704378117232247e-05,
      "loss": 1.3442,
      "step": 162
    },
    {
      "epoch": 0.08278313864906044,
      "grad_norm": 2.447350263595581,
      "learning_rate": 1.9700495157856162e-05,
      "loss": 1.3476,
      "step": 163
    },
    {
      "epoch": 0.08329101066531235,
      "grad_norm": 2.002796173095703,
      "learning_rate": 1.9696587251114936e-05,
      "loss": 1.6351,
      "step": 164
    },
    {
      "epoch": 0.08379888268156424,
      "grad_norm": 1.765633225440979,
      "learning_rate": 1.969265440705878e-05,
      "loss": 1.3999,
      "step": 165
    },
    {
      "epoch": 0.08430675469781615,
      "grad_norm": 1.6512019634246826,
      "learning_rate": 1.9688696635802035e-05,
      "loss": 1.5612,
      "step": 166
    },
    {
      "epoch": 0.08481462671406806,
      "grad_norm": 2.272329330444336,
      "learning_rate": 1.968471394752314e-05,
      "loss": 1.286,
      "step": 167
    },
    {
      "epoch": 0.08532249873031995,
      "grad_norm": 2.384737730026245,
      "learning_rate": 1.9680706352464632e-05,
      "loss": 1.299,
      "step": 168
    },
    {
      "epoch": 0.08583037074657186,
      "grad_norm": 0.558256208896637,
      "learning_rate": 1.9676673860933083e-05,
      "loss": 0.2597,
      "step": 169
    },
    {
      "epoch": 0.08633824276282377,
      "grad_norm": 2.099900007247925,
      "learning_rate": 1.9672616483299106e-05,
      "loss": 1.3598,
      "step": 170
    },
    {
      "epoch": 0.08684611477907567,
      "grad_norm": 2.465440273284912,
      "learning_rate": 1.9668534229997305e-05,
      "loss": 1.201,
      "step": 171
    },
    {
      "epoch": 0.08735398679532758,
      "grad_norm": 1.5782811641693115,
      "learning_rate": 1.966442711152627e-05,
      "loss": 1.3735,
      "step": 172
    },
    {
      "epoch": 0.08786185881157949,
      "grad_norm": 1.8597896099090576,
      "learning_rate": 1.9660295138448533e-05,
      "loss": 1.4313,
      "step": 173
    },
    {
      "epoch": 0.08836973082783138,
      "grad_norm": 1.765465259552002,
      "learning_rate": 1.965613832139054e-05,
      "loss": 1.4627,
      "step": 174
    },
    {
      "epoch": 0.08887760284408329,
      "grad_norm": 1.6649190187454224,
      "learning_rate": 1.9651956671042642e-05,
      "loss": 1.633,
      "step": 175
    },
    {
      "epoch": 0.0893854748603352,
      "grad_norm": 1.8893892765045166,
      "learning_rate": 1.9647750198159043e-05,
      "loss": 1.597,
      "step": 176
    },
    {
      "epoch": 0.0898933468765871,
      "grad_norm": 1.6628423929214478,
      "learning_rate": 1.9643518913557792e-05,
      "loss": 1.6589,
      "step": 177
    },
    {
      "epoch": 0.090401218892839,
      "grad_norm": 1.9656569957733154,
      "learning_rate": 1.9639262828120748e-05,
      "loss": 1.436,
      "step": 178
    },
    {
      "epoch": 0.09090909090909091,
      "grad_norm": 1.9969520568847656,
      "learning_rate": 1.9634981952793544e-05,
      "loss": 1.4815,
      "step": 179
    },
    {
      "epoch": 0.0914169629253428,
      "grad_norm": 2.0888116359710693,
      "learning_rate": 1.963067629858558e-05,
      "loss": 1.4511,
      "step": 180
    },
    {
      "epoch": 0.09192483494159472,
      "grad_norm": 1.9178920984268188,
      "learning_rate": 1.9626345876569974e-05,
      "loss": 1.367,
      "step": 181
    },
    {
      "epoch": 0.09243270695784662,
      "grad_norm": 2.327199935913086,
      "learning_rate": 1.9621990697883536e-05,
      "loss": 1.3266,
      "step": 182
    },
    {
      "epoch": 0.09294057897409853,
      "grad_norm": 1.8885021209716797,
      "learning_rate": 1.9617610773726746e-05,
      "loss": 1.3464,
      "step": 183
    },
    {
      "epoch": 0.09344845099035043,
      "grad_norm": 2.1692769527435303,
      "learning_rate": 1.9613206115363736e-05,
      "loss": 1.5941,
      "step": 184
    },
    {
      "epoch": 0.09395632300660234,
      "grad_norm": 1.5790897607803345,
      "learning_rate": 1.9608776734122234e-05,
      "loss": 1.5025,
      "step": 185
    },
    {
      "epoch": 0.09446419502285425,
      "grad_norm": 2.0684800148010254,
      "learning_rate": 1.9604322641393556e-05,
      "loss": 1.5111,
      "step": 186
    },
    {
      "epoch": 0.09497206703910614,
      "grad_norm": 2.5110013484954834,
      "learning_rate": 1.9599843848632568e-05,
      "loss": 1.4123,
      "step": 187
    },
    {
      "epoch": 0.09547993905535805,
      "grad_norm": 1.8878841400146484,
      "learning_rate": 1.959534036735766e-05,
      "loss": 1.4055,
      "step": 188
    },
    {
      "epoch": 0.09598781107160996,
      "grad_norm": 1.6657261848449707,
      "learning_rate": 1.959081220915071e-05,
      "loss": 1.7038,
      "step": 189
    },
    {
      "epoch": 0.09649568308786186,
      "grad_norm": 2.0334601402282715,
      "learning_rate": 1.9586259385657066e-05,
      "loss": 1.5934,
      "step": 190
    },
    {
      "epoch": 0.09700355510411376,
      "grad_norm": 1.8854690790176392,
      "learning_rate": 1.958168190858551e-05,
      "loss": 1.7543,
      "step": 191
    },
    {
      "epoch": 0.09751142712036567,
      "grad_norm": 1.7778549194335938,
      "learning_rate": 1.957707978970822e-05,
      "loss": 1.4673,
      "step": 192
    },
    {
      "epoch": 0.09801929913661757,
      "grad_norm": 1.6139943599700928,
      "learning_rate": 1.957245304086075e-05,
      "loss": 1.5247,
      "step": 193
    },
    {
      "epoch": 0.09852717115286948,
      "grad_norm": 2.0140955448150635,
      "learning_rate": 1.9567801673942e-05,
      "loss": 1.6386,
      "step": 194
    },
    {
      "epoch": 0.09903504316912139,
      "grad_norm": 1.8387404680252075,
      "learning_rate": 1.9563125700914178e-05,
      "loss": 1.5482,
      "step": 195
    },
    {
      "epoch": 0.09954291518537328,
      "grad_norm": 2.418991804122925,
      "learning_rate": 1.9558425133802773e-05,
      "loss": 1.272,
      "step": 196
    },
    {
      "epoch": 0.10005078720162519,
      "grad_norm": 1.6561681032180786,
      "learning_rate": 1.9553699984696526e-05,
      "loss": 1.588,
      "step": 197
    },
    {
      "epoch": 0.1005586592178771,
      "grad_norm": 1.572954773902893,
      "learning_rate": 1.954895026574739e-05,
      "loss": 1.8015,
      "step": 198
    },
    {
      "epoch": 0.101066531234129,
      "grad_norm": 1.933001160621643,
      "learning_rate": 1.9544175989170524e-05,
      "loss": 1.4857,
      "step": 199
    },
    {
      "epoch": 0.1015744032503809,
      "grad_norm": 1.5697592496871948,
      "learning_rate": 1.953937716724422e-05,
      "loss": 1.5252,
      "step": 200
    },
    {
      "epoch": 0.10208227526663281,
      "grad_norm": 2.5573041439056396,
      "learning_rate": 1.9534553812309915e-05,
      "loss": 1.4089,
      "step": 201
    },
    {
      "epoch": 0.10259014728288471,
      "grad_norm": 2.043156623840332,
      "learning_rate": 1.952970593677212e-05,
      "loss": 1.5563,
      "step": 202
    },
    {
      "epoch": 0.10309801929913662,
      "grad_norm": 2.2919042110443115,
      "learning_rate": 1.9524833553098427e-05,
      "loss": 1.3823,
      "step": 203
    },
    {
      "epoch": 0.10360589131538853,
      "grad_norm": 1.8231582641601562,
      "learning_rate": 1.951993667381944e-05,
      "loss": 1.6335,
      "step": 204
    },
    {
      "epoch": 0.10411376333164042,
      "grad_norm": 2.7078065872192383,
      "learning_rate": 1.951501531152877e-05,
      "loss": 1.3149,
      "step": 205
    },
    {
      "epoch": 0.10462163534789233,
      "grad_norm": 1.694219946861267,
      "learning_rate": 1.9510069478882992e-05,
      "loss": 1.5816,
      "step": 206
    },
    {
      "epoch": 0.10512950736414424,
      "grad_norm": 1.9638980627059937,
      "learning_rate": 1.9505099188601608e-05,
      "loss": 1.4232,
      "step": 207
    },
    {
      "epoch": 0.10563737938039613,
      "grad_norm": 1.720871925354004,
      "learning_rate": 1.950010445346702e-05,
      "loss": 1.5198,
      "step": 208
    },
    {
      "epoch": 0.10614525139664804,
      "grad_norm": 1.6361560821533203,
      "learning_rate": 1.9495085286324507e-05,
      "loss": 1.5962,
      "step": 209
    },
    {
      "epoch": 0.10665312341289995,
      "grad_norm": 2.454713821411133,
      "learning_rate": 1.949004170008216e-05,
      "loss": 1.1473,
      "step": 210
    },
    {
      "epoch": 0.10716099542915185,
      "grad_norm": 1.7801949977874756,
      "learning_rate": 1.948497370771089e-05,
      "loss": 1.2549,
      "step": 211
    },
    {
      "epoch": 0.10766886744540376,
      "grad_norm": 2.275209665298462,
      "learning_rate": 1.947988132224437e-05,
      "loss": 1.4693,
      "step": 212
    },
    {
      "epoch": 0.10817673946165567,
      "grad_norm": 2.096883535385132,
      "learning_rate": 1.9474764556778995e-05,
      "loss": 1.6335,
      "step": 213
    },
    {
      "epoch": 0.10868461147790756,
      "grad_norm": 1.7155479192733765,
      "learning_rate": 1.9469623424473872e-05,
      "loss": 1.3781,
      "step": 214
    },
    {
      "epoch": 0.10919248349415947,
      "grad_norm": 2.3470253944396973,
      "learning_rate": 1.9464457938550777e-05,
      "loss": 1.3552,
      "step": 215
    },
    {
      "epoch": 0.10970035551041138,
      "grad_norm": 1.6862317323684692,
      "learning_rate": 1.94592681122941e-05,
      "loss": 1.5208,
      "step": 216
    },
    {
      "epoch": 0.11020822752666327,
      "grad_norm": 1.914600133895874,
      "learning_rate": 1.945405395905084e-05,
      "loss": 1.5037,
      "step": 217
    },
    {
      "epoch": 0.11071609954291518,
      "grad_norm": 1.9791793823242188,
      "learning_rate": 1.9448815492230565e-05,
      "loss": 1.4713,
      "step": 218
    },
    {
      "epoch": 0.11122397155916709,
      "grad_norm": 2.2346582412719727,
      "learning_rate": 1.9443552725305365e-05,
      "loss": 1.4726,
      "step": 219
    },
    {
      "epoch": 0.11173184357541899,
      "grad_norm": 2.368903398513794,
      "learning_rate": 1.9438265671809814e-05,
      "loss": 1.0064,
      "step": 220
    },
    {
      "epoch": 0.1122397155916709,
      "grad_norm": 2.2253165245056152,
      "learning_rate": 1.943295434534097e-05,
      "loss": 1.5413,
      "step": 221
    },
    {
      "epoch": 0.1127475876079228,
      "grad_norm": 2.02980637550354,
      "learning_rate": 1.942761875955829e-05,
      "loss": 1.3832,
      "step": 222
    },
    {
      "epoch": 0.11325545962417471,
      "grad_norm": 1.889204502105713,
      "learning_rate": 1.942225892818363e-05,
      "loss": 1.4722,
      "step": 223
    },
    {
      "epoch": 0.11376333164042661,
      "grad_norm": 2.5287232398986816,
      "learning_rate": 1.941687486500121e-05,
      "loss": 1.3209,
      "step": 224
    },
    {
      "epoch": 0.11427120365667852,
      "grad_norm": 1.8791983127593994,
      "learning_rate": 1.9411466583857553e-05,
      "loss": 1.4135,
      "step": 225
    },
    {
      "epoch": 0.11477907567293043,
      "grad_norm": 2.2831013202667236,
      "learning_rate": 1.940603409866147e-05,
      "loss": 1.3759,
      "step": 226
    },
    {
      "epoch": 0.11528694768918232,
      "grad_norm": 2.0638115406036377,
      "learning_rate": 1.9400577423384024e-05,
      "loss": 1.567,
      "step": 227
    },
    {
      "epoch": 0.11579481970543423,
      "grad_norm": 1.8206024169921875,
      "learning_rate": 1.9395096572058482e-05,
      "loss": 1.3574,
      "step": 228
    },
    {
      "epoch": 0.11630269172168614,
      "grad_norm": 2.0053048133850098,
      "learning_rate": 1.9389591558780293e-05,
      "loss": 1.59,
      "step": 229
    },
    {
      "epoch": 0.11681056373793804,
      "grad_norm": 1.729561686515808,
      "learning_rate": 1.938406239770704e-05,
      "loss": 1.4971,
      "step": 230
    },
    {
      "epoch": 0.11731843575418995,
      "grad_norm": 2.176643133163452,
      "learning_rate": 1.9378509103058406e-05,
      "loss": 1.5178,
      "step": 231
    },
    {
      "epoch": 0.11782630777044185,
      "grad_norm": 1.8067771196365356,
      "learning_rate": 1.9372931689116147e-05,
      "loss": 1.5915,
      "step": 232
    },
    {
      "epoch": 0.11833417978669375,
      "grad_norm": 2.685448408126831,
      "learning_rate": 1.9367330170224043e-05,
      "loss": 1.0551,
      "step": 233
    },
    {
      "epoch": 0.11884205180294566,
      "grad_norm": 1.656010389328003,
      "learning_rate": 1.9361704560787872e-05,
      "loss": 1.4525,
      "step": 234
    },
    {
      "epoch": 0.11934992381919757,
      "grad_norm": 1.7234470844268799,
      "learning_rate": 1.9356054875275353e-05,
      "loss": 1.393,
      "step": 235
    },
    {
      "epoch": 0.11985779583544946,
      "grad_norm": 2.0601320266723633,
      "learning_rate": 1.935038112821614e-05,
      "loss": 1.375,
      "step": 236
    },
    {
      "epoch": 0.12036566785170137,
      "grad_norm": 1.6750483512878418,
      "learning_rate": 1.9344683334201764e-05,
      "loss": 1.7454,
      "step": 237
    },
    {
      "epoch": 0.12087353986795328,
      "grad_norm": 1.990843415260315,
      "learning_rate": 1.9338961507885586e-05,
      "loss": 1.5386,
      "step": 238
    },
    {
      "epoch": 0.12138141188420518,
      "grad_norm": 2.328029155731201,
      "learning_rate": 1.9333215663982786e-05,
      "loss": 1.3595,
      "step": 239
    },
    {
      "epoch": 0.12188928390045708,
      "grad_norm": 2.0774049758911133,
      "learning_rate": 1.9327445817270304e-05,
      "loss": 1.366,
      "step": 240
    },
    {
      "epoch": 0.122397155916709,
      "grad_norm": 2.1023411750793457,
      "learning_rate": 1.932165198258682e-05,
      "loss": 1.306,
      "step": 241
    },
    {
      "epoch": 0.12290502793296089,
      "grad_norm": 1.935680866241455,
      "learning_rate": 1.931583417483269e-05,
      "loss": 1.8575,
      "step": 242
    },
    {
      "epoch": 0.1234128999492128,
      "grad_norm": 1.8602055311203003,
      "learning_rate": 1.9309992408969938e-05,
      "loss": 1.3252,
      "step": 243
    },
    {
      "epoch": 0.12392077196546471,
      "grad_norm": 2.098102569580078,
      "learning_rate": 1.9304126700022188e-05,
      "loss": 1.5078,
      "step": 244
    },
    {
      "epoch": 0.1244286439817166,
      "grad_norm": 2.305022954940796,
      "learning_rate": 1.929823706307466e-05,
      "loss": 1.5595,
      "step": 245
    },
    {
      "epoch": 0.12493651599796851,
      "grad_norm": 1.9000664949417114,
      "learning_rate": 1.9292323513274083e-05,
      "loss": 1.4804,
      "step": 246
    },
    {
      "epoch": 0.12544438801422042,
      "grad_norm": 2.0502431392669678,
      "learning_rate": 1.928638606582871e-05,
      "loss": 1.4233,
      "step": 247
    },
    {
      "epoch": 0.12595226003047233,
      "grad_norm": 1.6554498672485352,
      "learning_rate": 1.928042473600824e-05,
      "loss": 1.56,
      "step": 248
    },
    {
      "epoch": 0.12646013204672424,
      "grad_norm": 1.7347345352172852,
      "learning_rate": 1.9274439539143796e-05,
      "loss": 1.6861,
      "step": 249
    },
    {
      "epoch": 0.12696800406297612,
      "grad_norm": 1.698886513710022,
      "learning_rate": 1.9268430490627882e-05,
      "loss": 1.4357,
      "step": 250
    },
    {
      "epoch": 0.12747587607922803,
      "grad_norm": 1.5664055347442627,
      "learning_rate": 1.9262397605914335e-05,
      "loss": 1.5929,
      "step": 251
    },
    {
      "epoch": 0.12798374809547994,
      "grad_norm": 2.1273763179779053,
      "learning_rate": 1.9256340900518297e-05,
      "loss": 1.6916,
      "step": 252
    },
    {
      "epoch": 0.12849162011173185,
      "grad_norm": 1.813781499862671,
      "learning_rate": 1.9250260390016183e-05,
      "loss": 1.6156,
      "step": 253
    },
    {
      "epoch": 0.12899949212798376,
      "grad_norm": 2.629873275756836,
      "learning_rate": 1.9244156090045614e-05,
      "loss": 1.3784,
      "step": 254
    },
    {
      "epoch": 0.12950736414423566,
      "grad_norm": 1.9181764125823975,
      "learning_rate": 1.9238028016305395e-05,
      "loss": 1.5728,
      "step": 255
    },
    {
      "epoch": 0.13001523616048755,
      "grad_norm": 1.8989523649215698,
      "learning_rate": 1.9231876184555478e-05,
      "loss": 1.4443,
      "step": 256
    },
    {
      "epoch": 0.13052310817673946,
      "grad_norm": 2.246098279953003,
      "learning_rate": 1.9225700610616905e-05,
      "loss": 1.373,
      "step": 257
    },
    {
      "epoch": 0.13103098019299136,
      "grad_norm": 1.9531599283218384,
      "learning_rate": 1.9219501310371785e-05,
      "loss": 1.4015,
      "step": 258
    },
    {
      "epoch": 0.13153885220924327,
      "grad_norm": 1.9696428775787354,
      "learning_rate": 1.9213278299763247e-05,
      "loss": 1.4289,
      "step": 259
    },
    {
      "epoch": 0.13204672422549518,
      "grad_norm": 2.4111690521240234,
      "learning_rate": 1.920703159479538e-05,
      "loss": 1.3672,
      "step": 260
    },
    {
      "epoch": 0.1325545962417471,
      "grad_norm": 1.8648861646652222,
      "learning_rate": 1.9200761211533242e-05,
      "loss": 1.4389,
      "step": 261
    },
    {
      "epoch": 0.13306246825799897,
      "grad_norm": 1.9896775484085083,
      "learning_rate": 1.919446716610275e-05,
      "loss": 1.5265,
      "step": 262
    },
    {
      "epoch": 0.13357034027425088,
      "grad_norm": 1.900830626487732,
      "learning_rate": 1.9188149474690702e-05,
      "loss": 1.4751,
      "step": 263
    },
    {
      "epoch": 0.1340782122905028,
      "grad_norm": 2.4136266708374023,
      "learning_rate": 1.9181808153544693e-05,
      "loss": 1.2784,
      "step": 264
    },
    {
      "epoch": 0.1345860843067547,
      "grad_norm": 2.2382752895355225,
      "learning_rate": 1.9175443218973088e-05,
      "loss": 1.3597,
      "step": 265
    },
    {
      "epoch": 0.1350939563230066,
      "grad_norm": 2.2490243911743164,
      "learning_rate": 1.916905468734499e-05,
      "loss": 1.4279,
      "step": 266
    },
    {
      "epoch": 0.13560182833925852,
      "grad_norm": 2.242929458618164,
      "learning_rate": 1.9162642575090177e-05,
      "loss": 1.4268,
      "step": 267
    },
    {
      "epoch": 0.1361097003555104,
      "grad_norm": 2.0219061374664307,
      "learning_rate": 1.915620689869908e-05,
      "loss": 1.5397,
      "step": 268
    },
    {
      "epoch": 0.1366175723717623,
      "grad_norm": 2.5027315616607666,
      "learning_rate": 1.9149747674722728e-05,
      "loss": 1.3594,
      "step": 269
    },
    {
      "epoch": 0.13712544438801422,
      "grad_norm": 2.541565418243408,
      "learning_rate": 1.9143264919772706e-05,
      "loss": 1.6437,
      "step": 270
    },
    {
      "epoch": 0.13763331640426613,
      "grad_norm": 1.9443773031234741,
      "learning_rate": 1.9136758650521115e-05,
      "loss": 1.4999,
      "step": 271
    },
    {
      "epoch": 0.13814118842051804,
      "grad_norm": 2.0174036026000977,
      "learning_rate": 1.913022888370053e-05,
      "loss": 1.4428,
      "step": 272
    },
    {
      "epoch": 0.13864906043676994,
      "grad_norm": 2.9639198780059814,
      "learning_rate": 1.9123675636103965e-05,
      "loss": 1.0563,
      "step": 273
    },
    {
      "epoch": 0.13915693245302183,
      "grad_norm": 2.349069118499756,
      "learning_rate": 1.911709892458481e-05,
      "loss": 1.4261,
      "step": 274
    },
    {
      "epoch": 0.13966480446927373,
      "grad_norm": 2.796888828277588,
      "learning_rate": 1.9110498766056795e-05,
      "loss": 1.3594,
      "step": 275
    },
    {
      "epoch": 0.14017267648552564,
      "grad_norm": 1.9367467164993286,
      "learning_rate": 1.9103875177493965e-05,
      "loss": 1.2106,
      "step": 276
    },
    {
      "epoch": 0.14068054850177755,
      "grad_norm": 2.081439256668091,
      "learning_rate": 1.9097228175930614e-05,
      "loss": 1.7384,
      "step": 277
    },
    {
      "epoch": 0.14118842051802946,
      "grad_norm": 2.059976100921631,
      "learning_rate": 1.909055777846124e-05,
      "loss": 1.6941,
      "step": 278
    },
    {
      "epoch": 0.14169629253428137,
      "grad_norm": 2.0120351314544678,
      "learning_rate": 1.9083864002240528e-05,
      "loss": 1.6482,
      "step": 279
    },
    {
      "epoch": 0.14220416455053325,
      "grad_norm": 2.1116793155670166,
      "learning_rate": 1.9077146864483276e-05,
      "loss": 1.4704,
      "step": 280
    },
    {
      "epoch": 0.14271203656678516,
      "grad_norm": 1.8414626121520996,
      "learning_rate": 1.9070406382464357e-05,
      "loss": 1.5199,
      "step": 281
    },
    {
      "epoch": 0.14321990858303707,
      "grad_norm": 2.3146474361419678,
      "learning_rate": 1.906364257351869e-05,
      "loss": 1.3412,
      "step": 282
    },
    {
      "epoch": 0.14372778059928898,
      "grad_norm": 3.1995961666107178,
      "learning_rate": 1.9056855455041184e-05,
      "loss": 1.1945,
      "step": 283
    },
    {
      "epoch": 0.1442356526155409,
      "grad_norm": 2.307894229888916,
      "learning_rate": 1.905004504448669e-05,
      "loss": 1.4024,
      "step": 284
    },
    {
      "epoch": 0.1447435246317928,
      "grad_norm": 2.1073455810546875,
      "learning_rate": 1.9043211359369968e-05,
      "loss": 1.4765,
      "step": 285
    },
    {
      "epoch": 0.1452513966480447,
      "grad_norm": 2.1424057483673096,
      "learning_rate": 1.9036354417265625e-05,
      "loss": 1.4475,
      "step": 286
    },
    {
      "epoch": 0.1457592686642966,
      "grad_norm": 2.280087471008301,
      "learning_rate": 1.9029474235808084e-05,
      "loss": 1.4765,
      "step": 287
    },
    {
      "epoch": 0.1462671406805485,
      "grad_norm": 2.5047733783721924,
      "learning_rate": 1.902257083269154e-05,
      "loss": 1.4661,
      "step": 288
    },
    {
      "epoch": 0.1467750126968004,
      "grad_norm": 2.0760483741760254,
      "learning_rate": 1.9015644225669898e-05,
      "loss": 1.3863,
      "step": 289
    },
    {
      "epoch": 0.14728288471305231,
      "grad_norm": 2.574784755706787,
      "learning_rate": 1.900869443255674e-05,
      "loss": 1.3514,
      "step": 290
    },
    {
      "epoch": 0.14779075672930422,
      "grad_norm": 1.9647001028060913,
      "learning_rate": 1.9001721471225288e-05,
      "loss": 1.6358,
      "step": 291
    },
    {
      "epoch": 0.14829862874555613,
      "grad_norm": 2.3125619888305664,
      "learning_rate": 1.8994725359608337e-05,
      "loss": 1.2977,
      "step": 292
    },
    {
      "epoch": 0.148806500761808,
      "grad_norm": 2.3946354389190674,
      "learning_rate": 1.898770611569822e-05,
      "loss": 1.1799,
      "step": 293
    },
    {
      "epoch": 0.14931437277805992,
      "grad_norm": 2.557532548904419,
      "learning_rate": 1.898066375754676e-05,
      "loss": 1.2864,
      "step": 294
    },
    {
      "epoch": 0.14982224479431183,
      "grad_norm": 1.8328592777252197,
      "learning_rate": 1.8973598303265227e-05,
      "loss": 1.5309,
      "step": 295
    },
    {
      "epoch": 0.15033011681056374,
      "grad_norm": 2.9495561122894287,
      "learning_rate": 1.896650977102429e-05,
      "loss": 1.3253,
      "step": 296
    },
    {
      "epoch": 0.15083798882681565,
      "grad_norm": 2.62872576713562,
      "learning_rate": 1.8959398179053967e-05,
      "loss": 1.5211,
      "step": 297
    },
    {
      "epoch": 0.15134586084306756,
      "grad_norm": 2.3727867603302,
      "learning_rate": 1.895226354564358e-05,
      "loss": 1.3479,
      "step": 298
    },
    {
      "epoch": 0.15185373285931944,
      "grad_norm": 1.9571844339370728,
      "learning_rate": 1.894510588914171e-05,
      "loss": 1.4394,
      "step": 299
    },
    {
      "epoch": 0.15236160487557135,
      "grad_norm": 2.380986213684082,
      "learning_rate": 1.8937925227956138e-05,
      "loss": 1.4961,
      "step": 300
    },
    {
      "epoch": 0.15286947689182326,
      "grad_norm": 2.1453568935394287,
      "learning_rate": 1.8930721580553823e-05,
      "loss": 1.4865,
      "step": 301
    },
    {
      "epoch": 0.15337734890807517,
      "grad_norm": 2.148232936859131,
      "learning_rate": 1.8923494965460835e-05,
      "loss": 1.4999,
      "step": 302
    },
    {
      "epoch": 0.15388522092432708,
      "grad_norm": 2.053626298904419,
      "learning_rate": 1.8916245401262302e-05,
      "loss": 1.4948,
      "step": 303
    },
    {
      "epoch": 0.15439309294057899,
      "grad_norm": 2.15143084526062,
      "learning_rate": 1.8908972906602377e-05,
      "loss": 1.3438,
      "step": 304
    },
    {
      "epoch": 0.15490096495683087,
      "grad_norm": 1.9328835010528564,
      "learning_rate": 1.8901677500184193e-05,
      "loss": 1.6481,
      "step": 305
    },
    {
      "epoch": 0.15540883697308278,
      "grad_norm": 2.4750730991363525,
      "learning_rate": 1.8894359200769795e-05,
      "loss": 1.414,
      "step": 306
    },
    {
      "epoch": 0.15591670898933468,
      "grad_norm": 2.187335968017578,
      "learning_rate": 1.88870180271801e-05,
      "loss": 1.6663,
      "step": 307
    },
    {
      "epoch": 0.1564245810055866,
      "grad_norm": 1.9440557956695557,
      "learning_rate": 1.8879653998294874e-05,
      "loss": 1.574,
      "step": 308
    },
    {
      "epoch": 0.1569324530218385,
      "grad_norm": 2.0348010063171387,
      "learning_rate": 1.887226713305264e-05,
      "loss": 1.5302,
      "step": 309
    },
    {
      "epoch": 0.1574403250380904,
      "grad_norm": 2.1762094497680664,
      "learning_rate": 1.8864857450450656e-05,
      "loss": 1.3841,
      "step": 310
    },
    {
      "epoch": 0.1579481970543423,
      "grad_norm": 2.0718870162963867,
      "learning_rate": 1.885742496954486e-05,
      "loss": 1.4259,
      "step": 311
    },
    {
      "epoch": 0.1584560690705942,
      "grad_norm": 2.1066360473632812,
      "learning_rate": 1.8849969709449832e-05,
      "loss": 1.4205,
      "step": 312
    },
    {
      "epoch": 0.1589639410868461,
      "grad_norm": 2.104362726211548,
      "learning_rate": 1.8842491689338723e-05,
      "loss": 1.4979,
      "step": 313
    },
    {
      "epoch": 0.15947181310309802,
      "grad_norm": 2.0534019470214844,
      "learning_rate": 1.8834990928443223e-05,
      "loss": 1.4079,
      "step": 314
    },
    {
      "epoch": 0.15997968511934993,
      "grad_norm": 2.407684087753296,
      "learning_rate": 1.8827467446053495e-05,
      "loss": 1.2353,
      "step": 315
    },
    {
      "epoch": 0.16048755713560184,
      "grad_norm": 1.9434267282485962,
      "learning_rate": 1.881992126151816e-05,
      "loss": 1.4704,
      "step": 316
    },
    {
      "epoch": 0.16099542915185372,
      "grad_norm": 2.3935511112213135,
      "learning_rate": 1.88123523942442e-05,
      "loss": 1.5185,
      "step": 317
    },
    {
      "epoch": 0.16150330116810563,
      "grad_norm": 2.342310667037964,
      "learning_rate": 1.8804760863696935e-05,
      "loss": 1.4252,
      "step": 318
    },
    {
      "epoch": 0.16201117318435754,
      "grad_norm": 2.0927910804748535,
      "learning_rate": 1.8797146689399986e-05,
      "loss": 1.4787,
      "step": 319
    },
    {
      "epoch": 0.16251904520060945,
      "grad_norm": 2.240511417388916,
      "learning_rate": 1.8789509890935195e-05,
      "loss": 1.4495,
      "step": 320
    },
    {
      "epoch": 0.16302691721686136,
      "grad_norm": 2.1083669662475586,
      "learning_rate": 1.878185048794259e-05,
      "loss": 1.3573,
      "step": 321
    },
    {
      "epoch": 0.16353478923311326,
      "grad_norm": 2.706357002258301,
      "learning_rate": 1.8774168500120327e-05,
      "loss": 1.1445,
      "step": 322
    },
    {
      "epoch": 0.16404266124936517,
      "grad_norm": 2.1849541664123535,
      "learning_rate": 1.8766463947224656e-05,
      "loss": 1.5005,
      "step": 323
    },
    {
      "epoch": 0.16455053326561705,
      "grad_norm": 1.9791027307510376,
      "learning_rate": 1.875873684906985e-05,
      "loss": 1.522,
      "step": 324
    },
    {
      "epoch": 0.16505840528186896,
      "grad_norm": 3.4028851985931396,
      "learning_rate": 1.8750987225528172e-05,
      "loss": 1.2043,
      "step": 325
    },
    {
      "epoch": 0.16556627729812087,
      "grad_norm": 2.6301186084747314,
      "learning_rate": 1.874321509652981e-05,
      "loss": 1.3337,
      "step": 326
    },
    {
      "epoch": 0.16607414931437278,
      "grad_norm": 2.031865119934082,
      "learning_rate": 1.8735420482062823e-05,
      "loss": 1.7132,
      "step": 327
    },
    {
      "epoch": 0.1665820213306247,
      "grad_norm": 2.090599298477173,
      "learning_rate": 1.872760340217311e-05,
      "loss": 1.577,
      "step": 328
    },
    {
      "epoch": 0.1670898933468766,
      "grad_norm": 2.234833240509033,
      "learning_rate": 1.8719763876964336e-05,
      "loss": 1.2331,
      "step": 329
    },
    {
      "epoch": 0.16759776536312848,
      "grad_norm": 2.4577765464782715,
      "learning_rate": 1.87119019265979e-05,
      "loss": 1.4527,
      "step": 330
    },
    {
      "epoch": 0.1681056373793804,
      "grad_norm": 1.89016592502594,
      "learning_rate": 1.8704017571292863e-05,
      "loss": 1.4465,
      "step": 331
    },
    {
      "epoch": 0.1686135093956323,
      "grad_norm": 2.2821614742279053,
      "learning_rate": 1.8696110831325913e-05,
      "loss": 1.3836,
      "step": 332
    },
    {
      "epoch": 0.1691213814118842,
      "grad_norm": 2.25083327293396,
      "learning_rate": 1.8688181727031302e-05,
      "loss": 1.5524,
      "step": 333
    },
    {
      "epoch": 0.16962925342813612,
      "grad_norm": 2.1324195861816406,
      "learning_rate": 1.8680230278800803e-05,
      "loss": 1.5654,
      "step": 334
    },
    {
      "epoch": 0.17013712544438803,
      "grad_norm": 1.9769184589385986,
      "learning_rate": 1.8672256507083643e-05,
      "loss": 1.5077,
      "step": 335
    },
    {
      "epoch": 0.1706449974606399,
      "grad_norm": 2.0667829513549805,
      "learning_rate": 1.8664260432386468e-05,
      "loss": 1.5829,
      "step": 336
    },
    {
      "epoch": 0.17115286947689182,
      "grad_norm": 2.8132925033569336,
      "learning_rate": 1.8656242075273285e-05,
      "loss": 1.1102,
      "step": 337
    },
    {
      "epoch": 0.17166074149314373,
      "grad_norm": 2.055262565612793,
      "learning_rate": 1.864820145636539e-05,
      "loss": 1.3644,
      "step": 338
    },
    {
      "epoch": 0.17216861350939563,
      "grad_norm": 2.576754570007324,
      "learning_rate": 1.8640138596341344e-05,
      "loss": 1.1964,
      "step": 339
    },
    {
      "epoch": 0.17267648552564754,
      "grad_norm": 2.026365280151367,
      "learning_rate": 1.8632053515936908e-05,
      "loss": 1.4661,
      "step": 340
    },
    {
      "epoch": 0.17318435754189945,
      "grad_norm": 2.220334768295288,
      "learning_rate": 1.8623946235944984e-05,
      "loss": 1.5105,
      "step": 341
    },
    {
      "epoch": 0.17369222955815133,
      "grad_norm": 1.902352213859558,
      "learning_rate": 1.8615816777215568e-05,
      "loss": 1.3642,
      "step": 342
    },
    {
      "epoch": 0.17420010157440324,
      "grad_norm": 2.278432846069336,
      "learning_rate": 1.860766516065569e-05,
      "loss": 1.2551,
      "step": 343
    },
    {
      "epoch": 0.17470797359065515,
      "grad_norm": 2.4950084686279297,
      "learning_rate": 1.859949140722937e-05,
      "loss": 1.49,
      "step": 344
    },
    {
      "epoch": 0.17521584560690706,
      "grad_norm": 2.029480218887329,
      "learning_rate": 1.859129553795756e-05,
      "loss": 1.3534,
      "step": 345
    },
    {
      "epoch": 0.17572371762315897,
      "grad_norm": 2.1549057960510254,
      "learning_rate": 1.8583077573918075e-05,
      "loss": 1.3466,
      "step": 346
    },
    {
      "epoch": 0.17623158963941088,
      "grad_norm": 2.1388282775878906,
      "learning_rate": 1.8574837536245573e-05,
      "loss": 1.6887,
      "step": 347
    },
    {
      "epoch": 0.17673946165566276,
      "grad_norm": 2.1772992610931396,
      "learning_rate": 1.8566575446131465e-05,
      "loss": 1.3554,
      "step": 348
    },
    {
      "epoch": 0.17724733367191467,
      "grad_norm": 2.1744778156280518,
      "learning_rate": 1.855829132482388e-05,
      "loss": 1.4455,
      "step": 349
    },
    {
      "epoch": 0.17775520568816658,
      "grad_norm": 2.47438383102417,
      "learning_rate": 1.8549985193627614e-05,
      "loss": 1.0442,
      "step": 350
    },
    {
      "epoch": 0.1782630777044185,
      "grad_norm": 2.6652379035949707,
      "learning_rate": 1.8541657073904056e-05,
      "loss": 1.5324,
      "step": 351
    },
    {
      "epoch": 0.1787709497206704,
      "grad_norm": 1.9979355335235596,
      "learning_rate": 1.8533306987071143e-05,
      "loss": 1.3954,
      "step": 352
    },
    {
      "epoch": 0.1792788217369223,
      "grad_norm": 2.422628164291382,
      "learning_rate": 1.8524934954603315e-05,
      "loss": 1.1543,
      "step": 353
    },
    {
      "epoch": 0.1797866937531742,
      "grad_norm": 2.561239242553711,
      "learning_rate": 1.8516540998031448e-05,
      "loss": 1.4323,
      "step": 354
    },
    {
      "epoch": 0.1802945657694261,
      "grad_norm": 2.8377954959869385,
      "learning_rate": 1.85081251389428e-05,
      "loss": 1.2745,
      "step": 355
    },
    {
      "epoch": 0.180802437785678,
      "grad_norm": 2.4836831092834473,
      "learning_rate": 1.8499687398980958e-05,
      "loss": 1.6612,
      "step": 356
    },
    {
      "epoch": 0.18131030980192991,
      "grad_norm": 2.547248601913452,
      "learning_rate": 1.849122779984578e-05,
      "loss": 1.1906,
      "step": 357
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 2.7639479637145996,
      "learning_rate": 1.8482746363293343e-05,
      "loss": 1.4388,
      "step": 358
    },
    {
      "epoch": 0.18232605383443373,
      "grad_norm": 2.050381898880005,
      "learning_rate": 1.8474243111135887e-05,
      "loss": 1.3234,
      "step": 359
    },
    {
      "epoch": 0.1828339258506856,
      "grad_norm": 2.310777187347412,
      "learning_rate": 1.8465718065241748e-05,
      "loss": 1.4396,
      "step": 360
    },
    {
      "epoch": 0.18334179786693752,
      "grad_norm": 2.1302127838134766,
      "learning_rate": 1.8457171247535317e-05,
      "loss": 1.3624,
      "step": 361
    },
    {
      "epoch": 0.18384966988318943,
      "grad_norm": 2.225740671157837,
      "learning_rate": 1.8448602679996973e-05,
      "loss": 1.4544,
      "step": 362
    },
    {
      "epoch": 0.18435754189944134,
      "grad_norm": 2.29659104347229,
      "learning_rate": 1.8440012384663038e-05,
      "loss": 1.4831,
      "step": 363
    },
    {
      "epoch": 0.18486541391569325,
      "grad_norm": 2.4151954650878906,
      "learning_rate": 1.8431400383625705e-05,
      "loss": 1.6177,
      "step": 364
    },
    {
      "epoch": 0.18537328593194516,
      "grad_norm": 2.5019686222076416,
      "learning_rate": 1.8422766699032987e-05,
      "loss": 1.6529,
      "step": 365
    },
    {
      "epoch": 0.18588115794819707,
      "grad_norm": 2.6340701580047607,
      "learning_rate": 1.8414111353088673e-05,
      "loss": 1.4929,
      "step": 366
    },
    {
      "epoch": 0.18638902996444895,
      "grad_norm": 2.0728466510772705,
      "learning_rate": 1.840543436805225e-05,
      "loss": 1.2841,
      "step": 367
    },
    {
      "epoch": 0.18689690198070086,
      "grad_norm": 2.3754184246063232,
      "learning_rate": 1.8396735766238857e-05,
      "loss": 1.5152,
      "step": 368
    },
    {
      "epoch": 0.18740477399695277,
      "grad_norm": 2.1963157653808594,
      "learning_rate": 1.838801557001923e-05,
      "loss": 1.5506,
      "step": 369
    },
    {
      "epoch": 0.18791264601320468,
      "grad_norm": 2.961599588394165,
      "learning_rate": 1.8379273801819638e-05,
      "loss": 1.4706,
      "step": 370
    },
    {
      "epoch": 0.18842051802945659,
      "grad_norm": 2.4010231494903564,
      "learning_rate": 1.8370510484121827e-05,
      "loss": 1.5459,
      "step": 371
    },
    {
      "epoch": 0.1889283900457085,
      "grad_norm": 2.4878487586975098,
      "learning_rate": 1.836172563946297e-05,
      "loss": 1.3344,
      "step": 372
    },
    {
      "epoch": 0.18943626206196038,
      "grad_norm": 2.9341530799865723,
      "learning_rate": 1.835291929043559e-05,
      "loss": 1.1486,
      "step": 373
    },
    {
      "epoch": 0.18994413407821228,
      "grad_norm": 2.265747308731079,
      "learning_rate": 1.8344091459687523e-05,
      "loss": 1.4896,
      "step": 374
    },
    {
      "epoch": 0.1904520060944642,
      "grad_norm": 2.425539016723633,
      "learning_rate": 1.8335242169921855e-05,
      "loss": 1.34,
      "step": 375
    },
    {
      "epoch": 0.1909598781107161,
      "grad_norm": 2.2222049236297607,
      "learning_rate": 1.8326371443896844e-05,
      "loss": 1.3314,
      "step": 376
    },
    {
      "epoch": 0.191467750126968,
      "grad_norm": 2.7927303314208984,
      "learning_rate": 1.8317479304425892e-05,
      "loss": 1.5547,
      "step": 377
    },
    {
      "epoch": 0.19197562214321992,
      "grad_norm": 2.29571795463562,
      "learning_rate": 1.8308565774377467e-05,
      "loss": 1.6539,
      "step": 378
    },
    {
      "epoch": 0.1924834941594718,
      "grad_norm": 2.2622365951538086,
      "learning_rate": 1.8299630876675044e-05,
      "loss": 1.0428,
      "step": 379
    },
    {
      "epoch": 0.1929913661757237,
      "grad_norm": 2.2128021717071533,
      "learning_rate": 1.8290674634297058e-05,
      "loss": 1.7003,
      "step": 380
    },
    {
      "epoch": 0.19349923819197562,
      "grad_norm": 2.8428139686584473,
      "learning_rate": 1.8281697070276823e-05,
      "loss": 1.3185,
      "step": 381
    },
    {
      "epoch": 0.19400711020822753,
      "grad_norm": 2.158491849899292,
      "learning_rate": 1.8272698207702507e-05,
      "loss": 1.5285,
      "step": 382
    },
    {
      "epoch": 0.19451498222447944,
      "grad_norm": 2.125875949859619,
      "learning_rate": 1.8263678069717038e-05,
      "loss": 1.3319,
      "step": 383
    },
    {
      "epoch": 0.19502285424073135,
      "grad_norm": 2.0204598903656006,
      "learning_rate": 1.825463667951807e-05,
      "loss": 1.5469,
      "step": 384
    },
    {
      "epoch": 0.19553072625698323,
      "grad_norm": 2.247217893600464,
      "learning_rate": 1.82455740603579e-05,
      "loss": 1.7199,
      "step": 385
    },
    {
      "epoch": 0.19603859827323514,
      "grad_norm": 2.4135122299194336,
      "learning_rate": 1.8236490235543434e-05,
      "loss": 1.2538,
      "step": 386
    },
    {
      "epoch": 0.19654647028948705,
      "grad_norm": 2.0119616985321045,
      "learning_rate": 1.8227385228436105e-05,
      "loss": 1.6421,
      "step": 387
    },
    {
      "epoch": 0.19705434230573896,
      "grad_norm": 2.754758596420288,
      "learning_rate": 1.8218259062451823e-05,
      "loss": 1.4222,
      "step": 388
    },
    {
      "epoch": 0.19756221432199086,
      "grad_norm": 3.040740728378296,
      "learning_rate": 1.8209111761060917e-05,
      "loss": 0.8824,
      "step": 389
    },
    {
      "epoch": 0.19807008633824277,
      "grad_norm": 2.2685205936431885,
      "learning_rate": 1.8199943347788068e-05,
      "loss": 1.4771,
      "step": 390
    },
    {
      "epoch": 0.19857795835449465,
      "grad_norm": 2.4334073066711426,
      "learning_rate": 1.819075384621226e-05,
      "loss": 1.3698,
      "step": 391
    },
    {
      "epoch": 0.19908583037074656,
      "grad_norm": 2.461564302444458,
      "learning_rate": 1.8181543279966693e-05,
      "loss": 1.3328,
      "step": 392
    },
    {
      "epoch": 0.19959370238699847,
      "grad_norm": 2.8636112213134766,
      "learning_rate": 1.817231167273876e-05,
      "loss": 1.3873,
      "step": 393
    },
    {
      "epoch": 0.20010157440325038,
      "grad_norm": 2.322828769683838,
      "learning_rate": 1.8163059048269955e-05,
      "loss": 1.5203,
      "step": 394
    },
    {
      "epoch": 0.2006094464195023,
      "grad_norm": 2.6736671924591064,
      "learning_rate": 1.8153785430355822e-05,
      "loss": 1.3255,
      "step": 395
    },
    {
      "epoch": 0.2011173184357542,
      "grad_norm": 2.41998553276062,
      "learning_rate": 1.814449084284591e-05,
      "loss": 1.5725,
      "step": 396
    },
    {
      "epoch": 0.20162519045200608,
      "grad_norm": 2.028114080429077,
      "learning_rate": 1.813517530964368e-05,
      "loss": 1.5688,
      "step": 397
    },
    {
      "epoch": 0.202133062468258,
      "grad_norm": 3.2604825496673584,
      "learning_rate": 1.8125838854706462e-05,
      "loss": 1.3424,
      "step": 398
    },
    {
      "epoch": 0.2026409344845099,
      "grad_norm": 2.9950945377349854,
      "learning_rate": 1.8116481502045403e-05,
      "loss": 1.1741,
      "step": 399
    },
    {
      "epoch": 0.2031488065007618,
      "grad_norm": 3.022077798843384,
      "learning_rate": 1.8107103275725385e-05,
      "loss": 1.4031,
      "step": 400
    },
    {
      "epoch": 0.20365667851701372,
      "grad_norm": 2.1941399574279785,
      "learning_rate": 1.8097704199864975e-05,
      "loss": 1.5143,
      "step": 401
    },
    {
      "epoch": 0.20416455053326563,
      "grad_norm": 2.6506152153015137,
      "learning_rate": 1.808828429863636e-05,
      "loss": 1.253,
      "step": 402
    },
    {
      "epoch": 0.20467242254951754,
      "grad_norm": 2.522321939468384,
      "learning_rate": 1.8078843596265284e-05,
      "loss": 1.3241,
      "step": 403
    },
    {
      "epoch": 0.20518029456576942,
      "grad_norm": 1.9919332265853882,
      "learning_rate": 1.8069382117030986e-05,
      "loss": 1.4723,
      "step": 404
    },
    {
      "epoch": 0.20568816658202133,
      "grad_norm": 2.273698329925537,
      "learning_rate": 1.8059899885266138e-05,
      "loss": 1.5205,
      "step": 405
    },
    {
      "epoch": 0.20619603859827323,
      "grad_norm": 2.337557792663574,
      "learning_rate": 1.8050396925356784e-05,
      "loss": 1.5634,
      "step": 406
    },
    {
      "epoch": 0.20670391061452514,
      "grad_norm": 2.289808750152588,
      "learning_rate": 1.8040873261742283e-05,
      "loss": 1.4679,
      "step": 407
    },
    {
      "epoch": 0.20721178263077705,
      "grad_norm": 2.309823989868164,
      "learning_rate": 1.8031328918915217e-05,
      "loss": 1.4815,
      "step": 408
    },
    {
      "epoch": 0.20771965464702896,
      "grad_norm": 2.8153891563415527,
      "learning_rate": 1.802176392142138e-05,
      "loss": 1.015,
      "step": 409
    },
    {
      "epoch": 0.20822752666328084,
      "grad_norm": 2.379129409790039,
      "learning_rate": 1.8012178293859656e-05,
      "loss": 1.5316,
      "step": 410
    },
    {
      "epoch": 0.20873539867953275,
      "grad_norm": 2.3453006744384766,
      "learning_rate": 1.8002572060882007e-05,
      "loss": 1.3609,
      "step": 411
    },
    {
      "epoch": 0.20924327069578466,
      "grad_norm": 2.95465350151062,
      "learning_rate": 1.7992945247193377e-05,
      "loss": 1.5909,
      "step": 412
    },
    {
      "epoch": 0.20975114271203657,
      "grad_norm": 2.9034945964813232,
      "learning_rate": 1.7983297877551632e-05,
      "loss": 1.3234,
      "step": 413
    },
    {
      "epoch": 0.21025901472828848,
      "grad_norm": 1.9625252485275269,
      "learning_rate": 1.797362997676752e-05,
      "loss": 1.6168,
      "step": 414
    },
    {
      "epoch": 0.2107668867445404,
      "grad_norm": 2.2009334564208984,
      "learning_rate": 1.796394156970458e-05,
      "loss": 1.5079,
      "step": 415
    },
    {
      "epoch": 0.21127475876079227,
      "grad_norm": 3.1377625465393066,
      "learning_rate": 1.7954232681279088e-05,
      "loss": 1.0786,
      "step": 416
    },
    {
      "epoch": 0.21178263077704418,
      "grad_norm": 2.8287923336029053,
      "learning_rate": 1.794450333645999e-05,
      "loss": 1.4505,
      "step": 417
    },
    {
      "epoch": 0.2122905027932961,
      "grad_norm": 2.2335076332092285,
      "learning_rate": 1.7934753560268857e-05,
      "loss": 1.6506,
      "step": 418
    },
    {
      "epoch": 0.212798374809548,
      "grad_norm": 2.296365737915039,
      "learning_rate": 1.792498337777978e-05,
      "loss": 1.3601,
      "step": 419
    },
    {
      "epoch": 0.2133062468257999,
      "grad_norm": 2.6674273014068604,
      "learning_rate": 1.7915192814119353e-05,
      "loss": 1.4654,
      "step": 420
    },
    {
      "epoch": 0.21381411884205181,
      "grad_norm": 2.8880250453948975,
      "learning_rate": 1.790538189446657e-05,
      "loss": 1.3702,
      "step": 421
    },
    {
      "epoch": 0.2143219908583037,
      "grad_norm": 2.449025869369507,
      "learning_rate": 1.7895550644052786e-05,
      "loss": 1.4645,
      "step": 422
    },
    {
      "epoch": 0.2148298628745556,
      "grad_norm": 1.928270697593689,
      "learning_rate": 1.7885699088161633e-05,
      "loss": 1.4804,
      "step": 423
    },
    {
      "epoch": 0.21533773489080751,
      "grad_norm": 2.3959450721740723,
      "learning_rate": 1.7875827252128975e-05,
      "loss": 1.2716,
      "step": 424
    },
    {
      "epoch": 0.21584560690705942,
      "grad_norm": 2.897495985031128,
      "learning_rate": 1.786593516134282e-05,
      "loss": 1.6,
      "step": 425
    },
    {
      "epoch": 0.21635347892331133,
      "grad_norm": 2.6640288829803467,
      "learning_rate": 1.7856022841243267e-05,
      "loss": 1.4547,
      "step": 426
    },
    {
      "epoch": 0.21686135093956324,
      "grad_norm": 2.1698009967803955,
      "learning_rate": 1.7846090317322457e-05,
      "loss": 1.3702,
      "step": 427
    },
    {
      "epoch": 0.21736922295581512,
      "grad_norm": 2.8702659606933594,
      "learning_rate": 1.7836137615124466e-05,
      "loss": 1.2881,
      "step": 428
    },
    {
      "epoch": 0.21787709497206703,
      "grad_norm": 2.8904287815093994,
      "learning_rate": 1.7826164760245284e-05,
      "loss": 1.5613,
      "step": 429
    },
    {
      "epoch": 0.21838496698831894,
      "grad_norm": 2.9009711742401123,
      "learning_rate": 1.7816171778332715e-05,
      "loss": 1.3943,
      "step": 430
    },
    {
      "epoch": 0.21889283900457085,
      "grad_norm": 2.75502872467041,
      "learning_rate": 1.7806158695086333e-05,
      "loss": 1.4311,
      "step": 431
    },
    {
      "epoch": 0.21940071102082276,
      "grad_norm": 2.89510178565979,
      "learning_rate": 1.7796125536257408e-05,
      "loss": 1.3384,
      "step": 432
    },
    {
      "epoch": 0.21990858303707467,
      "grad_norm": 2.7613089084625244,
      "learning_rate": 1.778607232764883e-05,
      "loss": 1.734,
      "step": 433
    },
    {
      "epoch": 0.22041645505332655,
      "grad_norm": 1.9892759323120117,
      "learning_rate": 1.7775999095115066e-05,
      "loss": 1.4728,
      "step": 434
    },
    {
      "epoch": 0.22092432706957846,
      "grad_norm": 2.4320971965789795,
      "learning_rate": 1.7765905864562066e-05,
      "loss": 1.4656,
      "step": 435
    },
    {
      "epoch": 0.22143219908583037,
      "grad_norm": 2.669489622116089,
      "learning_rate": 1.7755792661947224e-05,
      "loss": 1.2149,
      "step": 436
    },
    {
      "epoch": 0.22194007110208228,
      "grad_norm": 2.6985087394714355,
      "learning_rate": 1.774565951327929e-05,
      "loss": 1.3109,
      "step": 437
    },
    {
      "epoch": 0.22244794311833418,
      "grad_norm": 2.665987014770508,
      "learning_rate": 1.7735506444618303e-05,
      "loss": 1.4917,
      "step": 438
    },
    {
      "epoch": 0.2229558151345861,
      "grad_norm": 2.214789628982544,
      "learning_rate": 1.7725333482075546e-05,
      "loss": 1.6784,
      "step": 439
    },
    {
      "epoch": 0.22346368715083798,
      "grad_norm": 2.9302868843078613,
      "learning_rate": 1.771514065181346e-05,
      "loss": 1.2476,
      "step": 440
    },
    {
      "epoch": 0.22397155916708988,
      "grad_norm": 2.3863744735717773,
      "learning_rate": 1.7704927980045578e-05,
      "loss": 1.3971,
      "step": 441
    },
    {
      "epoch": 0.2244794311833418,
      "grad_norm": 2.391859769821167,
      "learning_rate": 1.7694695493036454e-05,
      "loss": 1.3805,
      "step": 442
    },
    {
      "epoch": 0.2249873031995937,
      "grad_norm": 2.776182174682617,
      "learning_rate": 1.7684443217101616e-05,
      "loss": 1.3091,
      "step": 443
    },
    {
      "epoch": 0.2254951752158456,
      "grad_norm": 2.4615769386291504,
      "learning_rate": 1.7674171178607476e-05,
      "loss": 1.457,
      "step": 444
    },
    {
      "epoch": 0.22600304723209752,
      "grad_norm": 2.2038397789001465,
      "learning_rate": 1.7663879403971274e-05,
      "loss": 1.4441,
      "step": 445
    },
    {
      "epoch": 0.22651091924834943,
      "grad_norm": 2.2213821411132812,
      "learning_rate": 1.7653567919661004e-05,
      "loss": 1.3748,
      "step": 446
    },
    {
      "epoch": 0.2270187912646013,
      "grad_norm": 2.3839704990386963,
      "learning_rate": 1.7643236752195357e-05,
      "loss": 1.6147,
      "step": 447
    },
    {
      "epoch": 0.22752666328085322,
      "grad_norm": 2.652604341506958,
      "learning_rate": 1.7632885928143624e-05,
      "loss": 1.5667,
      "step": 448
    },
    {
      "epoch": 0.22803453529710513,
      "grad_norm": 2.5281620025634766,
      "learning_rate": 1.762251547412567e-05,
      "loss": 1.2654,
      "step": 449
    },
    {
      "epoch": 0.22854240731335704,
      "grad_norm": 2.3414971828460693,
      "learning_rate": 1.761212541681183e-05,
      "loss": 1.4149,
      "step": 450
    },
    {
      "epoch": 0.22905027932960895,
      "grad_norm": 3.3338890075683594,
      "learning_rate": 1.760171578292287e-05,
      "loss": 1.5189,
      "step": 451
    },
    {
      "epoch": 0.22955815134586086,
      "grad_norm": 2.4341635704040527,
      "learning_rate": 1.7591286599229874e-05,
      "loss": 1.5695,
      "step": 452
    },
    {
      "epoch": 0.23006602336211274,
      "grad_norm": 2.1505444049835205,
      "learning_rate": 1.7580837892554233e-05,
      "loss": 1.3496,
      "step": 453
    },
    {
      "epoch": 0.23057389537836465,
      "grad_norm": 2.65849232673645,
      "learning_rate": 1.7570369689767534e-05,
      "loss": 1.4188,
      "step": 454
    },
    {
      "epoch": 0.23108176739461656,
      "grad_norm": 2.5335099697113037,
      "learning_rate": 1.75598820177915e-05,
      "loss": 1.5386,
      "step": 455
    },
    {
      "epoch": 0.23158963941086846,
      "grad_norm": 2.5862152576446533,
      "learning_rate": 1.7549374903597923e-05,
      "loss": 1.5935,
      "step": 456
    },
    {
      "epoch": 0.23209751142712037,
      "grad_norm": 2.326003074645996,
      "learning_rate": 1.7538848374208606e-05,
      "loss": 1.3387,
      "step": 457
    },
    {
      "epoch": 0.23260538344337228,
      "grad_norm": 2.2396774291992188,
      "learning_rate": 1.7528302456695278e-05,
      "loss": 1.565,
      "step": 458
    },
    {
      "epoch": 0.23311325545962416,
      "grad_norm": 3.605147361755371,
      "learning_rate": 1.751773717817953e-05,
      "loss": 0.9666,
      "step": 459
    },
    {
      "epoch": 0.23362112747587607,
      "grad_norm": 2.740504503250122,
      "learning_rate": 1.750715256583274e-05,
      "loss": 1.4752,
      "step": 460
    },
    {
      "epoch": 0.23412899949212798,
      "grad_norm": 3.1214797496795654,
      "learning_rate": 1.7496548646876018e-05,
      "loss": 1.1117,
      "step": 461
    },
    {
      "epoch": 0.2346368715083799,
      "grad_norm": 2.327601432800293,
      "learning_rate": 1.748592544858011e-05,
      "loss": 1.2953,
      "step": 462
    },
    {
      "epoch": 0.2351447435246318,
      "grad_norm": 2.6740219593048096,
      "learning_rate": 1.7475282998265367e-05,
      "loss": 1.462,
      "step": 463
    },
    {
      "epoch": 0.2356526155408837,
      "grad_norm": 2.490809202194214,
      "learning_rate": 1.746462132330163e-05,
      "loss": 1.234,
      "step": 464
    },
    {
      "epoch": 0.2361604875571356,
      "grad_norm": 3.0028162002563477,
      "learning_rate": 1.745394045110819e-05,
      "loss": 1.0599,
      "step": 465
    },
    {
      "epoch": 0.2366683595733875,
      "grad_norm": 2.8784072399139404,
      "learning_rate": 1.7443240409153714e-05,
      "loss": 1.188,
      "step": 466
    },
    {
      "epoch": 0.2371762315896394,
      "grad_norm": 2.4966988563537598,
      "learning_rate": 1.743252122495616e-05,
      "loss": 1.5535,
      "step": 467
    },
    {
      "epoch": 0.23768410360589132,
      "grad_norm": 2.3101346492767334,
      "learning_rate": 1.7421782926082714e-05,
      "loss": 1.4142,
      "step": 468
    },
    {
      "epoch": 0.23819197562214323,
      "grad_norm": 2.4662463665008545,
      "learning_rate": 1.7411025540149737e-05,
      "loss": 1.4697,
      "step": 469
    },
    {
      "epoch": 0.23869984763839514,
      "grad_norm": 2.2677195072174072,
      "learning_rate": 1.7400249094822656e-05,
      "loss": 1.3498,
      "step": 470
    },
    {
      "epoch": 0.23920771965464702,
      "grad_norm": 2.796475410461426,
      "learning_rate": 1.7389453617815926e-05,
      "loss": 1.6972,
      "step": 471
    },
    {
      "epoch": 0.23971559167089893,
      "grad_norm": 2.44164776802063,
      "learning_rate": 1.737863913689295e-05,
      "loss": 1.5431,
      "step": 472
    },
    {
      "epoch": 0.24022346368715083,
      "grad_norm": 3.1222140789031982,
      "learning_rate": 1.736780567986599e-05,
      "loss": 1.1978,
      "step": 473
    },
    {
      "epoch": 0.24073133570340274,
      "grad_norm": 3.480477809906006,
      "learning_rate": 1.735695327459613e-05,
      "loss": 1.0457,
      "step": 474
    },
    {
      "epoch": 0.24123920771965465,
      "grad_norm": 2.646369695663452,
      "learning_rate": 1.734608194899317e-05,
      "loss": 1.2979,
      "step": 475
    },
    {
      "epoch": 0.24174707973590656,
      "grad_norm": 2.220792531967163,
      "learning_rate": 1.7335191731015564e-05,
      "loss": 1.4697,
      "step": 476
    },
    {
      "epoch": 0.24225495175215844,
      "grad_norm": 2.121004819869995,
      "learning_rate": 1.7324282648670374e-05,
      "loss": 1.5301,
      "step": 477
    },
    {
      "epoch": 0.24276282376841035,
      "grad_norm": 2.2976834774017334,
      "learning_rate": 1.731335473001316e-05,
      "loss": 1.4953,
      "step": 478
    },
    {
      "epoch": 0.24327069578466226,
      "grad_norm": 2.562124252319336,
      "learning_rate": 1.730240800314792e-05,
      "loss": 1.2582,
      "step": 479
    },
    {
      "epoch": 0.24377856780091417,
      "grad_norm": 2.4853553771972656,
      "learning_rate": 1.7291442496227042e-05,
      "loss": 1.3838,
      "step": 480
    },
    {
      "epoch": 0.24428643981716608,
      "grad_norm": 2.6778106689453125,
      "learning_rate": 1.7280458237451192e-05,
      "loss": 1.5141,
      "step": 481
    },
    {
      "epoch": 0.244794311833418,
      "grad_norm": 2.7512123584747314,
      "learning_rate": 1.7269455255069276e-05,
      "loss": 1.342,
      "step": 482
    },
    {
      "epoch": 0.2453021838496699,
      "grad_norm": 2.8781282901763916,
      "learning_rate": 1.7258433577378342e-05,
      "loss": 1.3965,
      "step": 483
    },
    {
      "epoch": 0.24581005586592178,
      "grad_norm": 2.9807615280151367,
      "learning_rate": 1.7247393232723523e-05,
      "loss": 1.5302,
      "step": 484
    },
    {
      "epoch": 0.2463179278821737,
      "grad_norm": 3.537425994873047,
      "learning_rate": 1.723633424949796e-05,
      "loss": 1.3961,
      "step": 485
    },
    {
      "epoch": 0.2468257998984256,
      "grad_norm": 3.2156057357788086,
      "learning_rate": 1.722525665614272e-05,
      "loss": 1.4332,
      "step": 486
    },
    {
      "epoch": 0.2473336719146775,
      "grad_norm": 2.931434392929077,
      "learning_rate": 1.7214160481146748e-05,
      "loss": 1.5141,
      "step": 487
    },
    {
      "epoch": 0.24784154393092941,
      "grad_norm": 2.7063584327697754,
      "learning_rate": 1.7203045753046755e-05,
      "loss": 1.4324,
      "step": 488
    },
    {
      "epoch": 0.24834941594718132,
      "grad_norm": 2.5695347785949707,
      "learning_rate": 1.7191912500427184e-05,
      "loss": 1.4655,
      "step": 489
    },
    {
      "epoch": 0.2488572879634332,
      "grad_norm": 2.729949951171875,
      "learning_rate": 1.7180760751920106e-05,
      "loss": 1.2106,
      "step": 490
    },
    {
      "epoch": 0.2493651599796851,
      "grad_norm": 2.229206085205078,
      "learning_rate": 1.7169590536205166e-05,
      "loss": 1.698,
      "step": 491
    },
    {
      "epoch": 0.24987303199593702,
      "grad_norm": 2.490645408630371,
      "learning_rate": 1.7158401882009507e-05,
      "loss": 1.5399,
      "step": 492
    },
    {
      "epoch": 0.25038090401218893,
      "grad_norm": 2.817081928253174,
      "learning_rate": 1.7147194818107674e-05,
      "loss": 1.2956,
      "step": 493
    },
    {
      "epoch": 0.25038090401218893,
      "eval_loss": 1.272118330001831,
      "eval_runtime": 169.0072,
      "eval_samples_per_second": 5.822,
      "eval_steps_per_second": 1.456,
      "step": 493
    },
    {
      "epoch": 0.25088877602844084,
      "grad_norm": 2.6739213466644287,
      "learning_rate": 1.713596937332158e-05,
      "loss": 1.2104,
      "step": 494
    },
    {
      "epoch": 0.25139664804469275,
      "grad_norm": 2.5604422092437744,
      "learning_rate": 1.7124725576520386e-05,
      "loss": 1.4294,
      "step": 495
    },
    {
      "epoch": 0.25190452006094466,
      "grad_norm": 3.073133707046509,
      "learning_rate": 1.711346345662047e-05,
      "loss": 1.3513,
      "step": 496
    },
    {
      "epoch": 0.25241239207719657,
      "grad_norm": 2.7818005084991455,
      "learning_rate": 1.710218304258532e-05,
      "loss": 1.3464,
      "step": 497
    },
    {
      "epoch": 0.2529202640934485,
      "grad_norm": 2.26126766204834,
      "learning_rate": 1.7090884363425482e-05,
      "loss": 1.4703,
      "step": 498
    },
    {
      "epoch": 0.25342813610970033,
      "grad_norm": 2.490053415298462,
      "learning_rate": 1.7079567448198463e-05,
      "loss": 1.7689,
      "step": 499
    },
    {
      "epoch": 0.25393600812595224,
      "grad_norm": 2.396192789077759,
      "learning_rate": 1.706823232600868e-05,
      "loss": 1.6666,
      "step": 500
    },
    {
      "epoch": 0.25444388014220415,
      "grad_norm": 2.811347484588623,
      "learning_rate": 1.7056879026007373e-05,
      "loss": 1.3238,
      "step": 501
    },
    {
      "epoch": 0.25495175215845606,
      "grad_norm": 2.5664217472076416,
      "learning_rate": 1.7045507577392518e-05,
      "loss": 1.39,
      "step": 502
    },
    {
      "epoch": 0.25545962417470797,
      "grad_norm": 2.662076473236084,
      "learning_rate": 1.7034118009408784e-05,
      "loss": 1.3517,
      "step": 503
    },
    {
      "epoch": 0.2559674961909599,
      "grad_norm": 2.6712756156921387,
      "learning_rate": 1.7022710351347423e-05,
      "loss": 1.5732,
      "step": 504
    },
    {
      "epoch": 0.2564753682072118,
      "grad_norm": 2.152195930480957,
      "learning_rate": 1.7011284632546225e-05,
      "loss": 1.3897,
      "step": 505
    },
    {
      "epoch": 0.2569832402234637,
      "grad_norm": 2.2648797035217285,
      "learning_rate": 1.6999840882389413e-05,
      "loss": 1.4887,
      "step": 506
    },
    {
      "epoch": 0.2574911122397156,
      "grad_norm": 3.0260188579559326,
      "learning_rate": 1.698837913030759e-05,
      "loss": 1.3154,
      "step": 507
    },
    {
      "epoch": 0.2579989842559675,
      "grad_norm": 2.791656494140625,
      "learning_rate": 1.6976899405777656e-05,
      "loss": 1.3579,
      "step": 508
    },
    {
      "epoch": 0.2585068562722194,
      "grad_norm": 2.6979916095733643,
      "learning_rate": 1.6965401738322734e-05,
      "loss": 1.3458,
      "step": 509
    },
    {
      "epoch": 0.25901472828847133,
      "grad_norm": 2.2963948249816895,
      "learning_rate": 1.6953886157512084e-05,
      "loss": 1.4368,
      "step": 510
    },
    {
      "epoch": 0.2595226003047232,
      "grad_norm": 2.48846697807312,
      "learning_rate": 1.6942352692961036e-05,
      "loss": 1.3519,
      "step": 511
    },
    {
      "epoch": 0.2600304723209751,
      "grad_norm": 2.709425449371338,
      "learning_rate": 1.6930801374330924e-05,
      "loss": 1.4138,
      "step": 512
    },
    {
      "epoch": 0.260538344337227,
      "grad_norm": 2.7645037174224854,
      "learning_rate": 1.691923223132899e-05,
      "loss": 1.1771,
      "step": 513
    },
    {
      "epoch": 0.2610462163534789,
      "grad_norm": 3.7189676761627197,
      "learning_rate": 1.690764529370831e-05,
      "loss": 1.0428,
      "step": 514
    },
    {
      "epoch": 0.2615540883697308,
      "grad_norm": 3.617924928665161,
      "learning_rate": 1.689604059126774e-05,
      "loss": 0.8653,
      "step": 515
    },
    {
      "epoch": 0.26206196038598273,
      "grad_norm": 2.869488000869751,
      "learning_rate": 1.6884418153851805e-05,
      "loss": 1.5021,
      "step": 516
    },
    {
      "epoch": 0.26256983240223464,
      "grad_norm": 2.827406883239746,
      "learning_rate": 1.687277801135065e-05,
      "loss": 1.7214,
      "step": 517
    },
    {
      "epoch": 0.26307770441848655,
      "grad_norm": 2.7554526329040527,
      "learning_rate": 1.6861120193699953e-05,
      "loss": 1.5105,
      "step": 518
    },
    {
      "epoch": 0.26358557643473846,
      "grad_norm": 3.9668350219726562,
      "learning_rate": 1.6849444730880846e-05,
      "loss": 1.1446,
      "step": 519
    },
    {
      "epoch": 0.26409344845099036,
      "grad_norm": 3.158369779586792,
      "learning_rate": 1.683775165291984e-05,
      "loss": 1.0159,
      "step": 520
    },
    {
      "epoch": 0.2646013204672423,
      "grad_norm": 2.6982619762420654,
      "learning_rate": 1.6826040989888754e-05,
      "loss": 1.611,
      "step": 521
    },
    {
      "epoch": 0.2651091924834942,
      "grad_norm": 4.317055702209473,
      "learning_rate": 1.681431277190462e-05,
      "loss": 1.3542,
      "step": 522
    },
    {
      "epoch": 0.2656170644997461,
      "grad_norm": 3.3620643615722656,
      "learning_rate": 1.680256702912963e-05,
      "loss": 1.3332,
      "step": 523
    },
    {
      "epoch": 0.26612493651599795,
      "grad_norm": 4.196297645568848,
      "learning_rate": 1.6790803791771034e-05,
      "loss": 0.9123,
      "step": 524
    },
    {
      "epoch": 0.26663280853224985,
      "grad_norm": 2.9033923149108887,
      "learning_rate": 1.677902309008108e-05,
      "loss": 1.4143,
      "step": 525
    },
    {
      "epoch": 0.26714068054850176,
      "grad_norm": 2.843406915664673,
      "learning_rate": 1.6767224954356928e-05,
      "loss": 1.1294,
      "step": 526
    },
    {
      "epoch": 0.2676485525647537,
      "grad_norm": 2.732457160949707,
      "learning_rate": 1.6755409414940584e-05,
      "loss": 1.7313,
      "step": 527
    },
    {
      "epoch": 0.2681564245810056,
      "grad_norm": 2.676978826522827,
      "learning_rate": 1.6743576502218792e-05,
      "loss": 1.464,
      "step": 528
    },
    {
      "epoch": 0.2686642965972575,
      "grad_norm": 2.5006861686706543,
      "learning_rate": 1.6731726246622996e-05,
      "loss": 1.5256,
      "step": 529
    },
    {
      "epoch": 0.2691721686135094,
      "grad_norm": 2.2941715717315674,
      "learning_rate": 1.6719858678629232e-05,
      "loss": 1.3606,
      "step": 530
    },
    {
      "epoch": 0.2696800406297613,
      "grad_norm": 2.733076572418213,
      "learning_rate": 1.6707973828758055e-05,
      "loss": 1.1881,
      "step": 531
    },
    {
      "epoch": 0.2701879126460132,
      "grad_norm": 3.1281354427337646,
      "learning_rate": 1.6696071727574476e-05,
      "loss": 1.2662,
      "step": 532
    },
    {
      "epoch": 0.2706957846622651,
      "grad_norm": 2.3248023986816406,
      "learning_rate": 1.6684152405687867e-05,
      "loss": 1.6241,
      "step": 533
    },
    {
      "epoch": 0.27120365667851704,
      "grad_norm": 2.4165563583374023,
      "learning_rate": 1.667221589375189e-05,
      "loss": 1.3266,
      "step": 534
    },
    {
      "epoch": 0.27171152869476894,
      "grad_norm": 2.5224907398223877,
      "learning_rate": 1.6660262222464404e-05,
      "loss": 1.3281,
      "step": 535
    },
    {
      "epoch": 0.2722194007110208,
      "grad_norm": 2.712475061416626,
      "learning_rate": 1.6648291422567414e-05,
      "loss": 1.3519,
      "step": 536
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 2.8065361976623535,
      "learning_rate": 1.6636303524846968e-05,
      "loss": 1.5687,
      "step": 537
    },
    {
      "epoch": 0.2732351447435246,
      "grad_norm": 2.669524669647217,
      "learning_rate": 1.6624298560133087e-05,
      "loss": 1.6927,
      "step": 538
    },
    {
      "epoch": 0.2737430167597765,
      "grad_norm": 2.4854934215545654,
      "learning_rate": 1.661227655929968e-05,
      "loss": 1.4282,
      "step": 539
    },
    {
      "epoch": 0.27425088877602843,
      "grad_norm": 2.914283037185669,
      "learning_rate": 1.6600237553264474e-05,
      "loss": 1.4625,
      "step": 540
    },
    {
      "epoch": 0.27475876079228034,
      "grad_norm": 3.190369129180908,
      "learning_rate": 1.6588181572988928e-05,
      "loss": 1.373,
      "step": 541
    },
    {
      "epoch": 0.27526663280853225,
      "grad_norm": 3.2698330879211426,
      "learning_rate": 1.6576108649478153e-05,
      "loss": 1.3348,
      "step": 542
    },
    {
      "epoch": 0.27577450482478416,
      "grad_norm": 2.661933422088623,
      "learning_rate": 1.6564018813780838e-05,
      "loss": 1.443,
      "step": 543
    },
    {
      "epoch": 0.27628237684103607,
      "grad_norm": 2.999220848083496,
      "learning_rate": 1.655191209698916e-05,
      "loss": 1.416,
      "step": 544
    },
    {
      "epoch": 0.276790248857288,
      "grad_norm": 2.747131824493408,
      "learning_rate": 1.6539788530238716e-05,
      "loss": 1.4417,
      "step": 545
    },
    {
      "epoch": 0.2772981208735399,
      "grad_norm": 3.005300283432007,
      "learning_rate": 1.6527648144708436e-05,
      "loss": 1.1945,
      "step": 546
    },
    {
      "epoch": 0.2778059928897918,
      "grad_norm": 2.605773687362671,
      "learning_rate": 1.65154909716205e-05,
      "loss": 1.4557,
      "step": 547
    },
    {
      "epoch": 0.27831386490604365,
      "grad_norm": 2.42523193359375,
      "learning_rate": 1.6503317042240264e-05,
      "loss": 1.3029,
      "step": 548
    },
    {
      "epoch": 0.27882173692229556,
      "grad_norm": 2.553007125854492,
      "learning_rate": 1.6491126387876176e-05,
      "loss": 1.5922,
      "step": 549
    },
    {
      "epoch": 0.27932960893854747,
      "grad_norm": 3.1910340785980225,
      "learning_rate": 1.6478919039879705e-05,
      "loss": 1.169,
      "step": 550
    },
    {
      "epoch": 0.2798374809547994,
      "grad_norm": 3.14784574508667,
      "learning_rate": 1.6466695029645237e-05,
      "loss": 1.4954,
      "step": 551
    },
    {
      "epoch": 0.2803453529710513,
      "grad_norm": 2.7142114639282227,
      "learning_rate": 1.645445438861002e-05,
      "loss": 1.2445,
      "step": 552
    },
    {
      "epoch": 0.2808532249873032,
      "grad_norm": 3.649052619934082,
      "learning_rate": 1.644219714825407e-05,
      "loss": 1.1579,
      "step": 553
    },
    {
      "epoch": 0.2813610970035551,
      "grad_norm": 2.5138401985168457,
      "learning_rate": 1.6429923340100096e-05,
      "loss": 1.4395,
      "step": 554
    },
    {
      "epoch": 0.281868969019807,
      "grad_norm": 2.99912428855896,
      "learning_rate": 1.6417632995713407e-05,
      "loss": 1.2584,
      "step": 555
    },
    {
      "epoch": 0.2823768410360589,
      "grad_norm": 3.645402193069458,
      "learning_rate": 1.6405326146701846e-05,
      "loss": 1.2956,
      "step": 556
    },
    {
      "epoch": 0.28288471305231083,
      "grad_norm": 2.8912413120269775,
      "learning_rate": 1.63930028247157e-05,
      "loss": 1.2831,
      "step": 557
    },
    {
      "epoch": 0.28339258506856274,
      "grad_norm": 2.8437440395355225,
      "learning_rate": 1.6380663061447618e-05,
      "loss": 1.3912,
      "step": 558
    },
    {
      "epoch": 0.28390045708481465,
      "grad_norm": 3.3983263969421387,
      "learning_rate": 1.6368306888632538e-05,
      "loss": 1.3817,
      "step": 559
    },
    {
      "epoch": 0.2844083291010665,
      "grad_norm": 2.768794298171997,
      "learning_rate": 1.63559343380476e-05,
      "loss": 1.517,
      "step": 560
    },
    {
      "epoch": 0.2849162011173184,
      "grad_norm": 3.3941545486450195,
      "learning_rate": 1.634354544151205e-05,
      "loss": 0.9587,
      "step": 561
    },
    {
      "epoch": 0.2854240731335703,
      "grad_norm": 3.603623390197754,
      "learning_rate": 1.6331140230887185e-05,
      "loss": 1.0481,
      "step": 562
    },
    {
      "epoch": 0.28593194514982223,
      "grad_norm": 2.6169986724853516,
      "learning_rate": 1.6318718738076263e-05,
      "loss": 1.2115,
      "step": 563
    },
    {
      "epoch": 0.28643981716607414,
      "grad_norm": 2.8581392765045166,
      "learning_rate": 1.6306280995024394e-05,
      "loss": 1.4153,
      "step": 564
    },
    {
      "epoch": 0.28694768918232605,
      "grad_norm": 3.1200432777404785,
      "learning_rate": 1.6293827033718505e-05,
      "loss": 1.5795,
      "step": 565
    },
    {
      "epoch": 0.28745556119857796,
      "grad_norm": 2.9050920009613037,
      "learning_rate": 1.628135688618721e-05,
      "loss": 1.4716,
      "step": 566
    },
    {
      "epoch": 0.28796343321482987,
      "grad_norm": 3.6316444873809814,
      "learning_rate": 1.6268870584500775e-05,
      "loss": 1.4701,
      "step": 567
    },
    {
      "epoch": 0.2884713052310818,
      "grad_norm": 3.67140531539917,
      "learning_rate": 1.6256368160770982e-05,
      "loss": 1.2652,
      "step": 568
    },
    {
      "epoch": 0.2889791772473337,
      "grad_norm": 2.786526679992676,
      "learning_rate": 1.6243849647151098e-05,
      "loss": 1.4331,
      "step": 569
    },
    {
      "epoch": 0.2894870492635856,
      "grad_norm": 2.9930903911590576,
      "learning_rate": 1.623131507583576e-05,
      "loss": 1.392,
      "step": 570
    },
    {
      "epoch": 0.2899949212798375,
      "grad_norm": 2.4112536907196045,
      "learning_rate": 1.6218764479060906e-05,
      "loss": 1.6523,
      "step": 571
    },
    {
      "epoch": 0.2905027932960894,
      "grad_norm": 2.9992754459381104,
      "learning_rate": 1.6206197889103674e-05,
      "loss": 1.3157,
      "step": 572
    },
    {
      "epoch": 0.29101066531234127,
      "grad_norm": 2.8552370071411133,
      "learning_rate": 1.619361533828235e-05,
      "loss": 1.1532,
      "step": 573
    },
    {
      "epoch": 0.2915185373285932,
      "grad_norm": 3.1149823665618896,
      "learning_rate": 1.618101685895626e-05,
      "loss": 1.3478,
      "step": 574
    },
    {
      "epoch": 0.2920264093448451,
      "grad_norm": 3.108031988143921,
      "learning_rate": 1.6168402483525696e-05,
      "loss": 1.1179,
      "step": 575
    },
    {
      "epoch": 0.292534281361097,
      "grad_norm": 2.5380311012268066,
      "learning_rate": 1.615577224443183e-05,
      "loss": 1.5684,
      "step": 576
    },
    {
      "epoch": 0.2930421533773489,
      "grad_norm": 2.935041904449463,
      "learning_rate": 1.6143126174156624e-05,
      "loss": 1.5189,
      "step": 577
    },
    {
      "epoch": 0.2935500253936008,
      "grad_norm": 2.550523519515991,
      "learning_rate": 1.613046430522277e-05,
      "loss": 1.4392,
      "step": 578
    },
    {
      "epoch": 0.2940578974098527,
      "grad_norm": 2.2801826000213623,
      "learning_rate": 1.6117786670193578e-05,
      "loss": 1.5579,
      "step": 579
    },
    {
      "epoch": 0.29456576942610463,
      "grad_norm": 2.9374067783355713,
      "learning_rate": 1.610509330167291e-05,
      "loss": 1.41,
      "step": 580
    },
    {
      "epoch": 0.29507364144235654,
      "grad_norm": 3.2625834941864014,
      "learning_rate": 1.6092384232305086e-05,
      "loss": 1.4237,
      "step": 581
    },
    {
      "epoch": 0.29558151345860845,
      "grad_norm": 2.602550745010376,
      "learning_rate": 1.6079659494774813e-05,
      "loss": 1.4022,
      "step": 582
    },
    {
      "epoch": 0.29608938547486036,
      "grad_norm": 2.5818839073181152,
      "learning_rate": 1.606691912180708e-05,
      "loss": 1.5549,
      "step": 583
    },
    {
      "epoch": 0.29659725749111227,
      "grad_norm": 2.6626718044281006,
      "learning_rate": 1.6054163146167104e-05,
      "loss": 1.1246,
      "step": 584
    },
    {
      "epoch": 0.2971051295073641,
      "grad_norm": 3.313922166824341,
      "learning_rate": 1.6041391600660215e-05,
      "loss": 1.5451,
      "step": 585
    },
    {
      "epoch": 0.297613001523616,
      "grad_norm": 2.426194190979004,
      "learning_rate": 1.6028604518131787e-05,
      "loss": 1.4977,
      "step": 586
    },
    {
      "epoch": 0.29812087353986794,
      "grad_norm": 2.4327762126922607,
      "learning_rate": 1.6015801931467153e-05,
      "loss": 1.5168,
      "step": 587
    },
    {
      "epoch": 0.29862874555611985,
      "grad_norm": 3.15324068069458,
      "learning_rate": 1.6002983873591526e-05,
      "loss": 1.4526,
      "step": 588
    },
    {
      "epoch": 0.29913661757237175,
      "grad_norm": 2.9152519702911377,
      "learning_rate": 1.5990150377469894e-05,
      "loss": 1.4495,
      "step": 589
    },
    {
      "epoch": 0.29964448958862366,
      "grad_norm": 2.939667224884033,
      "learning_rate": 1.5977301476106953e-05,
      "loss": 1.4074,
      "step": 590
    },
    {
      "epoch": 0.3001523616048756,
      "grad_norm": 3.2030749320983887,
      "learning_rate": 1.596443720254702e-05,
      "loss": 1.3787,
      "step": 591
    },
    {
      "epoch": 0.3006602336211275,
      "grad_norm": 3.0066051483154297,
      "learning_rate": 1.595155758987396e-05,
      "loss": 1.2458,
      "step": 592
    },
    {
      "epoch": 0.3011681056373794,
      "grad_norm": 3.0819432735443115,
      "learning_rate": 1.5938662671211052e-05,
      "loss": 1.3818,
      "step": 593
    },
    {
      "epoch": 0.3016759776536313,
      "grad_norm": 2.645960807800293,
      "learning_rate": 1.592575247972097e-05,
      "loss": 1.5243,
      "step": 594
    },
    {
      "epoch": 0.3021838496698832,
      "grad_norm": 4.147794723510742,
      "learning_rate": 1.591282704860565e-05,
      "loss": 1.3021,
      "step": 595
    },
    {
      "epoch": 0.3026917216861351,
      "grad_norm": 3.1667566299438477,
      "learning_rate": 1.5899886411106223e-05,
      "loss": 1.502,
      "step": 596
    },
    {
      "epoch": 0.30319959370238697,
      "grad_norm": 3.224921941757202,
      "learning_rate": 1.5886930600502935e-05,
      "loss": 1.2507,
      "step": 597
    },
    {
      "epoch": 0.3037074657186389,
      "grad_norm": 2.8849408626556396,
      "learning_rate": 1.587395965011504e-05,
      "loss": 1.5756,
      "step": 598
    },
    {
      "epoch": 0.3042153377348908,
      "grad_norm": 2.8263800144195557,
      "learning_rate": 1.5860973593300747e-05,
      "loss": 1.6592,
      "step": 599
    },
    {
      "epoch": 0.3047232097511427,
      "grad_norm": 2.590561866760254,
      "learning_rate": 1.5847972463457097e-05,
      "loss": 1.3933,
      "step": 600
    },
    {
      "epoch": 0.3052310817673946,
      "grad_norm": 3.864027738571167,
      "learning_rate": 1.58349562940199e-05,
      "loss": 1.4139,
      "step": 601
    },
    {
      "epoch": 0.3057389537836465,
      "grad_norm": 3.146385669708252,
      "learning_rate": 1.582192511846365e-05,
      "loss": 1.4181,
      "step": 602
    },
    {
      "epoch": 0.3062468257998984,
      "grad_norm": 3.3183696269989014,
      "learning_rate": 1.580887897030143e-05,
      "loss": 1.1924,
      "step": 603
    },
    {
      "epoch": 0.30675469781615033,
      "grad_norm": 3.0171616077423096,
      "learning_rate": 1.5795817883084823e-05,
      "loss": 1.2576,
      "step": 604
    },
    {
      "epoch": 0.30726256983240224,
      "grad_norm": 2.751277208328247,
      "learning_rate": 1.5782741890403842e-05,
      "loss": 1.5396,
      "step": 605
    },
    {
      "epoch": 0.30777044184865415,
      "grad_norm": 2.6898446083068848,
      "learning_rate": 1.576965102588683e-05,
      "loss": 1.4069,
      "step": 606
    },
    {
      "epoch": 0.30827831386490606,
      "grad_norm": 2.6060962677001953,
      "learning_rate": 1.5756545323200373e-05,
      "loss": 1.4862,
      "step": 607
    },
    {
      "epoch": 0.30878618588115797,
      "grad_norm": 2.941141128540039,
      "learning_rate": 1.574342481604922e-05,
      "loss": 1.4382,
      "step": 608
    },
    {
      "epoch": 0.3092940578974099,
      "grad_norm": 3.190333127975464,
      "learning_rate": 1.573028953817619e-05,
      "loss": 1.5053,
      "step": 609
    },
    {
      "epoch": 0.30980192991366173,
      "grad_norm": 2.9648282527923584,
      "learning_rate": 1.5717139523362094e-05,
      "loss": 1.2999,
      "step": 610
    },
    {
      "epoch": 0.31030980192991364,
      "grad_norm": 3.3651657104492188,
      "learning_rate": 1.5703974805425645e-05,
      "loss": 1.3128,
      "step": 611
    },
    {
      "epoch": 0.31081767394616555,
      "grad_norm": 4.03769063949585,
      "learning_rate": 1.569079541822336e-05,
      "loss": 1.1174,
      "step": 612
    },
    {
      "epoch": 0.31132554596241746,
      "grad_norm": 3.595374584197998,
      "learning_rate": 1.5677601395649486e-05,
      "loss": 1.1697,
      "step": 613
    },
    {
      "epoch": 0.31183341797866937,
      "grad_norm": 3.3900370597839355,
      "learning_rate": 1.5664392771635912e-05,
      "loss": 1.6163,
      "step": 614
    },
    {
      "epoch": 0.3123412899949213,
      "grad_norm": 3.554189443588257,
      "learning_rate": 1.5651169580152075e-05,
      "loss": 1.1483,
      "step": 615
    },
    {
      "epoch": 0.3128491620111732,
      "grad_norm": 3.3712856769561768,
      "learning_rate": 1.5637931855204877e-05,
      "loss": 1.4755,
      "step": 616
    },
    {
      "epoch": 0.3133570340274251,
      "grad_norm": 2.823914051055908,
      "learning_rate": 1.5624679630838594e-05,
      "loss": 1.5944,
      "step": 617
    },
    {
      "epoch": 0.313864906043677,
      "grad_norm": 3.103543519973755,
      "learning_rate": 1.56114129411348e-05,
      "loss": 1.1896,
      "step": 618
    },
    {
      "epoch": 0.3143727780599289,
      "grad_norm": 3.144697666168213,
      "learning_rate": 1.5598131820212256e-05,
      "loss": 1.3939,
      "step": 619
    },
    {
      "epoch": 0.3148806500761808,
      "grad_norm": 2.5628550052642822,
      "learning_rate": 1.5584836302226846e-05,
      "loss": 1.4688,
      "step": 620
    },
    {
      "epoch": 0.31538852209243273,
      "grad_norm": 2.8005852699279785,
      "learning_rate": 1.5571526421371484e-05,
      "loss": 1.3955,
      "step": 621
    },
    {
      "epoch": 0.3158963941086846,
      "grad_norm": 3.2499237060546875,
      "learning_rate": 1.555820221187601e-05,
      "loss": 1.4999,
      "step": 622
    },
    {
      "epoch": 0.3164042661249365,
      "grad_norm": 3.178745985031128,
      "learning_rate": 1.5544863708007122e-05,
      "loss": 1.2467,
      "step": 623
    },
    {
      "epoch": 0.3169121381411884,
      "grad_norm": 2.5890581607818604,
      "learning_rate": 1.553151094406828e-05,
      "loss": 1.3692,
      "step": 624
    },
    {
      "epoch": 0.3174200101574403,
      "grad_norm": 2.8289737701416016,
      "learning_rate": 1.5518143954399612e-05,
      "loss": 1.3892,
      "step": 625
    },
    {
      "epoch": 0.3179278821736922,
      "grad_norm": 2.4055609703063965,
      "learning_rate": 1.550476277337784e-05,
      "loss": 1.4142,
      "step": 626
    },
    {
      "epoch": 0.31843575418994413,
      "grad_norm": 2.517207384109497,
      "learning_rate": 1.549136743541617e-05,
      "loss": 1.3777,
      "step": 627
    },
    {
      "epoch": 0.31894362620619604,
      "grad_norm": 3.7871336936950684,
      "learning_rate": 1.5477957974964225e-05,
      "loss": 1.4155,
      "step": 628
    },
    {
      "epoch": 0.31945149822244795,
      "grad_norm": 3.072606325149536,
      "learning_rate": 1.5464534426507954e-05,
      "loss": 1.3369,
      "step": 629
    },
    {
      "epoch": 0.31995937023869986,
      "grad_norm": 3.0552282333374023,
      "learning_rate": 1.545109682456952e-05,
      "loss": 1.5726,
      "step": 630
    },
    {
      "epoch": 0.32046724225495177,
      "grad_norm": 4.044310569763184,
      "learning_rate": 1.5437645203707242e-05,
      "loss": 1.3222,
      "step": 631
    },
    {
      "epoch": 0.3209751142712037,
      "grad_norm": 2.970475912094116,
      "learning_rate": 1.5424179598515486e-05,
      "loss": 1.3176,
      "step": 632
    },
    {
      "epoch": 0.3214829862874556,
      "grad_norm": 2.8568050861358643,
      "learning_rate": 1.5410700043624587e-05,
      "loss": 1.724,
      "step": 633
    },
    {
      "epoch": 0.32199085830370744,
      "grad_norm": 3.2946465015411377,
      "learning_rate": 1.539720657370075e-05,
      "loss": 1.3501,
      "step": 634
    },
    {
      "epoch": 0.32249873031995935,
      "grad_norm": 3.082104206085205,
      "learning_rate": 1.5383699223445967e-05,
      "loss": 1.3687,
      "step": 635
    },
    {
      "epoch": 0.32300660233621126,
      "grad_norm": 2.9089434146881104,
      "learning_rate": 1.537017802759793e-05,
      "loss": 1.2199,
      "step": 636
    },
    {
      "epoch": 0.32351447435246317,
      "grad_norm": 2.8768131732940674,
      "learning_rate": 1.5356643020929942e-05,
      "loss": 1.5327,
      "step": 637
    },
    {
      "epoch": 0.3240223463687151,
      "grad_norm": 2.816420793533325,
      "learning_rate": 1.5343094238250814e-05,
      "loss": 1.3783,
      "step": 638
    },
    {
      "epoch": 0.324530218384967,
      "grad_norm": 2.742537260055542,
      "learning_rate": 1.5329531714404787e-05,
      "loss": 1.3853,
      "step": 639
    },
    {
      "epoch": 0.3250380904012189,
      "grad_norm": 2.976881742477417,
      "learning_rate": 1.531595548427145e-05,
      "loss": 1.5038,
      "step": 640
    },
    {
      "epoch": 0.3255459624174708,
      "grad_norm": 2.651549816131592,
      "learning_rate": 1.5302365582765643e-05,
      "loss": 1.6156,
      "step": 641
    },
    {
      "epoch": 0.3260538344337227,
      "grad_norm": 2.5804250240325928,
      "learning_rate": 1.5288762044837348e-05,
      "loss": 1.4925,
      "step": 642
    },
    {
      "epoch": 0.3265617064499746,
      "grad_norm": 2.674237012863159,
      "learning_rate": 1.527514490547163e-05,
      "loss": 1.3758,
      "step": 643
    },
    {
      "epoch": 0.32706957846622653,
      "grad_norm": 3.6240010261535645,
      "learning_rate": 1.526151419968853e-05,
      "loss": 1.4575,
      "step": 644
    },
    {
      "epoch": 0.32757745048247844,
      "grad_norm": 3.1033742427825928,
      "learning_rate": 1.5247869962542985e-05,
      "loss": 1.5951,
      "step": 645
    },
    {
      "epoch": 0.32808532249873035,
      "grad_norm": 3.0908331871032715,
      "learning_rate": 1.5234212229124718e-05,
      "loss": 1.3567,
      "step": 646
    },
    {
      "epoch": 0.3285931945149822,
      "grad_norm": 3.233978509902954,
      "learning_rate": 1.522054103455818e-05,
      "loss": 1.3405,
      "step": 647
    },
    {
      "epoch": 0.3291010665312341,
      "grad_norm": 3.537057399749756,
      "learning_rate": 1.520685641400242e-05,
      "loss": 1.4314,
      "step": 648
    },
    {
      "epoch": 0.329608938547486,
      "grad_norm": 3.0778417587280273,
      "learning_rate": 1.5193158402651032e-05,
      "loss": 1.2213,
      "step": 649
    },
    {
      "epoch": 0.33011681056373793,
      "grad_norm": 4.0384440422058105,
      "learning_rate": 1.5179447035732038e-05,
      "loss": 1.3354,
      "step": 650
    },
    {
      "epoch": 0.33062468257998984,
      "grad_norm": 2.6769659519195557,
      "learning_rate": 1.516572234850781e-05,
      "loss": 1.4348,
      "step": 651
    },
    {
      "epoch": 0.33113255459624175,
      "grad_norm": 2.5838680267333984,
      "learning_rate": 1.5151984376274982e-05,
      "loss": 1.3927,
      "step": 652
    },
    {
      "epoch": 0.33164042661249366,
      "grad_norm": 2.274228572845459,
      "learning_rate": 1.5138233154364344e-05,
      "loss": 1.5037,
      "step": 653
    },
    {
      "epoch": 0.33214829862874556,
      "grad_norm": 3.4756743907928467,
      "learning_rate": 1.512446871814077e-05,
      "loss": 1.2171,
      "step": 654
    },
    {
      "epoch": 0.3326561706449975,
      "grad_norm": 3.236459732055664,
      "learning_rate": 1.5110691103003118e-05,
      "loss": 1.177,
      "step": 655
    },
    {
      "epoch": 0.3331640426612494,
      "grad_norm": 3.806804895401001,
      "learning_rate": 1.5096900344384126e-05,
      "loss": 1.3514,
      "step": 656
    },
    {
      "epoch": 0.3336719146775013,
      "grad_norm": 3.0247037410736084,
      "learning_rate": 1.5083096477750347e-05,
      "loss": 1.3816,
      "step": 657
    },
    {
      "epoch": 0.3341797866937532,
      "grad_norm": 2.83921480178833,
      "learning_rate": 1.5069279538602048e-05,
      "loss": 1.624,
      "step": 658
    },
    {
      "epoch": 0.33468765871000505,
      "grad_norm": 2.76662540435791,
      "learning_rate": 1.5055449562473102e-05,
      "loss": 1.4202,
      "step": 659
    },
    {
      "epoch": 0.33519553072625696,
      "grad_norm": 2.6441972255706787,
      "learning_rate": 1.5041606584930914e-05,
      "loss": 1.5947,
      "step": 660
    },
    {
      "epoch": 0.33570340274250887,
      "grad_norm": 2.4214117527008057,
      "learning_rate": 1.5027750641576333e-05,
      "loss": 1.4026,
      "step": 661
    },
    {
      "epoch": 0.3362112747587608,
      "grad_norm": 3.095984935760498,
      "learning_rate": 1.5013881768043547e-05,
      "loss": 1.5424,
      "step": 662
    },
    {
      "epoch": 0.3367191467750127,
      "grad_norm": 3.10390305519104,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.2982,
      "step": 663
    },
    {
      "epoch": 0.3372270187912646,
      "grad_norm": 3.0772242546081543,
      "learning_rate": 1.4986105373146296e-05,
      "loss": 1.3818,
      "step": 664
    },
    {
      "epoch": 0.3377348908075165,
      "grad_norm": 2.9678761959075928,
      "learning_rate": 1.49721979232161e-05,
      "loss": 1.229,
      "step": 665
    },
    {
      "epoch": 0.3382427628237684,
      "grad_norm": 3.0405545234680176,
      "learning_rate": 1.4958277685976077e-05,
      "loss": 1.5045,
      "step": 666
    },
    {
      "epoch": 0.3387506348400203,
      "grad_norm": 4.125433444976807,
      "learning_rate": 1.4944344697225754e-05,
      "loss": 1.0499,
      "step": 667
    },
    {
      "epoch": 0.33925850685627224,
      "grad_norm": 2.586026906967163,
      "learning_rate": 1.493039899279747e-05,
      "loss": 1.549,
      "step": 668
    },
    {
      "epoch": 0.33976637887252414,
      "grad_norm": 3.4039981365203857,
      "learning_rate": 1.4916440608556253e-05,
      "loss": 1.6548,
      "step": 669
    },
    {
      "epoch": 0.34027425088877605,
      "grad_norm": 3.3261430263519287,
      "learning_rate": 1.4902469580399746e-05,
      "loss": 1.2327,
      "step": 670
    },
    {
      "epoch": 0.3407821229050279,
      "grad_norm": 3.39119815826416,
      "learning_rate": 1.4888485944258113e-05,
      "loss": 1.0486,
      "step": 671
    },
    {
      "epoch": 0.3412899949212798,
      "grad_norm": 2.656126022338867,
      "learning_rate": 1.4874489736093933e-05,
      "loss": 1.5682,
      "step": 672
    },
    {
      "epoch": 0.3417978669375317,
      "grad_norm": 2.970787286758423,
      "learning_rate": 1.4860480991902127e-05,
      "loss": 1.2382,
      "step": 673
    },
    {
      "epoch": 0.34230573895378363,
      "grad_norm": 2.8525571823120117,
      "learning_rate": 1.4846459747709849e-05,
      "loss": 1.4311,
      "step": 674
    },
    {
      "epoch": 0.34281361097003554,
      "grad_norm": 3.1475439071655273,
      "learning_rate": 1.4832426039576401e-05,
      "loss": 1.2665,
      "step": 675
    },
    {
      "epoch": 0.34332148298628745,
      "grad_norm": 2.9784014225006104,
      "learning_rate": 1.481837990359315e-05,
      "loss": 1.2435,
      "step": 676
    },
    {
      "epoch": 0.34382935500253936,
      "grad_norm": 3.6366403102874756,
      "learning_rate": 1.4804321375883404e-05,
      "loss": 1.4673,
      "step": 677
    },
    {
      "epoch": 0.34433722701879127,
      "grad_norm": 3.0844566822052,
      "learning_rate": 1.4790250492602352e-05,
      "loss": 1.3489,
      "step": 678
    },
    {
      "epoch": 0.3448450990350432,
      "grad_norm": 2.902252674102783,
      "learning_rate": 1.4776167289936964e-05,
      "loss": 1.4117,
      "step": 679
    },
    {
      "epoch": 0.3453529710512951,
      "grad_norm": 3.076643943786621,
      "learning_rate": 1.4762071804105881e-05,
      "loss": 1.2952,
      "step": 680
    },
    {
      "epoch": 0.345860843067547,
      "grad_norm": 2.8510043621063232,
      "learning_rate": 1.4747964071359343e-05,
      "loss": 1.4523,
      "step": 681
    },
    {
      "epoch": 0.3463687150837989,
      "grad_norm": 2.854254722595215,
      "learning_rate": 1.4733844127979073e-05,
      "loss": 1.6375,
      "step": 682
    },
    {
      "epoch": 0.34687658710005076,
      "grad_norm": 3.7380759716033936,
      "learning_rate": 1.4719712010278211e-05,
      "loss": 1.4145,
      "step": 683
    },
    {
      "epoch": 0.34738445911630267,
      "grad_norm": 2.862888813018799,
      "learning_rate": 1.4705567754601204e-05,
      "loss": 1.4546,
      "step": 684
    },
    {
      "epoch": 0.3478923311325546,
      "grad_norm": 3.2201626300811768,
      "learning_rate": 1.4691411397323706e-05,
      "loss": 1.5113,
      "step": 685
    },
    {
      "epoch": 0.3484002031488065,
      "grad_norm": 3.1785452365875244,
      "learning_rate": 1.46772429748525e-05,
      "loss": 1.3296,
      "step": 686
    },
    {
      "epoch": 0.3489080751650584,
      "grad_norm": 2.7606334686279297,
      "learning_rate": 1.4663062523625398e-05,
      "loss": 1.433,
      "step": 687
    },
    {
      "epoch": 0.3494159471813103,
      "grad_norm": 3.220571756362915,
      "learning_rate": 1.4648870080111144e-05,
      "loss": 1.2379,
      "step": 688
    },
    {
      "epoch": 0.3499238191975622,
      "grad_norm": 2.676091194152832,
      "learning_rate": 1.463466568080933e-05,
      "loss": 1.3755,
      "step": 689
    },
    {
      "epoch": 0.3504316912138141,
      "grad_norm": 3.6125404834747314,
      "learning_rate": 1.4620449362250286e-05,
      "loss": 1.2007,
      "step": 690
    },
    {
      "epoch": 0.35093956323006603,
      "grad_norm": 2.850972890853882,
      "learning_rate": 1.4606221160995002e-05,
      "loss": 1.5413,
      "step": 691
    },
    {
      "epoch": 0.35144743524631794,
      "grad_norm": 2.8337199687957764,
      "learning_rate": 1.459198111363503e-05,
      "loss": 1.628,
      "step": 692
    },
    {
      "epoch": 0.35195530726256985,
      "grad_norm": 2.9772043228149414,
      "learning_rate": 1.4577729256792377e-05,
      "loss": 1.4941,
      "step": 693
    },
    {
      "epoch": 0.35246317927882176,
      "grad_norm": 2.7553727626800537,
      "learning_rate": 1.4563465627119428e-05,
      "loss": 1.2939,
      "step": 694
    },
    {
      "epoch": 0.35297105129507367,
      "grad_norm": 4.402785301208496,
      "learning_rate": 1.4549190261298848e-05,
      "loss": 1.1561,
      "step": 695
    },
    {
      "epoch": 0.3534789233113255,
      "grad_norm": 3.322864055633545,
      "learning_rate": 1.4534903196043477e-05,
      "loss": 1.4288,
      "step": 696
    },
    {
      "epoch": 0.35398679532757743,
      "grad_norm": 3.631467342376709,
      "learning_rate": 1.4520604468096252e-05,
      "loss": 1.2479,
      "step": 697
    },
    {
      "epoch": 0.35449466734382934,
      "grad_norm": 3.06563138961792,
      "learning_rate": 1.4506294114230093e-05,
      "loss": 1.3205,
      "step": 698
    },
    {
      "epoch": 0.35500253936008125,
      "grad_norm": 2.893291711807251,
      "learning_rate": 1.4491972171247825e-05,
      "loss": 1.4051,
      "step": 699
    },
    {
      "epoch": 0.35551041137633316,
      "grad_norm": 3.3483219146728516,
      "learning_rate": 1.4477638675982081e-05,
      "loss": 1.5237,
      "step": 700
    },
    {
      "epoch": 0.35601828339258507,
      "grad_norm": 3.449445962905884,
      "learning_rate": 1.44632936652952e-05,
      "loss": 1.2882,
      "step": 701
    },
    {
      "epoch": 0.356526155408837,
      "grad_norm": 3.7622361183166504,
      "learning_rate": 1.444893717607913e-05,
      "loss": 1.4397,
      "step": 702
    },
    {
      "epoch": 0.3570340274250889,
      "grad_norm": 3.3970916271209717,
      "learning_rate": 1.443456924525535e-05,
      "loss": 1.4158,
      "step": 703
    },
    {
      "epoch": 0.3575418994413408,
      "grad_norm": 3.098458766937256,
      "learning_rate": 1.4420189909774757e-05,
      "loss": 1.371,
      "step": 704
    },
    {
      "epoch": 0.3580497714575927,
      "grad_norm": 2.868194818496704,
      "learning_rate": 1.4405799206617586e-05,
      "loss": 1.0695,
      "step": 705
    },
    {
      "epoch": 0.3585576434738446,
      "grad_norm": 3.3876736164093018,
      "learning_rate": 1.439139717279329e-05,
      "loss": 1.2454,
      "step": 706
    },
    {
      "epoch": 0.3590655154900965,
      "grad_norm": 3.4045627117156982,
      "learning_rate": 1.437698384534048e-05,
      "loss": 1.2066,
      "step": 707
    },
    {
      "epoch": 0.3595733875063484,
      "grad_norm": 2.855515480041504,
      "learning_rate": 1.4362559261326805e-05,
      "loss": 1.6359,
      "step": 708
    },
    {
      "epoch": 0.3600812595226003,
      "grad_norm": 2.887098550796509,
      "learning_rate": 1.4348123457848857e-05,
      "loss": 1.1533,
      "step": 709
    },
    {
      "epoch": 0.3605891315388522,
      "grad_norm": 3.2437286376953125,
      "learning_rate": 1.433367647203209e-05,
      "loss": 1.4143,
      "step": 710
    },
    {
      "epoch": 0.3610970035551041,
      "grad_norm": 3.3504650592803955,
      "learning_rate": 1.4319218341030719e-05,
      "loss": 1.4759,
      "step": 711
    },
    {
      "epoch": 0.361604875571356,
      "grad_norm": 3.7517337799072266,
      "learning_rate": 1.4304749102027608e-05,
      "loss": 1.3028,
      "step": 712
    },
    {
      "epoch": 0.3621127475876079,
      "grad_norm": 2.6952829360961914,
      "learning_rate": 1.4290268792234203e-05,
      "loss": 1.3487,
      "step": 713
    },
    {
      "epoch": 0.36262061960385983,
      "grad_norm": 3.4969964027404785,
      "learning_rate": 1.427577744889041e-05,
      "loss": 1.5259,
      "step": 714
    },
    {
      "epoch": 0.36312849162011174,
      "grad_norm": 2.790510892868042,
      "learning_rate": 1.4261275109264516e-05,
      "loss": 1.3438,
      "step": 715
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 3.0047624111175537,
      "learning_rate": 1.424676181065309e-05,
      "loss": 1.3558,
      "step": 716
    },
    {
      "epoch": 0.36414423565261556,
      "grad_norm": 3.4911231994628906,
      "learning_rate": 1.4232237590380882e-05,
      "loss": 1.4434,
      "step": 717
    },
    {
      "epoch": 0.36465210766886746,
      "grad_norm": 3.183309555053711,
      "learning_rate": 1.421770248580073e-05,
      "loss": 1.1461,
      "step": 718
    },
    {
      "epoch": 0.3651599796851194,
      "grad_norm": 3.691046714782715,
      "learning_rate": 1.4203156534293463e-05,
      "loss": 1.2286,
      "step": 719
    },
    {
      "epoch": 0.3656678517013712,
      "grad_norm": 3.438445568084717,
      "learning_rate": 1.4188599773267808e-05,
      "loss": 1.0223,
      "step": 720
    },
    {
      "epoch": 0.36617572371762314,
      "grad_norm": 2.969409704208374,
      "learning_rate": 1.4174032240160291e-05,
      "loss": 1.4687,
      "step": 721
    },
    {
      "epoch": 0.36668359573387505,
      "grad_norm": 3.7304344177246094,
      "learning_rate": 1.4159453972435135e-05,
      "loss": 1.0545,
      "step": 722
    },
    {
      "epoch": 0.36719146775012695,
      "grad_norm": 3.4797556400299072,
      "learning_rate": 1.4144865007584184e-05,
      "loss": 1.4136,
      "step": 723
    },
    {
      "epoch": 0.36769933976637886,
      "grad_norm": 3.0512192249298096,
      "learning_rate": 1.4130265383126778e-05,
      "loss": 1.3685,
      "step": 724
    },
    {
      "epoch": 0.36820721178263077,
      "grad_norm": 3.239837646484375,
      "learning_rate": 1.4115655136609677e-05,
      "loss": 1.4191,
      "step": 725
    },
    {
      "epoch": 0.3687150837988827,
      "grad_norm": 3.2800753116607666,
      "learning_rate": 1.4101034305606962e-05,
      "loss": 1.2906,
      "step": 726
    },
    {
      "epoch": 0.3692229558151346,
      "grad_norm": 3.322382926940918,
      "learning_rate": 1.4086402927719929e-05,
      "loss": 1.3609,
      "step": 727
    },
    {
      "epoch": 0.3697308278313865,
      "grad_norm": 3.090886354446411,
      "learning_rate": 1.4071761040576996e-05,
      "loss": 1.3124,
      "step": 728
    },
    {
      "epoch": 0.3702386998476384,
      "grad_norm": 3.5921313762664795,
      "learning_rate": 1.4057108681833623e-05,
      "loss": 1.4419,
      "step": 729
    },
    {
      "epoch": 0.3707465718638903,
      "grad_norm": 3.024637460708618,
      "learning_rate": 1.4042445889172178e-05,
      "loss": 1.5245,
      "step": 730
    },
    {
      "epoch": 0.3712544438801422,
      "grad_norm": 2.6643972396850586,
      "learning_rate": 1.402777270030188e-05,
      "loss": 1.3496,
      "step": 731
    },
    {
      "epoch": 0.37176231589639414,
      "grad_norm": 3.0148916244506836,
      "learning_rate": 1.4013089152958681e-05,
      "loss": 1.5488,
      "step": 732
    },
    {
      "epoch": 0.372270187912646,
      "grad_norm": 3.736215591430664,
      "learning_rate": 1.3998395284905166e-05,
      "loss": 1.4689,
      "step": 733
    },
    {
      "epoch": 0.3727780599288979,
      "grad_norm": 3.5619914531707764,
      "learning_rate": 1.3983691133930473e-05,
      "loss": 1.2974,
      "step": 734
    },
    {
      "epoch": 0.3732859319451498,
      "grad_norm": 2.7390873432159424,
      "learning_rate": 1.3968976737850172e-05,
      "loss": 0.876,
      "step": 735
    },
    {
      "epoch": 0.3737938039614017,
      "grad_norm": 3.622314691543579,
      "learning_rate": 1.3954252134506193e-05,
      "loss": 1.1496,
      "step": 736
    },
    {
      "epoch": 0.3743016759776536,
      "grad_norm": 2.7309963703155518,
      "learning_rate": 1.393951736176671e-05,
      "loss": 1.2773,
      "step": 737
    },
    {
      "epoch": 0.37480954799390553,
      "grad_norm": 3.10971736907959,
      "learning_rate": 1.392477245752605e-05,
      "loss": 1.3917,
      "step": 738
    },
    {
      "epoch": 0.37531742001015744,
      "grad_norm": 2.9807276725769043,
      "learning_rate": 1.3910017459704602e-05,
      "loss": 1.5002,
      "step": 739
    },
    {
      "epoch": 0.37582529202640935,
      "grad_norm": 3.2353007793426514,
      "learning_rate": 1.3895252406248708e-05,
      "loss": 1.2697,
      "step": 740
    },
    {
      "epoch": 0.37633316404266126,
      "grad_norm": 3.410163402557373,
      "learning_rate": 1.3880477335130573e-05,
      "loss": 1.3525,
      "step": 741
    },
    {
      "epoch": 0.37684103605891317,
      "grad_norm": 2.844461679458618,
      "learning_rate": 1.3865692284348162e-05,
      "loss": 1.4138,
      "step": 742
    },
    {
      "epoch": 0.3773489080751651,
      "grad_norm": 2.8310110569000244,
      "learning_rate": 1.3850897291925104e-05,
      "loss": 1.3118,
      "step": 743
    },
    {
      "epoch": 0.377856780091417,
      "grad_norm": 3.3003883361816406,
      "learning_rate": 1.3836092395910605e-05,
      "loss": 1.1658,
      "step": 744
    },
    {
      "epoch": 0.37836465210766884,
      "grad_norm": 3.317352533340454,
      "learning_rate": 1.382127763437933e-05,
      "loss": 1.3725,
      "step": 745
    },
    {
      "epoch": 0.37887252412392075,
      "grad_norm": 2.9160075187683105,
      "learning_rate": 1.3806453045431326e-05,
      "loss": 1.4349,
      "step": 746
    },
    {
      "epoch": 0.37938039614017266,
      "grad_norm": 3.4108498096466064,
      "learning_rate": 1.3791618667191902e-05,
      "loss": 1.3338,
      "step": 747
    },
    {
      "epoch": 0.37988826815642457,
      "grad_norm": 4.158885478973389,
      "learning_rate": 1.3776774537811548e-05,
      "loss": 1.081,
      "step": 748
    },
    {
      "epoch": 0.3803961401726765,
      "grad_norm": 3.3611350059509277,
      "learning_rate": 1.3761920695465838e-05,
      "loss": 1.4261,
      "step": 749
    },
    {
      "epoch": 0.3809040121889284,
      "grad_norm": 3.7590487003326416,
      "learning_rate": 1.3747057178355316e-05,
      "loss": 1.2509,
      "step": 750
    },
    {
      "epoch": 0.3814118842051803,
      "grad_norm": 3.6111934185028076,
      "learning_rate": 1.3732184024705413e-05,
      "loss": 1.3899,
      "step": 751
    },
    {
      "epoch": 0.3819197562214322,
      "grad_norm": 4.133301734924316,
      "learning_rate": 1.3717301272766336e-05,
      "loss": 1.5087,
      "step": 752
    },
    {
      "epoch": 0.3824276282376841,
      "grad_norm": 3.427138090133667,
      "learning_rate": 1.3702408960812989e-05,
      "loss": 1.3835,
      "step": 753
    },
    {
      "epoch": 0.382935500253936,
      "grad_norm": 3.1767027378082275,
      "learning_rate": 1.3687507127144848e-05,
      "loss": 1.3515,
      "step": 754
    },
    {
      "epoch": 0.38344337227018793,
      "grad_norm": 3.710176706314087,
      "learning_rate": 1.3672595810085887e-05,
      "loss": 1.2997,
      "step": 755
    },
    {
      "epoch": 0.38395124428643984,
      "grad_norm": 2.8906760215759277,
      "learning_rate": 1.3657675047984462e-05,
      "loss": 1.3962,
      "step": 756
    },
    {
      "epoch": 0.3844591163026917,
      "grad_norm": 2.8986518383026123,
      "learning_rate": 1.3642744879213226e-05,
      "loss": 1.5398,
      "step": 757
    },
    {
      "epoch": 0.3849669883189436,
      "grad_norm": 3.346881628036499,
      "learning_rate": 1.3627805342169019e-05,
      "loss": 1.2606,
      "step": 758
    },
    {
      "epoch": 0.3854748603351955,
      "grad_norm": 3.2820770740509033,
      "learning_rate": 1.361285647527277e-05,
      "loss": 1.3161,
      "step": 759
    },
    {
      "epoch": 0.3859827323514474,
      "grad_norm": 2.738898992538452,
      "learning_rate": 1.359789831696941e-05,
      "loss": 1.3221,
      "step": 760
    },
    {
      "epoch": 0.38649060436769933,
      "grad_norm": 3.425286293029785,
      "learning_rate": 1.3582930905727772e-05,
      "loss": 1.3743,
      "step": 761
    },
    {
      "epoch": 0.38699847638395124,
      "grad_norm": 2.762843370437622,
      "learning_rate": 1.3567954280040463e-05,
      "loss": 1.5665,
      "step": 762
    },
    {
      "epoch": 0.38750634840020315,
      "grad_norm": 3.123188018798828,
      "learning_rate": 1.3552968478423809e-05,
      "loss": 1.5878,
      "step": 763
    },
    {
      "epoch": 0.38801422041645506,
      "grad_norm": 3.069780111312866,
      "learning_rate": 1.3537973539417719e-05,
      "loss": 1.4137,
      "step": 764
    },
    {
      "epoch": 0.38852209243270697,
      "grad_norm": 3.6509578227996826,
      "learning_rate": 1.3522969501585612e-05,
      "loss": 1.5572,
      "step": 765
    },
    {
      "epoch": 0.3890299644489589,
      "grad_norm": 3.1971583366394043,
      "learning_rate": 1.35079564035143e-05,
      "loss": 1.3368,
      "step": 766
    },
    {
      "epoch": 0.3895378364652108,
      "grad_norm": 3.548170566558838,
      "learning_rate": 1.34929342838139e-05,
      "loss": 1.2282,
      "step": 767
    },
    {
      "epoch": 0.3900457084814627,
      "grad_norm": 2.829338788986206,
      "learning_rate": 1.3477903181117731e-05,
      "loss": 1.3614,
      "step": 768
    },
    {
      "epoch": 0.3905535804977146,
      "grad_norm": 3.30485463142395,
      "learning_rate": 1.3462863134082214e-05,
      "loss": 1.4793,
      "step": 769
    },
    {
      "epoch": 0.39106145251396646,
      "grad_norm": 3.015761613845825,
      "learning_rate": 1.3447814181386763e-05,
      "loss": 1.5157,
      "step": 770
    },
    {
      "epoch": 0.39156932453021837,
      "grad_norm": 3.4275357723236084,
      "learning_rate": 1.343275636173371e-05,
      "loss": 1.2972,
      "step": 771
    },
    {
      "epoch": 0.3920771965464703,
      "grad_norm": 3.141019344329834,
      "learning_rate": 1.3417689713848179e-05,
      "loss": 1.2101,
      "step": 772
    },
    {
      "epoch": 0.3925850685627222,
      "grad_norm": 3.313812017440796,
      "learning_rate": 1.3402614276478008e-05,
      "loss": 1.4199,
      "step": 773
    },
    {
      "epoch": 0.3930929405789741,
      "grad_norm": 3.148071050643921,
      "learning_rate": 1.3387530088393628e-05,
      "loss": 1.5838,
      "step": 774
    },
    {
      "epoch": 0.393600812595226,
      "grad_norm": 3.4933178424835205,
      "learning_rate": 1.3372437188387988e-05,
      "loss": 1.4677,
      "step": 775
    },
    {
      "epoch": 0.3941086846114779,
      "grad_norm": 2.875379800796509,
      "learning_rate": 1.3357335615276434e-05,
      "loss": 1.3789,
      "step": 776
    },
    {
      "epoch": 0.3946165566277298,
      "grad_norm": 2.804950714111328,
      "learning_rate": 1.3342225407896614e-05,
      "loss": 1.3081,
      "step": 777
    },
    {
      "epoch": 0.39512442864398173,
      "grad_norm": 3.7760562896728516,
      "learning_rate": 1.3327106605108389e-05,
      "loss": 1.5066,
      "step": 778
    },
    {
      "epoch": 0.39563230066023364,
      "grad_norm": 3.2691030502319336,
      "learning_rate": 1.3311979245793723e-05,
      "loss": 1.3408,
      "step": 779
    },
    {
      "epoch": 0.39614017267648555,
      "grad_norm": 3.711343765258789,
      "learning_rate": 1.329684336885658e-05,
      "loss": 1.2556,
      "step": 780
    },
    {
      "epoch": 0.39664804469273746,
      "grad_norm": 3.3347604274749756,
      "learning_rate": 1.3281699013222836e-05,
      "loss": 1.4213,
      "step": 781
    },
    {
      "epoch": 0.3971559167089893,
      "grad_norm": 3.439558744430542,
      "learning_rate": 1.3266546217840172e-05,
      "loss": 0.9587,
      "step": 782
    },
    {
      "epoch": 0.3976637887252412,
      "grad_norm": 3.292663097381592,
      "learning_rate": 1.3251385021677967e-05,
      "loss": 1.1098,
      "step": 783
    },
    {
      "epoch": 0.3981716607414931,
      "grad_norm": 3.108430862426758,
      "learning_rate": 1.3236215463727218e-05,
      "loss": 1.308,
      "step": 784
    },
    {
      "epoch": 0.39867953275774504,
      "grad_norm": 2.932769298553467,
      "learning_rate": 1.322103758300041e-05,
      "loss": 1.4641,
      "step": 785
    },
    {
      "epoch": 0.39918740477399695,
      "grad_norm": 3.71151065826416,
      "learning_rate": 1.3205851418531442e-05,
      "loss": 1.4567,
      "step": 786
    },
    {
      "epoch": 0.39969527679024885,
      "grad_norm": 2.6987061500549316,
      "learning_rate": 1.3190657009375516e-05,
      "loss": 1.3758,
      "step": 787
    },
    {
      "epoch": 0.40020314880650076,
      "grad_norm": 2.942145586013794,
      "learning_rate": 1.3175454394609037e-05,
      "loss": 1.3166,
      "step": 788
    },
    {
      "epoch": 0.4007110208227527,
      "grad_norm": 4.66526985168457,
      "learning_rate": 1.3160243613329512e-05,
      "loss": 1.2639,
      "step": 789
    },
    {
      "epoch": 0.4012188928390046,
      "grad_norm": 2.9264190196990967,
      "learning_rate": 1.3145024704655453e-05,
      "loss": 1.5059,
      "step": 790
    },
    {
      "epoch": 0.4017267648552565,
      "grad_norm": 3.232764482498169,
      "learning_rate": 1.3129797707726268e-05,
      "loss": 1.3631,
      "step": 791
    },
    {
      "epoch": 0.4022346368715084,
      "grad_norm": 3.2342617511749268,
      "learning_rate": 1.3114562661702173e-05,
      "loss": 1.2905,
      "step": 792
    },
    {
      "epoch": 0.4027425088877603,
      "grad_norm": 5.082442283630371,
      "learning_rate": 1.3099319605764078e-05,
      "loss": 0.6581,
      "step": 793
    },
    {
      "epoch": 0.40325038090401216,
      "grad_norm": 3.325721025466919,
      "learning_rate": 1.3084068579113495e-05,
      "loss": 1.2834,
      "step": 794
    },
    {
      "epoch": 0.40375825292026407,
      "grad_norm": 3.943657636642456,
      "learning_rate": 1.3068809620972438e-05,
      "loss": 1.5556,
      "step": 795
    },
    {
      "epoch": 0.404266124936516,
      "grad_norm": 2.8499321937561035,
      "learning_rate": 1.3053542770583314e-05,
      "loss": 1.5058,
      "step": 796
    },
    {
      "epoch": 0.4047739969527679,
      "grad_norm": 4.564476013183594,
      "learning_rate": 1.303826806720883e-05,
      "loss": 1.3175,
      "step": 797
    },
    {
      "epoch": 0.4052818689690198,
      "grad_norm": 3.9246888160705566,
      "learning_rate": 1.3022985550131893e-05,
      "loss": 1.1799,
      "step": 798
    },
    {
      "epoch": 0.4057897409852717,
      "grad_norm": 3.540846347808838,
      "learning_rate": 1.300769525865549e-05,
      "loss": 1.6187,
      "step": 799
    },
    {
      "epoch": 0.4062976130015236,
      "grad_norm": 2.856062412261963,
      "learning_rate": 1.2992397232102623e-05,
      "loss": 1.3625,
      "step": 800
    },
    {
      "epoch": 0.4068054850177755,
      "grad_norm": 3.0858078002929688,
      "learning_rate": 1.2977091509816166e-05,
      "loss": 1.5088,
      "step": 801
    },
    {
      "epoch": 0.40731335703402743,
      "grad_norm": 3.1453182697296143,
      "learning_rate": 1.2961778131158798e-05,
      "loss": 1.2945,
      "step": 802
    },
    {
      "epoch": 0.40782122905027934,
      "grad_norm": 3.0279111862182617,
      "learning_rate": 1.2946457135512889e-05,
      "loss": 1.5215,
      "step": 803
    },
    {
      "epoch": 0.40832910106653125,
      "grad_norm": 3.2221486568450928,
      "learning_rate": 1.2931128562280388e-05,
      "loss": 1.1617,
      "step": 804
    },
    {
      "epoch": 0.40883697308278316,
      "grad_norm": 3.101377010345459,
      "learning_rate": 1.2915792450882745e-05,
      "loss": 1.2618,
      "step": 805
    },
    {
      "epoch": 0.40934484509903507,
      "grad_norm": 4.297948360443115,
      "learning_rate": 1.2900448840760779e-05,
      "loss": 0.8525,
      "step": 806
    },
    {
      "epoch": 0.4098527171152869,
      "grad_norm": 3.5713438987731934,
      "learning_rate": 1.288509777137461e-05,
      "loss": 1.1102,
      "step": 807
    },
    {
      "epoch": 0.41036058913153883,
      "grad_norm": 3.4977385997772217,
      "learning_rate": 1.2869739282203538e-05,
      "loss": 1.5499,
      "step": 808
    },
    {
      "epoch": 0.41086846114779074,
      "grad_norm": 3.0273525714874268,
      "learning_rate": 1.2854373412745936e-05,
      "loss": 1.3806,
      "step": 809
    },
    {
      "epoch": 0.41137633316404265,
      "grad_norm": 3.4045588970184326,
      "learning_rate": 1.2839000202519166e-05,
      "loss": 1.2375,
      "step": 810
    },
    {
      "epoch": 0.41188420518029456,
      "grad_norm": 3.0983994007110596,
      "learning_rate": 1.2823619691059471e-05,
      "loss": 1.3675,
      "step": 811
    },
    {
      "epoch": 0.41239207719654647,
      "grad_norm": 3.5127463340759277,
      "learning_rate": 1.2808231917921857e-05,
      "loss": 1.4256,
      "step": 812
    },
    {
      "epoch": 0.4128999492127984,
      "grad_norm": 3.695651054382324,
      "learning_rate": 1.2792836922680026e-05,
      "loss": 1.4067,
      "step": 813
    },
    {
      "epoch": 0.4134078212290503,
      "grad_norm": 3.353546380996704,
      "learning_rate": 1.2777434744926231e-05,
      "loss": 1.3569,
      "step": 814
    },
    {
      "epoch": 0.4139156932453022,
      "grad_norm": 4.223036289215088,
      "learning_rate": 1.2762025424271213e-05,
      "loss": 1.0206,
      "step": 815
    },
    {
      "epoch": 0.4144235652615541,
      "grad_norm": 3.227527379989624,
      "learning_rate": 1.274660900034408e-05,
      "loss": 1.3569,
      "step": 816
    },
    {
      "epoch": 0.414931437277806,
      "grad_norm": 3.7230629920959473,
      "learning_rate": 1.2731185512792203e-05,
      "loss": 1.0783,
      "step": 817
    },
    {
      "epoch": 0.4154393092940579,
      "grad_norm": 3.222513198852539,
      "learning_rate": 1.2715755001281125e-05,
      "loss": 1.4662,
      "step": 818
    },
    {
      "epoch": 0.4159471813103098,
      "grad_norm": 3.7021639347076416,
      "learning_rate": 1.2700317505494446e-05,
      "loss": 1.4293,
      "step": 819
    },
    {
      "epoch": 0.4164550533265617,
      "grad_norm": 3.0131332874298096,
      "learning_rate": 1.268487306513373e-05,
      "loss": 1.237,
      "step": 820
    },
    {
      "epoch": 0.4169629253428136,
      "grad_norm": 3.4458506107330322,
      "learning_rate": 1.2669421719918411e-05,
      "loss": 1.1909,
      "step": 821
    },
    {
      "epoch": 0.4174707973590655,
      "grad_norm": 3.5575385093688965,
      "learning_rate": 1.265396350958566e-05,
      "loss": 1.4145,
      "step": 822
    },
    {
      "epoch": 0.4179786693753174,
      "grad_norm": 3.4635229110717773,
      "learning_rate": 1.2638498473890323e-05,
      "loss": 1.3249,
      "step": 823
    },
    {
      "epoch": 0.4184865413915693,
      "grad_norm": 3.8467001914978027,
      "learning_rate": 1.2623026652604788e-05,
      "loss": 1.2442,
      "step": 824
    },
    {
      "epoch": 0.41899441340782123,
      "grad_norm": 2.934985399246216,
      "learning_rate": 1.2607548085518903e-05,
      "loss": 1.3453,
      "step": 825
    },
    {
      "epoch": 0.41950228542407314,
      "grad_norm": 3.92087721824646,
      "learning_rate": 1.2592062812439854e-05,
      "loss": 1.2389,
      "step": 826
    },
    {
      "epoch": 0.42001015744032505,
      "grad_norm": 3.172184705734253,
      "learning_rate": 1.257657087319208e-05,
      "loss": 1.4821,
      "step": 827
    },
    {
      "epoch": 0.42051802945657696,
      "grad_norm": 3.199178457260132,
      "learning_rate": 1.2561072307617161e-05,
      "loss": 1.244,
      "step": 828
    },
    {
      "epoch": 0.42102590147282887,
      "grad_norm": 3.8658840656280518,
      "learning_rate": 1.2545567155573725e-05,
      "loss": 1.2649,
      "step": 829
    },
    {
      "epoch": 0.4215337734890808,
      "grad_norm": 3.9423792362213135,
      "learning_rate": 1.2530055456937321e-05,
      "loss": 1.4782,
      "step": 830
    },
    {
      "epoch": 0.42204164550533263,
      "grad_norm": 3.3361122608184814,
      "learning_rate": 1.251453725160036e-05,
      "loss": 1.5312,
      "step": 831
    },
    {
      "epoch": 0.42254951752158454,
      "grad_norm": 4.270658016204834,
      "learning_rate": 1.249901257947197e-05,
      "loss": 1.1508,
      "step": 832
    },
    {
      "epoch": 0.42305738953783645,
      "grad_norm": 3.095872163772583,
      "learning_rate": 1.2483481480477908e-05,
      "loss": 1.3785,
      "step": 833
    },
    {
      "epoch": 0.42356526155408836,
      "grad_norm": 3.7208592891693115,
      "learning_rate": 1.246794399456047e-05,
      "loss": 1.4228,
      "step": 834
    },
    {
      "epoch": 0.42407313357034027,
      "grad_norm": 3.2617084980010986,
      "learning_rate": 1.2452400161678367e-05,
      "loss": 1.3602,
      "step": 835
    },
    {
      "epoch": 0.4245810055865922,
      "grad_norm": 4.436614990234375,
      "learning_rate": 1.2436850021806644e-05,
      "loss": 1.0053,
      "step": 836
    },
    {
      "epoch": 0.4250888776028441,
      "grad_norm": 3.6590428352355957,
      "learning_rate": 1.2421293614936553e-05,
      "loss": 1.3901,
      "step": 837
    },
    {
      "epoch": 0.425596749619096,
      "grad_norm": 4.084576606750488,
      "learning_rate": 1.2405730981075478e-05,
      "loss": 1.4956,
      "step": 838
    },
    {
      "epoch": 0.4261046216353479,
      "grad_norm": 3.201557159423828,
      "learning_rate": 1.239016216024681e-05,
      "loss": 1.3609,
      "step": 839
    },
    {
      "epoch": 0.4266124936515998,
      "grad_norm": 3.2476463317871094,
      "learning_rate": 1.2374587192489846e-05,
      "loss": 1.5228,
      "step": 840
    },
    {
      "epoch": 0.4271203656678517,
      "grad_norm": 4.044534683227539,
      "learning_rate": 1.23590061178597e-05,
      "loss": 1.0868,
      "step": 841
    },
    {
      "epoch": 0.42762823768410363,
      "grad_norm": 2.8839735984802246,
      "learning_rate": 1.2343418976427191e-05,
      "loss": 1.4421,
      "step": 842
    },
    {
      "epoch": 0.4281361097003555,
      "grad_norm": 3.148660182952881,
      "learning_rate": 1.2327825808278728e-05,
      "loss": 1.4414,
      "step": 843
    },
    {
      "epoch": 0.4286439817166074,
      "grad_norm": 3.5366671085357666,
      "learning_rate": 1.2312226653516237e-05,
      "loss": 1.2989,
      "step": 844
    },
    {
      "epoch": 0.4291518537328593,
      "grad_norm": 3.481353998184204,
      "learning_rate": 1.229662155225703e-05,
      "loss": 1.441,
      "step": 845
    },
    {
      "epoch": 0.4296597257491112,
      "grad_norm": 4.349026679992676,
      "learning_rate": 1.2281010544633708e-05,
      "loss": 1.2352,
      "step": 846
    },
    {
      "epoch": 0.4301675977653631,
      "grad_norm": 3.208542823791504,
      "learning_rate": 1.2265393670794079e-05,
      "loss": 1.2023,
      "step": 847
    },
    {
      "epoch": 0.43067546978161503,
      "grad_norm": 3.4003970623016357,
      "learning_rate": 1.2249770970901023e-05,
      "loss": 1.338,
      "step": 848
    },
    {
      "epoch": 0.43118334179786694,
      "grad_norm": 3.414480447769165,
      "learning_rate": 1.2234142485132398e-05,
      "loss": 1.1968,
      "step": 849
    },
    {
      "epoch": 0.43169121381411885,
      "grad_norm": 3.3189659118652344,
      "learning_rate": 1.221850825368096e-05,
      "loss": 1.4158,
      "step": 850
    },
    {
      "epoch": 0.43219908583037076,
      "grad_norm": 4.061149597167969,
      "learning_rate": 1.2202868316754226e-05,
      "loss": 1.1293,
      "step": 851
    },
    {
      "epoch": 0.43270695784662266,
      "grad_norm": 3.178410768508911,
      "learning_rate": 1.2187222714574392e-05,
      "loss": 1.4672,
      "step": 852
    },
    {
      "epoch": 0.4332148298628746,
      "grad_norm": 4.4542927742004395,
      "learning_rate": 1.2171571487378229e-05,
      "loss": 1.2674,
      "step": 853
    },
    {
      "epoch": 0.4337227018791265,
      "grad_norm": 3.5307891368865967,
      "learning_rate": 1.2155914675416964e-05,
      "loss": 1.4065,
      "step": 854
    },
    {
      "epoch": 0.4342305738953784,
      "grad_norm": 3.9772210121154785,
      "learning_rate": 1.2140252318956192e-05,
      "loss": 1.091,
      "step": 855
    },
    {
      "epoch": 0.43473844591163024,
      "grad_norm": 3.50232195854187,
      "learning_rate": 1.2124584458275772e-05,
      "loss": 1.4893,
      "step": 856
    },
    {
      "epoch": 0.43524631792788215,
      "grad_norm": 3.74629807472229,
      "learning_rate": 1.2108911133669706e-05,
      "loss": 1.0373,
      "step": 857
    },
    {
      "epoch": 0.43575418994413406,
      "grad_norm": 3.458662986755371,
      "learning_rate": 1.2093232385446057e-05,
      "loss": 1.3145,
      "step": 858
    },
    {
      "epoch": 0.43626206196038597,
      "grad_norm": 2.9476242065429688,
      "learning_rate": 1.2077548253926834e-05,
      "loss": 1.3316,
      "step": 859
    },
    {
      "epoch": 0.4367699339766379,
      "grad_norm": 3.3274056911468506,
      "learning_rate": 1.2061858779447894e-05,
      "loss": 1.6013,
      "step": 860
    },
    {
      "epoch": 0.4372778059928898,
      "grad_norm": 4.1184797286987305,
      "learning_rate": 1.2046164002358834e-05,
      "loss": 1.4292,
      "step": 861
    },
    {
      "epoch": 0.4377856780091417,
      "grad_norm": 3.278787851333618,
      "learning_rate": 1.2030463963022878e-05,
      "loss": 1.2706,
      "step": 862
    },
    {
      "epoch": 0.4382935500253936,
      "grad_norm": 3.9214258193969727,
      "learning_rate": 1.2014758701816798e-05,
      "loss": 1.3934,
      "step": 863
    },
    {
      "epoch": 0.4388014220416455,
      "grad_norm": 3.6918678283691406,
      "learning_rate": 1.1999048259130782e-05,
      "loss": 1.4788,
      "step": 864
    },
    {
      "epoch": 0.4393092940578974,
      "grad_norm": 3.303007125854492,
      "learning_rate": 1.1983332675368353e-05,
      "loss": 1.5519,
      "step": 865
    },
    {
      "epoch": 0.43981716607414933,
      "grad_norm": 3.0729122161865234,
      "learning_rate": 1.1967611990946254e-05,
      "loss": 1.4969,
      "step": 866
    },
    {
      "epoch": 0.44032503809040124,
      "grad_norm": 3.0283827781677246,
      "learning_rate": 1.1951886246294345e-05,
      "loss": 1.3992,
      "step": 867
    },
    {
      "epoch": 0.4408329101066531,
      "grad_norm": 3.1034414768218994,
      "learning_rate": 1.1936155481855496e-05,
      "loss": 1.3656,
      "step": 868
    },
    {
      "epoch": 0.441340782122905,
      "grad_norm": 5.4229960441589355,
      "learning_rate": 1.1920419738085493e-05,
      "loss": 0.8356,
      "step": 869
    },
    {
      "epoch": 0.4418486541391569,
      "grad_norm": 3.6526145935058594,
      "learning_rate": 1.1904679055452922e-05,
      "loss": 1.3332,
      "step": 870
    },
    {
      "epoch": 0.4423565261554088,
      "grad_norm": 3.88909912109375,
      "learning_rate": 1.1888933474439072e-05,
      "loss": 1.3471,
      "step": 871
    },
    {
      "epoch": 0.44286439817166073,
      "grad_norm": 3.9307477474212646,
      "learning_rate": 1.1873183035537833e-05,
      "loss": 1.2132,
      "step": 872
    },
    {
      "epoch": 0.44337227018791264,
      "grad_norm": 3.914536952972412,
      "learning_rate": 1.1857427779255581e-05,
      "loss": 1.3909,
      "step": 873
    },
    {
      "epoch": 0.44388014220416455,
      "grad_norm": 3.4930009841918945,
      "learning_rate": 1.1841667746111094e-05,
      "loss": 1.1192,
      "step": 874
    },
    {
      "epoch": 0.44438801422041646,
      "grad_norm": 3.1796231269836426,
      "learning_rate": 1.1825902976635422e-05,
      "loss": 1.2825,
      "step": 875
    },
    {
      "epoch": 0.44489588623666837,
      "grad_norm": 4.252844333648682,
      "learning_rate": 1.1810133511371799e-05,
      "loss": 1.2945,
      "step": 876
    },
    {
      "epoch": 0.4454037582529203,
      "grad_norm": 3.7375707626342773,
      "learning_rate": 1.179435939087554e-05,
      "loss": 1.3394,
      "step": 877
    },
    {
      "epoch": 0.4459116302691722,
      "grad_norm": 3.735342264175415,
      "learning_rate": 1.1778580655713928e-05,
      "loss": 1.129,
      "step": 878
    },
    {
      "epoch": 0.4464195022854241,
      "grad_norm": 3.234929084777832,
      "learning_rate": 1.1762797346466115e-05,
      "loss": 1.4438,
      "step": 879
    },
    {
      "epoch": 0.44692737430167595,
      "grad_norm": 0.7645437121391296,
      "learning_rate": 1.1747009503723017e-05,
      "loss": 0.2741,
      "step": 880
    },
    {
      "epoch": 0.44743524631792786,
      "grad_norm": 4.15008020401001,
      "learning_rate": 1.1731217168087207e-05,
      "loss": 1.3858,
      "step": 881
    },
    {
      "epoch": 0.44794311833417977,
      "grad_norm": 3.355151414871216,
      "learning_rate": 1.171542038017282e-05,
      "loss": 1.2419,
      "step": 882
    },
    {
      "epoch": 0.4484509903504317,
      "grad_norm": 3.976581573486328,
      "learning_rate": 1.1699619180605426e-05,
      "loss": 1.2607,
      "step": 883
    },
    {
      "epoch": 0.4489588623666836,
      "grad_norm": 4.275966167449951,
      "learning_rate": 1.1683813610021955e-05,
      "loss": 1.1482,
      "step": 884
    },
    {
      "epoch": 0.4494667343829355,
      "grad_norm": 3.5389294624328613,
      "learning_rate": 1.1668003709070576e-05,
      "loss": 1.3533,
      "step": 885
    },
    {
      "epoch": 0.4499746063991874,
      "grad_norm": 3.413022994995117,
      "learning_rate": 1.1652189518410586e-05,
      "loss": 1.2802,
      "step": 886
    },
    {
      "epoch": 0.4504824784154393,
      "grad_norm": 3.4573328495025635,
      "learning_rate": 1.1636371078712325e-05,
      "loss": 1.3234,
      "step": 887
    },
    {
      "epoch": 0.4509903504316912,
      "grad_norm": 3.2229926586151123,
      "learning_rate": 1.1620548430657052e-05,
      "loss": 1.3755,
      "step": 888
    },
    {
      "epoch": 0.45149822244794313,
      "grad_norm": 3.1548473834991455,
      "learning_rate": 1.1604721614936856e-05,
      "loss": 1.364,
      "step": 889
    },
    {
      "epoch": 0.45200609446419504,
      "grad_norm": 5.584766864776611,
      "learning_rate": 1.158889067225454e-05,
      "loss": 0.7164,
      "step": 890
    },
    {
      "epoch": 0.45251396648044695,
      "grad_norm": 4.067551612854004,
      "learning_rate": 1.1573055643323519e-05,
      "loss": 1.3354,
      "step": 891
    },
    {
      "epoch": 0.45302183849669886,
      "grad_norm": 3.604311943054199,
      "learning_rate": 1.1557216568867724e-05,
      "loss": 1.3106,
      "step": 892
    },
    {
      "epoch": 0.4535297105129507,
      "grad_norm": 3.1507656574249268,
      "learning_rate": 1.1541373489621477e-05,
      "loss": 1.5909,
      "step": 893
    },
    {
      "epoch": 0.4540375825292026,
      "grad_norm": 3.9540703296661377,
      "learning_rate": 1.1525526446329413e-05,
      "loss": 1.1574,
      "step": 894
    },
    {
      "epoch": 0.45454545454545453,
      "grad_norm": 3.316423177719116,
      "learning_rate": 1.1509675479746359e-05,
      "loss": 1.5137,
      "step": 895
    },
    {
      "epoch": 0.45505332656170644,
      "grad_norm": 3.792618751525879,
      "learning_rate": 1.1493820630637222e-05,
      "loss": 1.2956,
      "step": 896
    },
    {
      "epoch": 0.45556119857795835,
      "grad_norm": 3.3896944522857666,
      "learning_rate": 1.1477961939776906e-05,
      "loss": 1.2825,
      "step": 897
    },
    {
      "epoch": 0.45606907059421026,
      "grad_norm": 5.148727893829346,
      "learning_rate": 1.1462099447950192e-05,
      "loss": 0.8357,
      "step": 898
    },
    {
      "epoch": 0.45657694261046217,
      "grad_norm": 3.793823003768921,
      "learning_rate": 1.1446233195951629e-05,
      "loss": 1.0614,
      "step": 899
    },
    {
      "epoch": 0.4570848146267141,
      "grad_norm": 3.341181755065918,
      "learning_rate": 1.1430363224585445e-05,
      "loss": 1.3663,
      "step": 900
    },
    {
      "epoch": 0.457592686642966,
      "grad_norm": 3.8023338317871094,
      "learning_rate": 1.1414489574665424e-05,
      "loss": 1.2573,
      "step": 901
    },
    {
      "epoch": 0.4581005586592179,
      "grad_norm": 3.6153724193573,
      "learning_rate": 1.1398612287014823e-05,
      "loss": 1.3102,
      "step": 902
    },
    {
      "epoch": 0.4586084306754698,
      "grad_norm": 4.726499557495117,
      "learning_rate": 1.1382731402466246e-05,
      "loss": 1.1665,
      "step": 903
    },
    {
      "epoch": 0.4591163026917217,
      "grad_norm": 3.5814390182495117,
      "learning_rate": 1.1366846961861548e-05,
      "loss": 1.4517,
      "step": 904
    },
    {
      "epoch": 0.45962417470797357,
      "grad_norm": 4.046858787536621,
      "learning_rate": 1.135095900605173e-05,
      "loss": 1.2333,
      "step": 905
    },
    {
      "epoch": 0.4601320467242255,
      "grad_norm": 3.346405506134033,
      "learning_rate": 1.1335067575896834e-05,
      "loss": 1.3656,
      "step": 906
    },
    {
      "epoch": 0.4606399187404774,
      "grad_norm": 3.4278366565704346,
      "learning_rate": 1.1319172712265836e-05,
      "loss": 1.3663,
      "step": 907
    },
    {
      "epoch": 0.4611477907567293,
      "grad_norm": 3.110161066055298,
      "learning_rate": 1.1303274456036541e-05,
      "loss": 1.4959,
      "step": 908
    },
    {
      "epoch": 0.4616556627729812,
      "grad_norm": 4.720462799072266,
      "learning_rate": 1.1287372848095485e-05,
      "loss": 1.2964,
      "step": 909
    },
    {
      "epoch": 0.4621635347892331,
      "grad_norm": 3.7709381580352783,
      "learning_rate": 1.1271467929337818e-05,
      "loss": 1.5176,
      "step": 910
    },
    {
      "epoch": 0.462671406805485,
      "grad_norm": 3.399737596511841,
      "learning_rate": 1.1255559740667206e-05,
      "loss": 1.202,
      "step": 911
    },
    {
      "epoch": 0.46317927882173693,
      "grad_norm": 3.852874279022217,
      "learning_rate": 1.1239648322995724e-05,
      "loss": 1.4293,
      "step": 912
    },
    {
      "epoch": 0.46368715083798884,
      "grad_norm": 3.403273344039917,
      "learning_rate": 1.122373371724375e-05,
      "loss": 1.4322,
      "step": 913
    },
    {
      "epoch": 0.46419502285424075,
      "grad_norm": 4.433647155761719,
      "learning_rate": 1.120781596433987e-05,
      "loss": 1.0907,
      "step": 914
    },
    {
      "epoch": 0.46470289487049266,
      "grad_norm": 4.138065338134766,
      "learning_rate": 1.1191895105220747e-05,
      "loss": 1.5683,
      "step": 915
    },
    {
      "epoch": 0.46521076688674456,
      "grad_norm": 3.5094752311706543,
      "learning_rate": 1.1175971180831046e-05,
      "loss": 1.3521,
      "step": 916
    },
    {
      "epoch": 0.4657186389029964,
      "grad_norm": 3.576686143875122,
      "learning_rate": 1.1160044232123313e-05,
      "loss": 1.2947,
      "step": 917
    },
    {
      "epoch": 0.4662265109192483,
      "grad_norm": 3.317537546157837,
      "learning_rate": 1.114411430005787e-05,
      "loss": 1.2871,
      "step": 918
    },
    {
      "epoch": 0.46673438293550024,
      "grad_norm": 3.438308000564575,
      "learning_rate": 1.1128181425602712e-05,
      "loss": 1.3563,
      "step": 919
    },
    {
      "epoch": 0.46724225495175215,
      "grad_norm": 3.2957468032836914,
      "learning_rate": 1.1112245649733393e-05,
      "loss": 1.5019,
      "step": 920
    },
    {
      "epoch": 0.46775012696800405,
      "grad_norm": 3.8559672832489014,
      "learning_rate": 1.1096307013432948e-05,
      "loss": 1.2077,
      "step": 921
    },
    {
      "epoch": 0.46825799898425596,
      "grad_norm": 3.325054407119751,
      "learning_rate": 1.1080365557691743e-05,
      "loss": 1.2988,
      "step": 922
    },
    {
      "epoch": 0.46876587100050787,
      "grad_norm": 3.2300596237182617,
      "learning_rate": 1.1064421323507422e-05,
      "loss": 1.2613,
      "step": 923
    },
    {
      "epoch": 0.4692737430167598,
      "grad_norm": 5.3694963455200195,
      "learning_rate": 1.1048474351884756e-05,
      "loss": 0.9946,
      "step": 924
    },
    {
      "epoch": 0.4697816150330117,
      "grad_norm": 3.384295701980591,
      "learning_rate": 1.1032524683835557e-05,
      "loss": 1.5957,
      "step": 925
    },
    {
      "epoch": 0.4702894870492636,
      "grad_norm": 5.694066047668457,
      "learning_rate": 1.1016572360378579e-05,
      "loss": 0.9679,
      "step": 926
    },
    {
      "epoch": 0.4707973590655155,
      "grad_norm": 3.405916929244995,
      "learning_rate": 1.1000617422539405e-05,
      "loss": 1.4736,
      "step": 927
    },
    {
      "epoch": 0.4713052310817674,
      "grad_norm": 3.8782858848571777,
      "learning_rate": 1.0984659911350335e-05,
      "loss": 1.1301,
      "step": 928
    },
    {
      "epoch": 0.4718131030980193,
      "grad_norm": 3.81451678276062,
      "learning_rate": 1.0968699867850289e-05,
      "loss": 1.3413,
      "step": 929
    },
    {
      "epoch": 0.4723209751142712,
      "grad_norm": 3.7040867805480957,
      "learning_rate": 1.0952737333084704e-05,
      "loss": 1.226,
      "step": 930
    },
    {
      "epoch": 0.4728288471305231,
      "grad_norm": 3.4536123275756836,
      "learning_rate": 1.0936772348105418e-05,
      "loss": 1.4836,
      "step": 931
    },
    {
      "epoch": 0.473336719146775,
      "grad_norm": 4.051872253417969,
      "learning_rate": 1.0920804953970579e-05,
      "loss": 1.2695,
      "step": 932
    },
    {
      "epoch": 0.4738445911630269,
      "grad_norm": 3.3694775104522705,
      "learning_rate": 1.0904835191744515e-05,
      "loss": 1.1816,
      "step": 933
    },
    {
      "epoch": 0.4743524631792788,
      "grad_norm": 3.6979739665985107,
      "learning_rate": 1.0888863102497665e-05,
      "loss": 1.4489,
      "step": 934
    },
    {
      "epoch": 0.4748603351955307,
      "grad_norm": 4.128031253814697,
      "learning_rate": 1.0872888727306434e-05,
      "loss": 1.2585,
      "step": 935
    },
    {
      "epoch": 0.47536820721178263,
      "grad_norm": 3.5181219577789307,
      "learning_rate": 1.0856912107253115e-05,
      "loss": 1.3912,
      "step": 936
    },
    {
      "epoch": 0.47587607922803454,
      "grad_norm": 4.10791015625,
      "learning_rate": 1.0840933283425774e-05,
      "loss": 1.2379,
      "step": 937
    },
    {
      "epoch": 0.47638395124428645,
      "grad_norm": 5.002466201782227,
      "learning_rate": 1.0824952296918146e-05,
      "loss": 1.3092,
      "step": 938
    },
    {
      "epoch": 0.47689182326053836,
      "grad_norm": 3.118469476699829,
      "learning_rate": 1.080896918882952e-05,
      "loss": 1.3928,
      "step": 939
    },
    {
      "epoch": 0.47739969527679027,
      "grad_norm": 4.414337635040283,
      "learning_rate": 1.0792984000264653e-05,
      "loss": 1.0707,
      "step": 940
    },
    {
      "epoch": 0.4779075672930422,
      "grad_norm": 5.438792705535889,
      "learning_rate": 1.077699677233364e-05,
      "loss": 1.0545,
      "step": 941
    },
    {
      "epoch": 0.47841543930929403,
      "grad_norm": 4.211660861968994,
      "learning_rate": 1.0761007546151826e-05,
      "loss": 1.0931,
      "step": 942
    },
    {
      "epoch": 0.47892331132554594,
      "grad_norm": 3.749115467071533,
      "learning_rate": 1.0745016362839704e-05,
      "loss": 1.4633,
      "step": 943
    },
    {
      "epoch": 0.47943118334179785,
      "grad_norm": 4.394846439361572,
      "learning_rate": 1.0729023263522781e-05,
      "loss": 1.2719,
      "step": 944
    },
    {
      "epoch": 0.47993905535804976,
      "grad_norm": 3.9928226470947266,
      "learning_rate": 1.0713028289331515e-05,
      "loss": 1.4197,
      "step": 945
    },
    {
      "epoch": 0.48044692737430167,
      "grad_norm": 3.897555112838745,
      "learning_rate": 1.0697031481401166e-05,
      "loss": 1.1506,
      "step": 946
    },
    {
      "epoch": 0.4809547993905536,
      "grad_norm": 3.1774990558624268,
      "learning_rate": 1.0681032880871716e-05,
      "loss": 1.3888,
      "step": 947
    },
    {
      "epoch": 0.4814626714068055,
      "grad_norm": 3.9464094638824463,
      "learning_rate": 1.0665032528887761e-05,
      "loss": 1.0521,
      "step": 948
    },
    {
      "epoch": 0.4819705434230574,
      "grad_norm": 3.534813642501831,
      "learning_rate": 1.0649030466598398e-05,
      "loss": 1.3523,
      "step": 949
    },
    {
      "epoch": 0.4824784154393093,
      "grad_norm": 3.586951971054077,
      "learning_rate": 1.0633026735157124e-05,
      "loss": 1.1999,
      "step": 950
    },
    {
      "epoch": 0.4829862874555612,
      "grad_norm": 3.499028444290161,
      "learning_rate": 1.0617021375721727e-05,
      "loss": 1.5563,
      "step": 951
    },
    {
      "epoch": 0.4834941594718131,
      "grad_norm": 4.564541339874268,
      "learning_rate": 1.0601014429454184e-05,
      "loss": 1.2579,
      "step": 952
    },
    {
      "epoch": 0.48400203148806503,
      "grad_norm": 3.5675089359283447,
      "learning_rate": 1.0585005937520552e-05,
      "loss": 1.1718,
      "step": 953
    },
    {
      "epoch": 0.4845099035043169,
      "grad_norm": 3.861952781677246,
      "learning_rate": 1.0568995941090858e-05,
      "loss": 1.3146,
      "step": 954
    },
    {
      "epoch": 0.4850177755205688,
      "grad_norm": 3.5122156143188477,
      "learning_rate": 1.0552984481339008e-05,
      "loss": 1.3775,
      "step": 955
    },
    {
      "epoch": 0.4855256475368207,
      "grad_norm": 3.456569194793701,
      "learning_rate": 1.0536971599442667e-05,
      "loss": 1.4182,
      "step": 956
    },
    {
      "epoch": 0.4860335195530726,
      "grad_norm": 3.4301440715789795,
      "learning_rate": 1.052095733658315e-05,
      "loss": 1.3669,
      "step": 957
    },
    {
      "epoch": 0.4865413915693245,
      "grad_norm": 3.3819780349731445,
      "learning_rate": 1.0504941733945337e-05,
      "loss": 1.5379,
      "step": 958
    },
    {
      "epoch": 0.48704926358557643,
      "grad_norm": 3.7181880474090576,
      "learning_rate": 1.0488924832717544e-05,
      "loss": 1.3226,
      "step": 959
    },
    {
      "epoch": 0.48755713560182834,
      "grad_norm": 4.475208759307861,
      "learning_rate": 1.047290667409143e-05,
      "loss": 1.2804,
      "step": 960
    },
    {
      "epoch": 0.48806500761808025,
      "grad_norm": 4.995233535766602,
      "learning_rate": 1.0456887299261891e-05,
      "loss": 1.1545,
      "step": 961
    },
    {
      "epoch": 0.48857287963433216,
      "grad_norm": 4.604562759399414,
      "learning_rate": 1.0440866749426937e-05,
      "loss": 1.2169,
      "step": 962
    },
    {
      "epoch": 0.48908075165058407,
      "grad_norm": 3.4509825706481934,
      "learning_rate": 1.042484506578762e-05,
      "loss": 1.5351,
      "step": 963
    },
    {
      "epoch": 0.489588623666836,
      "grad_norm": 4.223725318908691,
      "learning_rate": 1.040882228954789e-05,
      "loss": 1.2196,
      "step": 964
    },
    {
      "epoch": 0.4900964956830879,
      "grad_norm": 5.32916259765625,
      "learning_rate": 1.0392798461914518e-05,
      "loss": 1.1461,
      "step": 965
    },
    {
      "epoch": 0.4906043676993398,
      "grad_norm": 3.5469343662261963,
      "learning_rate": 1.0376773624096979e-05,
      "loss": 1.2844,
      "step": 966
    },
    {
      "epoch": 0.49111223971559165,
      "grad_norm": 3.920461893081665,
      "learning_rate": 1.0360747817307335e-05,
      "loss": 1.0736,
      "step": 967
    },
    {
      "epoch": 0.49162011173184356,
      "grad_norm": 3.2652604579925537,
      "learning_rate": 1.0344721082760152e-05,
      "loss": 1.2714,
      "step": 968
    },
    {
      "epoch": 0.49212798374809547,
      "grad_norm": 3.248802661895752,
      "learning_rate": 1.0328693461672381e-05,
      "loss": 1.4837,
      "step": 969
    },
    {
      "epoch": 0.4926358557643474,
      "grad_norm": 3.43121600151062,
      "learning_rate": 1.0312664995263242e-05,
      "loss": 1.4444,
      "step": 970
    },
    {
      "epoch": 0.4931437277805993,
      "grad_norm": 3.92547869682312,
      "learning_rate": 1.0296635724754137e-05,
      "loss": 1.3583,
      "step": 971
    },
    {
      "epoch": 0.4936515997968512,
      "grad_norm": 3.1732635498046875,
      "learning_rate": 1.0280605691368542e-05,
      "loss": 1.4998,
      "step": 972
    },
    {
      "epoch": 0.4941594718131031,
      "grad_norm": 3.503230571746826,
      "learning_rate": 1.026457493633188e-05,
      "loss": 1.3833,
      "step": 973
    },
    {
      "epoch": 0.494667343829355,
      "grad_norm": 3.726642608642578,
      "learning_rate": 1.0248543500871449e-05,
      "loss": 1.1362,
      "step": 974
    },
    {
      "epoch": 0.4951752158456069,
      "grad_norm": 3.980401039123535,
      "learning_rate": 1.0232511426216273e-05,
      "loss": 1.4434,
      "step": 975
    },
    {
      "epoch": 0.49568308786185883,
      "grad_norm": 4.425255298614502,
      "learning_rate": 1.021647875359704e-05,
      "loss": 1.2277,
      "step": 976
    },
    {
      "epoch": 0.49619095987811074,
      "grad_norm": 4.443340301513672,
      "learning_rate": 1.020044552424597e-05,
      "loss": 1.2725,
      "step": 977
    },
    {
      "epoch": 0.49669883189436265,
      "grad_norm": 3.401463508605957,
      "learning_rate": 1.018441177939671e-05,
      "loss": 1.4152,
      "step": 978
    },
    {
      "epoch": 0.4972067039106145,
      "grad_norm": 4.191753387451172,
      "learning_rate": 1.0168377560284237e-05,
      "loss": 1.5727,
      "step": 979
    },
    {
      "epoch": 0.4977145759268664,
      "grad_norm": 3.79152512550354,
      "learning_rate": 1.0152342908144747e-05,
      "loss": 1.51,
      "step": 980
    },
    {
      "epoch": 0.4982224479431183,
      "grad_norm": 3.977187395095825,
      "learning_rate": 1.0136307864215552e-05,
      "loss": 1.1533,
      "step": 981
    },
    {
      "epoch": 0.4987303199593702,
      "grad_norm": 3.822143077850342,
      "learning_rate": 1.0120272469734969e-05,
      "loss": 1.3153,
      "step": 982
    },
    {
      "epoch": 0.49923819197562214,
      "grad_norm": 4.221085071563721,
      "learning_rate": 1.010423676594221e-05,
      "loss": 1.0479,
      "step": 983
    },
    {
      "epoch": 0.49974606399187405,
      "grad_norm": 3.6054933071136475,
      "learning_rate": 1.0088200794077296e-05,
      "loss": 1.3151,
      "step": 984
    },
    {
      "epoch": 0.500253936008126,
      "grad_norm": 4.690833568572998,
      "learning_rate": 1.007216459538093e-05,
      "loss": 1.0533,
      "step": 985
    },
    {
      "epoch": 0.5007618080243779,
      "grad_norm": 4.331236839294434,
      "learning_rate": 1.0056128211094395e-05,
      "loss": 1.4805,
      "step": 986
    },
    {
      "epoch": 0.5007618080243779,
      "eval_loss": 1.2702714204788208,
      "eval_runtime": 167.3873,
      "eval_samples_per_second": 5.879,
      "eval_steps_per_second": 1.47,
      "step": 986
    },
    {
      "epoch": 0.5012696800406298,
      "grad_norm": 3.8839874267578125,
      "learning_rate": 1.0040091682459461e-05,
      "loss": 1.4227,
      "step": 987
    },
    {
      "epoch": 0.5017775520568817,
      "grad_norm": 4.005825042724609,
      "learning_rate": 1.0024055050718257e-05,
      "loss": 1.2848,
      "step": 988
    },
    {
      "epoch": 0.5022854240731336,
      "grad_norm": 4.486266613006592,
      "learning_rate": 1.0008018357113187e-05,
      "loss": 1.321,
      "step": 989
    },
    {
      "epoch": 0.5027932960893855,
      "grad_norm": 4.866846084594727,
      "learning_rate": 9.991981642886814e-06,
      "loss": 1.3309,
      "step": 990
    },
    {
      "epoch": 0.5033011681056374,
      "grad_norm": 3.5652294158935547,
      "learning_rate": 9.975944949281748e-06,
      "loss": 1.3894,
      "step": 991
    },
    {
      "epoch": 0.5038090401218893,
      "grad_norm": 4.046600818634033,
      "learning_rate": 9.959908317540542e-06,
      "loss": 1.3046,
      "step": 992
    },
    {
      "epoch": 0.5043169121381412,
      "grad_norm": 3.955585479736328,
      "learning_rate": 9.943871788905606e-06,
      "loss": 1.1924,
      "step": 993
    },
    {
      "epoch": 0.5048247841543931,
      "grad_norm": 3.789905309677124,
      "learning_rate": 9.927835404619075e-06,
      "loss": 1.3519,
      "step": 994
    },
    {
      "epoch": 0.505332656170645,
      "grad_norm": 5.119289875030518,
      "learning_rate": 9.911799205922704e-06,
      "loss": 1.2782,
      "step": 995
    },
    {
      "epoch": 0.505840528186897,
      "grad_norm": 3.122493028640747,
      "learning_rate": 9.895763234057792e-06,
      "loss": 1.5082,
      "step": 996
    },
    {
      "epoch": 0.5063484002031488,
      "grad_norm": 3.873720407485962,
      "learning_rate": 9.879727530265038e-06,
      "loss": 1.2583,
      "step": 997
    },
    {
      "epoch": 0.5068562722194007,
      "grad_norm": 3.268399953842163,
      "learning_rate": 9.863692135784448e-06,
      "loss": 1.4216,
      "step": 998
    },
    {
      "epoch": 0.5073641442356526,
      "grad_norm": 3.9059388637542725,
      "learning_rate": 9.847657091855254e-06,
      "loss": 1.1708,
      "step": 999
    },
    {
      "epoch": 0.5078720162519045,
      "grad_norm": 3.699793577194214,
      "learning_rate": 9.831622439715766e-06,
      "loss": 1.3077,
      "step": 1000
    },
    {
      "epoch": 0.5083798882681564,
      "grad_norm": 3.7766926288604736,
      "learning_rate": 9.815588220603292e-06,
      "loss": 1.5583,
      "step": 1001
    },
    {
      "epoch": 0.5088877602844083,
      "grad_norm": 4.702587127685547,
      "learning_rate": 9.799554475754032e-06,
      "loss": 1.0589,
      "step": 1002
    },
    {
      "epoch": 0.5093956323006602,
      "grad_norm": 3.7550957202911377,
      "learning_rate": 9.783521246402963e-06,
      "loss": 1.0813,
      "step": 1003
    },
    {
      "epoch": 0.5099035043169121,
      "grad_norm": 4.080326557159424,
      "learning_rate": 9.767488573783728e-06,
      "loss": 1.3106,
      "step": 1004
    },
    {
      "epoch": 0.510411376333164,
      "grad_norm": 4.644419193267822,
      "learning_rate": 9.751456499128555e-06,
      "loss": 1.4889,
      "step": 1005
    },
    {
      "epoch": 0.5109192483494159,
      "grad_norm": 3.7005774974823,
      "learning_rate": 9.735425063668121e-06,
      "loss": 1.5305,
      "step": 1006
    },
    {
      "epoch": 0.5114271203656678,
      "grad_norm": 3.4154560565948486,
      "learning_rate": 9.719394308631461e-06,
      "loss": 1.4153,
      "step": 1007
    },
    {
      "epoch": 0.5119349923819198,
      "grad_norm": 4.038942813873291,
      "learning_rate": 9.703364275245864e-06,
      "loss": 1.4484,
      "step": 1008
    },
    {
      "epoch": 0.5124428643981717,
      "grad_norm": 4.789218425750732,
      "learning_rate": 9.687335004736763e-06,
      "loss": 1.3403,
      "step": 1009
    },
    {
      "epoch": 0.5129507364144236,
      "grad_norm": 4.188717842102051,
      "learning_rate": 9.671306538327625e-06,
      "loss": 1.1975,
      "step": 1010
    },
    {
      "epoch": 0.5134586084306755,
      "grad_norm": 3.6356616020202637,
      "learning_rate": 9.655278917239848e-06,
      "loss": 1.3069,
      "step": 1011
    },
    {
      "epoch": 0.5139664804469274,
      "grad_norm": 3.921940565109253,
      "learning_rate": 9.639252182692669e-06,
      "loss": 1.3313,
      "step": 1012
    },
    {
      "epoch": 0.5144743524631793,
      "grad_norm": 3.735460042953491,
      "learning_rate": 9.623226375903028e-06,
      "loss": 1.4116,
      "step": 1013
    },
    {
      "epoch": 0.5149822244794312,
      "grad_norm": 3.5081825256347656,
      "learning_rate": 9.607201538085484e-06,
      "loss": 1.263,
      "step": 1014
    },
    {
      "epoch": 0.5154900964956831,
      "grad_norm": 4.349667549133301,
      "learning_rate": 9.591177710452113e-06,
      "loss": 1.2996,
      "step": 1015
    },
    {
      "epoch": 0.515997968511935,
      "grad_norm": 3.45693039894104,
      "learning_rate": 9.575154934212385e-06,
      "loss": 1.3949,
      "step": 1016
    },
    {
      "epoch": 0.5165058405281869,
      "grad_norm": 4.386706829071045,
      "learning_rate": 9.559133250573062e-06,
      "loss": 1.2486,
      "step": 1017
    },
    {
      "epoch": 0.5170137125444388,
      "grad_norm": 3.824852705001831,
      "learning_rate": 9.543112700738112e-06,
      "loss": 1.6588,
      "step": 1018
    },
    {
      "epoch": 0.5175215845606908,
      "grad_norm": 4.599579811096191,
      "learning_rate": 9.527093325908572e-06,
      "loss": 1.0744,
      "step": 1019
    },
    {
      "epoch": 0.5180294565769427,
      "grad_norm": 4.259639263153076,
      "learning_rate": 9.511075167282459e-06,
      "loss": 1.1343,
      "step": 1020
    },
    {
      "epoch": 0.5185373285931946,
      "grad_norm": 4.185952186584473,
      "learning_rate": 9.495058266054665e-06,
      "loss": 1.1831,
      "step": 1021
    },
    {
      "epoch": 0.5190452006094464,
      "grad_norm": 3.6289265155792236,
      "learning_rate": 9.479042663416852e-06,
      "loss": 1.4118,
      "step": 1022
    },
    {
      "epoch": 0.5195530726256983,
      "grad_norm": 3.4601523876190186,
      "learning_rate": 9.463028400557338e-06,
      "loss": 1.3336,
      "step": 1023
    },
    {
      "epoch": 0.5200609446419502,
      "grad_norm": 4.5509257316589355,
      "learning_rate": 9.447015518660992e-06,
      "loss": 1.0343,
      "step": 1024
    },
    {
      "epoch": 0.5205688166582021,
      "grad_norm": 4.35545015335083,
      "learning_rate": 9.431004058909145e-06,
      "loss": 1.0353,
      "step": 1025
    },
    {
      "epoch": 0.521076688674454,
      "grad_norm": 3.8852431774139404,
      "learning_rate": 9.414994062479455e-06,
      "loss": 1.3905,
      "step": 1026
    },
    {
      "epoch": 0.5215845606907059,
      "grad_norm": 3.812549591064453,
      "learning_rate": 9.398985570545818e-06,
      "loss": 1.4305,
      "step": 1027
    },
    {
      "epoch": 0.5220924327069578,
      "grad_norm": 3.5209968090057373,
      "learning_rate": 9.382978624278276e-06,
      "loss": 1.3956,
      "step": 1028
    },
    {
      "epoch": 0.5226003047232097,
      "grad_norm": 4.683962821960449,
      "learning_rate": 9.366973264842877e-06,
      "loss": 1.3153,
      "step": 1029
    },
    {
      "epoch": 0.5231081767394616,
      "grad_norm": 4.334263324737549,
      "learning_rate": 9.350969533401601e-06,
      "loss": 1.1408,
      "step": 1030
    },
    {
      "epoch": 0.5236160487557135,
      "grad_norm": 3.7997000217437744,
      "learning_rate": 9.33496747111224e-06,
      "loss": 1.5354,
      "step": 1031
    },
    {
      "epoch": 0.5241239207719655,
      "grad_norm": 3.7128489017486572,
      "learning_rate": 9.318967119128289e-06,
      "loss": 1.4863,
      "step": 1032
    },
    {
      "epoch": 0.5246317927882174,
      "grad_norm": 3.7790944576263428,
      "learning_rate": 9.302968518598835e-06,
      "loss": 1.3404,
      "step": 1033
    },
    {
      "epoch": 0.5251396648044693,
      "grad_norm": 4.1233367919921875,
      "learning_rate": 9.286971710668486e-06,
      "loss": 1.1933,
      "step": 1034
    },
    {
      "epoch": 0.5256475368207212,
      "grad_norm": 4.708859443664551,
      "learning_rate": 9.27097673647722e-06,
      "loss": 1.1056,
      "step": 1035
    },
    {
      "epoch": 0.5261554088369731,
      "grad_norm": 3.791539430618286,
      "learning_rate": 9.2549836371603e-06,
      "loss": 1.4778,
      "step": 1036
    },
    {
      "epoch": 0.526663280853225,
      "grad_norm": 4.034219741821289,
      "learning_rate": 9.238992453848173e-06,
      "loss": 1.4119,
      "step": 1037
    },
    {
      "epoch": 0.5271711528694769,
      "grad_norm": 3.3890597820281982,
      "learning_rate": 9.223003227666364e-06,
      "loss": 1.3695,
      "step": 1038
    },
    {
      "epoch": 0.5276790248857288,
      "grad_norm": 3.4671132564544678,
      "learning_rate": 9.207015999735352e-06,
      "loss": 1.5091,
      "step": 1039
    },
    {
      "epoch": 0.5281868969019807,
      "grad_norm": 4.020588397979736,
      "learning_rate": 9.191030811170481e-06,
      "loss": 1.0718,
      "step": 1040
    },
    {
      "epoch": 0.5286947689182326,
      "grad_norm": 3.856794595718384,
      "learning_rate": 9.175047703081857e-06,
      "loss": 1.28,
      "step": 1041
    },
    {
      "epoch": 0.5292026409344845,
      "grad_norm": 3.7805190086364746,
      "learning_rate": 9.159066716574227e-06,
      "loss": 1.3552,
      "step": 1042
    },
    {
      "epoch": 0.5297105129507365,
      "grad_norm": 4.169523239135742,
      "learning_rate": 9.143087892746887e-06,
      "loss": 1.3731,
      "step": 1043
    },
    {
      "epoch": 0.5302183849669884,
      "grad_norm": 4.341207504272461,
      "learning_rate": 9.12711127269357e-06,
      "loss": 1.1904,
      "step": 1044
    },
    {
      "epoch": 0.5307262569832403,
      "grad_norm": 4.3100409507751465,
      "learning_rate": 9.11113689750234e-06,
      "loss": 1.311,
      "step": 1045
    },
    {
      "epoch": 0.5312341289994922,
      "grad_norm": 4.511486053466797,
      "learning_rate": 9.095164808255485e-06,
      "loss": 1.3542,
      "step": 1046
    },
    {
      "epoch": 0.531742001015744,
      "grad_norm": 4.631117343902588,
      "learning_rate": 9.079195046029425e-06,
      "loss": 0.9907,
      "step": 1047
    },
    {
      "epoch": 0.5322498730319959,
      "grad_norm": 3.4736275672912598,
      "learning_rate": 9.063227651894583e-06,
      "loss": 1.3579,
      "step": 1048
    },
    {
      "epoch": 0.5327577450482478,
      "grad_norm": 3.3849706649780273,
      "learning_rate": 9.0472626669153e-06,
      "loss": 1.3443,
      "step": 1049
    },
    {
      "epoch": 0.5332656170644997,
      "grad_norm": 3.844325542449951,
      "learning_rate": 9.031300132149713e-06,
      "loss": 1.3026,
      "step": 1050
    },
    {
      "epoch": 0.5337734890807516,
      "grad_norm": 3.62764835357666,
      "learning_rate": 9.015340088649668e-06,
      "loss": 1.0405,
      "step": 1051
    },
    {
      "epoch": 0.5342813610970035,
      "grad_norm": 3.797156810760498,
      "learning_rate": 8.999382577460598e-06,
      "loss": 1.3491,
      "step": 1052
    },
    {
      "epoch": 0.5347892331132554,
      "grad_norm": 5.074930667877197,
      "learning_rate": 8.983427639621421e-06,
      "loss": 1.1417,
      "step": 1053
    },
    {
      "epoch": 0.5352971051295073,
      "grad_norm": 4.06516170501709,
      "learning_rate": 8.967475316164446e-06,
      "loss": 1.2173,
      "step": 1054
    },
    {
      "epoch": 0.5358049771457593,
      "grad_norm": 3.873654365539551,
      "learning_rate": 8.951525648115251e-06,
      "loss": 1.4384,
      "step": 1055
    },
    {
      "epoch": 0.5363128491620112,
      "grad_norm": 3.887115716934204,
      "learning_rate": 8.93557867649258e-06,
      "loss": 1.3913,
      "step": 1056
    },
    {
      "epoch": 0.5368207211782631,
      "grad_norm": 4.213863372802734,
      "learning_rate": 8.919634442308258e-06,
      "loss": 1.2949,
      "step": 1057
    },
    {
      "epoch": 0.537328593194515,
      "grad_norm": 4.809872627258301,
      "learning_rate": 8.903692986567059e-06,
      "loss": 1.3473,
      "step": 1058
    },
    {
      "epoch": 0.5378364652107669,
      "grad_norm": 4.870054244995117,
      "learning_rate": 8.887754350266608e-06,
      "loss": 1.1729,
      "step": 1059
    },
    {
      "epoch": 0.5383443372270188,
      "grad_norm": 3.5638108253479004,
      "learning_rate": 8.871818574397293e-06,
      "loss": 1.4476,
      "step": 1060
    },
    {
      "epoch": 0.5388522092432707,
      "grad_norm": 4.44373893737793,
      "learning_rate": 8.855885699942134e-06,
      "loss": 1.4085,
      "step": 1061
    },
    {
      "epoch": 0.5393600812595226,
      "grad_norm": 4.298492908477783,
      "learning_rate": 8.839955767876687e-06,
      "loss": 0.9469,
      "step": 1062
    },
    {
      "epoch": 0.5398679532757745,
      "grad_norm": 4.122191905975342,
      "learning_rate": 8.824028819168955e-06,
      "loss": 1.1266,
      "step": 1063
    },
    {
      "epoch": 0.5403758252920264,
      "grad_norm": 3.8323357105255127,
      "learning_rate": 8.808104894779257e-06,
      "loss": 1.4753,
      "step": 1064
    },
    {
      "epoch": 0.5408836973082783,
      "grad_norm": 3.6288328170776367,
      "learning_rate": 8.792184035660136e-06,
      "loss": 1.3259,
      "step": 1065
    },
    {
      "epoch": 0.5413915693245303,
      "grad_norm": 3.873750925064087,
      "learning_rate": 8.77626628275625e-06,
      "loss": 1.4324,
      "step": 1066
    },
    {
      "epoch": 0.5418994413407822,
      "grad_norm": 5.460947036743164,
      "learning_rate": 8.76035167700428e-06,
      "loss": 1.3701,
      "step": 1067
    },
    {
      "epoch": 0.5424073133570341,
      "grad_norm": 4.058062553405762,
      "learning_rate": 8.744440259332798e-06,
      "loss": 1.1698,
      "step": 1068
    },
    {
      "epoch": 0.542915185373286,
      "grad_norm": 3.992889642715454,
      "learning_rate": 8.728532070662184e-06,
      "loss": 1.3678,
      "step": 1069
    },
    {
      "epoch": 0.5434230573895379,
      "grad_norm": 3.4362125396728516,
      "learning_rate": 8.712627151904518e-06,
      "loss": 1.363,
      "step": 1070
    },
    {
      "epoch": 0.5439309294057897,
      "grad_norm": 4.388937473297119,
      "learning_rate": 8.696725543963462e-06,
      "loss": 1.3305,
      "step": 1071
    },
    {
      "epoch": 0.5444388014220416,
      "grad_norm": 3.9520606994628906,
      "learning_rate": 8.680827287734167e-06,
      "loss": 1.2707,
      "step": 1072
    },
    {
      "epoch": 0.5449466734382935,
      "grad_norm": 5.032470226287842,
      "learning_rate": 8.66493242410317e-06,
      "loss": 1.1903,
      "step": 1073
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 4.099115371704102,
      "learning_rate": 8.649040993948273e-06,
      "loss": 1.4908,
      "step": 1074
    },
    {
      "epoch": 0.5459624174707973,
      "grad_norm": 4.2181549072265625,
      "learning_rate": 8.633153038138453e-06,
      "loss": 1.3798,
      "step": 1075
    },
    {
      "epoch": 0.5464702894870492,
      "grad_norm": 3.630211114883423,
      "learning_rate": 8.617268597533755e-06,
      "loss": 1.3977,
      "step": 1076
    },
    {
      "epoch": 0.5469781615033011,
      "grad_norm": 3.997370481491089,
      "learning_rate": 8.60138771298518e-06,
      "loss": 1.3037,
      "step": 1077
    },
    {
      "epoch": 0.547486033519553,
      "grad_norm": 4.776542663574219,
      "learning_rate": 8.585510425334579e-06,
      "loss": 1.4208,
      "step": 1078
    },
    {
      "epoch": 0.547993905535805,
      "grad_norm": 4.408464431762695,
      "learning_rate": 8.569636775414558e-06,
      "loss": 1.5326,
      "step": 1079
    },
    {
      "epoch": 0.5485017775520569,
      "grad_norm": 3.441190719604492,
      "learning_rate": 8.553766804048373e-06,
      "loss": 1.6373,
      "step": 1080
    },
    {
      "epoch": 0.5490096495683088,
      "grad_norm": 4.0465593338012695,
      "learning_rate": 8.537900552049812e-06,
      "loss": 1.2693,
      "step": 1081
    },
    {
      "epoch": 0.5495175215845607,
      "grad_norm": 4.380228042602539,
      "learning_rate": 8.522038060223094e-06,
      "loss": 1.4583,
      "step": 1082
    },
    {
      "epoch": 0.5500253936008126,
      "grad_norm": 3.919779062271118,
      "learning_rate": 8.50617936936278e-06,
      "loss": 1.413,
      "step": 1083
    },
    {
      "epoch": 0.5505332656170645,
      "grad_norm": 4.072299480438232,
      "learning_rate": 8.490324520253648e-06,
      "loss": 1.3373,
      "step": 1084
    },
    {
      "epoch": 0.5510411376333164,
      "grad_norm": 3.893017530441284,
      "learning_rate": 8.474473553670588e-06,
      "loss": 1.3729,
      "step": 1085
    },
    {
      "epoch": 0.5515490096495683,
      "grad_norm": 3.743358850479126,
      "learning_rate": 8.458626510378525e-06,
      "loss": 1.1618,
      "step": 1086
    },
    {
      "epoch": 0.5520568816658202,
      "grad_norm": 3.8513174057006836,
      "learning_rate": 8.442783431132282e-06,
      "loss": 1.2422,
      "step": 1087
    },
    {
      "epoch": 0.5525647536820721,
      "grad_norm": 3.660341501235962,
      "learning_rate": 8.426944356676481e-06,
      "loss": 1.4136,
      "step": 1088
    },
    {
      "epoch": 0.553072625698324,
      "grad_norm": 4.450745105743408,
      "learning_rate": 8.411109327745463e-06,
      "loss": 1.3378,
      "step": 1089
    },
    {
      "epoch": 0.553580497714576,
      "grad_norm": 3.3938448429107666,
      "learning_rate": 8.395278385063147e-06,
      "loss": 1.2288,
      "step": 1090
    },
    {
      "epoch": 0.5540883697308279,
      "grad_norm": 3.7265074253082275,
      "learning_rate": 8.379451569342948e-06,
      "loss": 1.2431,
      "step": 1091
    },
    {
      "epoch": 0.5545962417470798,
      "grad_norm": 3.6786811351776123,
      "learning_rate": 8.363628921287678e-06,
      "loss": 1.3273,
      "step": 1092
    },
    {
      "epoch": 0.5551041137633317,
      "grad_norm": 3.631423234939575,
      "learning_rate": 8.347810481589416e-06,
      "loss": 1.4747,
      "step": 1093
    },
    {
      "epoch": 0.5556119857795836,
      "grad_norm": 3.7244248390197754,
      "learning_rate": 8.331996290929429e-06,
      "loss": 1.5013,
      "step": 1094
    },
    {
      "epoch": 0.5561198577958355,
      "grad_norm": 3.9098987579345703,
      "learning_rate": 8.316186389978045e-06,
      "loss": 1.4644,
      "step": 1095
    },
    {
      "epoch": 0.5566277298120873,
      "grad_norm": 3.656759738922119,
      "learning_rate": 8.300380819394579e-06,
      "loss": 1.5651,
      "step": 1096
    },
    {
      "epoch": 0.5571356018283392,
      "grad_norm": 4.050074577331543,
      "learning_rate": 8.284579619827187e-06,
      "loss": 1.3435,
      "step": 1097
    },
    {
      "epoch": 0.5576434738445911,
      "grad_norm": 4.784929275512695,
      "learning_rate": 8.268782831912793e-06,
      "loss": 1.2313,
      "step": 1098
    },
    {
      "epoch": 0.558151345860843,
      "grad_norm": 4.976466178894043,
      "learning_rate": 8.252990496276986e-06,
      "loss": 1.264,
      "step": 1099
    },
    {
      "epoch": 0.5586592178770949,
      "grad_norm": 4.2961320877075195,
      "learning_rate": 8.237202653533886e-06,
      "loss": 1.1136,
      "step": 1100
    },
    {
      "epoch": 0.5591670898933468,
      "grad_norm": 3.6020753383636475,
      "learning_rate": 8.221419344286073e-06,
      "loss": 1.2281,
      "step": 1101
    },
    {
      "epoch": 0.5596749619095988,
      "grad_norm": 3.858781337738037,
      "learning_rate": 8.205640609124461e-06,
      "loss": 1.2915,
      "step": 1102
    },
    {
      "epoch": 0.5601828339258507,
      "grad_norm": 3.9297633171081543,
      "learning_rate": 8.189866488628204e-06,
      "loss": 1.3519,
      "step": 1103
    },
    {
      "epoch": 0.5606907059421026,
      "grad_norm": 3.6003174781799316,
      "learning_rate": 8.17409702336458e-06,
      "loss": 1.514,
      "step": 1104
    },
    {
      "epoch": 0.5611985779583545,
      "grad_norm": 4.255648612976074,
      "learning_rate": 8.158332253888907e-06,
      "loss": 1.3738,
      "step": 1105
    },
    {
      "epoch": 0.5617064499746064,
      "grad_norm": 4.09144401550293,
      "learning_rate": 8.142572220744422e-06,
      "loss": 1.1667,
      "step": 1106
    },
    {
      "epoch": 0.5622143219908583,
      "grad_norm": 3.893601417541504,
      "learning_rate": 8.126816964462172e-06,
      "loss": 1.3337,
      "step": 1107
    },
    {
      "epoch": 0.5627221940071102,
      "grad_norm": 4.821288585662842,
      "learning_rate": 8.11106652556093e-06,
      "loss": 1.2159,
      "step": 1108
    },
    {
      "epoch": 0.5632300660233621,
      "grad_norm": 3.5883355140686035,
      "learning_rate": 8.095320944547083e-06,
      "loss": 1.2955,
      "step": 1109
    },
    {
      "epoch": 0.563737938039614,
      "grad_norm": 4.82534646987915,
      "learning_rate": 8.079580261914512e-06,
      "loss": 1.507,
      "step": 1110
    },
    {
      "epoch": 0.5642458100558659,
      "grad_norm": 3.767213821411133,
      "learning_rate": 8.063844518144505e-06,
      "loss": 1.3607,
      "step": 1111
    },
    {
      "epoch": 0.5647536820721178,
      "grad_norm": 3.960946559906006,
      "learning_rate": 8.048113753705657e-06,
      "loss": 1.0787,
      "step": 1112
    },
    {
      "epoch": 0.5652615540883698,
      "grad_norm": 3.8261797428131104,
      "learning_rate": 8.032388009053747e-06,
      "loss": 1.3855,
      "step": 1113
    },
    {
      "epoch": 0.5657694261046217,
      "grad_norm": 3.765634298324585,
      "learning_rate": 8.016667324631649e-06,
      "loss": 1.1932,
      "step": 1114
    },
    {
      "epoch": 0.5662772981208736,
      "grad_norm": 4.079648017883301,
      "learning_rate": 8.00095174086922e-06,
      "loss": 1.3525,
      "step": 1115
    },
    {
      "epoch": 0.5667851701371255,
      "grad_norm": 4.009759426116943,
      "learning_rate": 7.985241298183209e-06,
      "loss": 1.3184,
      "step": 1116
    },
    {
      "epoch": 0.5672930421533774,
      "grad_norm": 4.056725978851318,
      "learning_rate": 7.969536036977124e-06,
      "loss": 1.3591,
      "step": 1117
    },
    {
      "epoch": 0.5678009141696293,
      "grad_norm": 5.053130149841309,
      "learning_rate": 7.95383599764117e-06,
      "loss": 1.3316,
      "step": 1118
    },
    {
      "epoch": 0.5683087861858812,
      "grad_norm": 3.995842933654785,
      "learning_rate": 7.938141220552107e-06,
      "loss": 1.4236,
      "step": 1119
    },
    {
      "epoch": 0.568816658202133,
      "grad_norm": 3.60331654548645,
      "learning_rate": 7.922451746073166e-06,
      "loss": 1.3767,
      "step": 1120
    },
    {
      "epoch": 0.5693245302183849,
      "grad_norm": 4.3413238525390625,
      "learning_rate": 7.906767614553947e-06,
      "loss": 1.1674,
      "step": 1121
    },
    {
      "epoch": 0.5698324022346368,
      "grad_norm": 4.532721042633057,
      "learning_rate": 7.8910888663303e-06,
      "loss": 1.4166,
      "step": 1122
    },
    {
      "epoch": 0.5703402742508887,
      "grad_norm": 4.4201765060424805,
      "learning_rate": 7.875415541724235e-06,
      "loss": 1.1205,
      "step": 1123
    },
    {
      "epoch": 0.5708481462671406,
      "grad_norm": 4.706449031829834,
      "learning_rate": 7.859747681043808e-06,
      "loss": 1.3193,
      "step": 1124
    },
    {
      "epoch": 0.5713560182833926,
      "grad_norm": 5.1450982093811035,
      "learning_rate": 7.84408532458304e-06,
      "loss": 1.3136,
      "step": 1125
    },
    {
      "epoch": 0.5718638902996445,
      "grad_norm": 3.987536668777466,
      "learning_rate": 7.828428512621776e-06,
      "loss": 1.2301,
      "step": 1126
    },
    {
      "epoch": 0.5723717623158964,
      "grad_norm": 4.997498035430908,
      "learning_rate": 7.812777285425608e-06,
      "loss": 1.124,
      "step": 1127
    },
    {
      "epoch": 0.5728796343321483,
      "grad_norm": 4.351808547973633,
      "learning_rate": 7.797131683245776e-06,
      "loss": 1.2459,
      "step": 1128
    },
    {
      "epoch": 0.5733875063484002,
      "grad_norm": 4.245429992675781,
      "learning_rate": 7.781491746319045e-06,
      "loss": 1.1971,
      "step": 1129
    },
    {
      "epoch": 0.5738953783646521,
      "grad_norm": 4.84749698638916,
      "learning_rate": 7.765857514867603e-06,
      "loss": 1.2181,
      "step": 1130
    },
    {
      "epoch": 0.574403250380904,
      "grad_norm": 3.8819117546081543,
      "learning_rate": 7.750229029098982e-06,
      "loss": 1.3041,
      "step": 1131
    },
    {
      "epoch": 0.5749111223971559,
      "grad_norm": 4.279530048370361,
      "learning_rate": 7.734606329205923e-06,
      "loss": 1.2113,
      "step": 1132
    },
    {
      "epoch": 0.5754189944134078,
      "grad_norm": 5.590778827667236,
      "learning_rate": 7.71898945536629e-06,
      "loss": 1.246,
      "step": 1133
    },
    {
      "epoch": 0.5759268664296597,
      "grad_norm": 3.8742496967315674,
      "learning_rate": 7.703378447742976e-06,
      "loss": 1.3175,
      "step": 1134
    },
    {
      "epoch": 0.5764347384459116,
      "grad_norm": 4.623325347900391,
      "learning_rate": 7.687773346483766e-06,
      "loss": 1.3656,
      "step": 1135
    },
    {
      "epoch": 0.5769426104621636,
      "grad_norm": 7.073203086853027,
      "learning_rate": 7.672174191721277e-06,
      "loss": 1.1737,
      "step": 1136
    },
    {
      "epoch": 0.5774504824784155,
      "grad_norm": 4.289577007293701,
      "learning_rate": 7.656581023572814e-06,
      "loss": 1.3797,
      "step": 1137
    },
    {
      "epoch": 0.5779583544946674,
      "grad_norm": 3.823345422744751,
      "learning_rate": 7.640993882140302e-06,
      "loss": 1.1435,
      "step": 1138
    },
    {
      "epoch": 0.5784662265109193,
      "grad_norm": 3.539424180984497,
      "learning_rate": 7.625412807510157e-06,
      "loss": 1.3426,
      "step": 1139
    },
    {
      "epoch": 0.5789740985271712,
      "grad_norm": 4.083397388458252,
      "learning_rate": 7.609837839753191e-06,
      "loss": 1.5188,
      "step": 1140
    },
    {
      "epoch": 0.5794819705434231,
      "grad_norm": 3.440096855163574,
      "learning_rate": 7.594269018924522e-06,
      "loss": 1.4336,
      "step": 1141
    },
    {
      "epoch": 0.579989842559675,
      "grad_norm": 4.4859490394592285,
      "learning_rate": 7.5787063850634476e-06,
      "loss": 1.226,
      "step": 1142
    },
    {
      "epoch": 0.5804977145759269,
      "grad_norm": 4.9387617111206055,
      "learning_rate": 7.56314997819336e-06,
      "loss": 0.9405,
      "step": 1143
    },
    {
      "epoch": 0.5810055865921788,
      "grad_norm": 3.6573925018310547,
      "learning_rate": 7.547599838321635e-06,
      "loss": 1.3936,
      "step": 1144
    },
    {
      "epoch": 0.5815134586084306,
      "grad_norm": 5.386067867279053,
      "learning_rate": 7.532056005439536e-06,
      "loss": 0.9549,
      "step": 1145
    },
    {
      "epoch": 0.5820213306246825,
      "grad_norm": 4.03516149520874,
      "learning_rate": 7.516518519522093e-06,
      "loss": 1.3277,
      "step": 1146
    },
    {
      "epoch": 0.5825292026409344,
      "grad_norm": 5.966615200042725,
      "learning_rate": 7.500987420528033e-06,
      "loss": 1.1483,
      "step": 1147
    },
    {
      "epoch": 0.5830370746571863,
      "grad_norm": 3.9208691120147705,
      "learning_rate": 7.485462748399641e-06,
      "loss": 1.1712,
      "step": 1148
    },
    {
      "epoch": 0.5835449466734383,
      "grad_norm": 3.6072921752929688,
      "learning_rate": 7.469944543062677e-06,
      "loss": 1.5522,
      "step": 1149
    },
    {
      "epoch": 0.5840528186896902,
      "grad_norm": 4.357779502868652,
      "learning_rate": 7.454432844426278e-06,
      "loss": 1.2472,
      "step": 1150
    },
    {
      "epoch": 0.5845606907059421,
      "grad_norm": 4.2969465255737305,
      "learning_rate": 7.43892769238284e-06,
      "loss": 1.6309,
      "step": 1151
    },
    {
      "epoch": 0.585068562722194,
      "grad_norm": 4.970481872558594,
      "learning_rate": 7.423429126807924e-06,
      "loss": 1.0273,
      "step": 1152
    },
    {
      "epoch": 0.5855764347384459,
      "grad_norm": 4.304748058319092,
      "learning_rate": 7.407937187560148e-06,
      "loss": 1.2925,
      "step": 1153
    },
    {
      "epoch": 0.5860843067546978,
      "grad_norm": 3.9706506729125977,
      "learning_rate": 7.3924519144811e-06,
      "loss": 1.3122,
      "step": 1154
    },
    {
      "epoch": 0.5865921787709497,
      "grad_norm": 3.9953420162200928,
      "learning_rate": 7.376973347395215e-06,
      "loss": 1.1798,
      "step": 1155
    },
    {
      "epoch": 0.5871000507872016,
      "grad_norm": 4.054521083831787,
      "learning_rate": 7.36150152610968e-06,
      "loss": 1.5718,
      "step": 1156
    },
    {
      "epoch": 0.5876079228034535,
      "grad_norm": 3.4766275882720947,
      "learning_rate": 7.346036490414342e-06,
      "loss": 1.3719,
      "step": 1157
    },
    {
      "epoch": 0.5881157948197054,
      "grad_norm": 5.536925315856934,
      "learning_rate": 7.330578280081595e-06,
      "loss": 1.3112,
      "step": 1158
    },
    {
      "epoch": 0.5886236668359573,
      "grad_norm": 3.708730459213257,
      "learning_rate": 7.315126934866269e-06,
      "loss": 1.2901,
      "step": 1159
    },
    {
      "epoch": 0.5891315388522093,
      "grad_norm": 4.949906349182129,
      "learning_rate": 7.299682494505557e-06,
      "loss": 1.1044,
      "step": 1160
    },
    {
      "epoch": 0.5896394108684612,
      "grad_norm": 4.265493392944336,
      "learning_rate": 7.284244998718879e-06,
      "loss": 1.2927,
      "step": 1161
    },
    {
      "epoch": 0.5901472828847131,
      "grad_norm": 4.239290714263916,
      "learning_rate": 7.268814487207796e-06,
      "loss": 1.083,
      "step": 1162
    },
    {
      "epoch": 0.590655154900965,
      "grad_norm": 4.395908355712891,
      "learning_rate": 7.253390999655923e-06,
      "loss": 1.2914,
      "step": 1163
    },
    {
      "epoch": 0.5911630269172169,
      "grad_norm": 4.987044334411621,
      "learning_rate": 7.237974575728788e-06,
      "loss": 1.4456,
      "step": 1164
    },
    {
      "epoch": 0.5916708989334688,
      "grad_norm": 4.285270690917969,
      "learning_rate": 7.222565255073775e-06,
      "loss": 1.3697,
      "step": 1165
    },
    {
      "epoch": 0.5921787709497207,
      "grad_norm": 4.593809127807617,
      "learning_rate": 7.2071630773199786e-06,
      "loss": 1.3608,
      "step": 1166
    },
    {
      "epoch": 0.5926866429659726,
      "grad_norm": 4.617862224578857,
      "learning_rate": 7.191768082078145e-06,
      "loss": 1.3649,
      "step": 1167
    },
    {
      "epoch": 0.5931945149822245,
      "grad_norm": 3.5487444400787354,
      "learning_rate": 7.176380308940536e-06,
      "loss": 1.3859,
      "step": 1168
    },
    {
      "epoch": 0.5937023869984764,
      "grad_norm": 3.997659921646118,
      "learning_rate": 7.160999797480834e-06,
      "loss": 1.4197,
      "step": 1169
    },
    {
      "epoch": 0.5942102590147282,
      "grad_norm": 4.572630405426025,
      "learning_rate": 7.145626587254069e-06,
      "loss": 1.1834,
      "step": 1170
    },
    {
      "epoch": 0.5947181310309801,
      "grad_norm": 4.793776988983154,
      "learning_rate": 7.130260717796467e-06,
      "loss": 1.2255,
      "step": 1171
    },
    {
      "epoch": 0.595226003047232,
      "grad_norm": 4.291995525360107,
      "learning_rate": 7.1149022286253884e-06,
      "loss": 1.181,
      "step": 1172
    },
    {
      "epoch": 0.595733875063484,
      "grad_norm": 3.9128329753875732,
      "learning_rate": 7.0995511592392234e-06,
      "loss": 1.4642,
      "step": 1173
    },
    {
      "epoch": 0.5962417470797359,
      "grad_norm": 3.90106463432312,
      "learning_rate": 7.084207549117262e-06,
      "loss": 1.4408,
      "step": 1174
    },
    {
      "epoch": 0.5967496190959878,
      "grad_norm": 4.521573066711426,
      "learning_rate": 7.068871437719611e-06,
      "loss": 1.2979,
      "step": 1175
    },
    {
      "epoch": 0.5972574911122397,
      "grad_norm": 4.508446216583252,
      "learning_rate": 7.0535428644871125e-06,
      "loss": 1.485,
      "step": 1176
    },
    {
      "epoch": 0.5977653631284916,
      "grad_norm": 3.9106104373931885,
      "learning_rate": 7.038221868841205e-06,
      "loss": 1.2453,
      "step": 1177
    },
    {
      "epoch": 0.5982732351447435,
      "grad_norm": 4.619490146636963,
      "learning_rate": 7.022908490183836e-06,
      "loss": 1.051,
      "step": 1178
    },
    {
      "epoch": 0.5987811071609954,
      "grad_norm": 5.208146095275879,
      "learning_rate": 7.00760276789738e-06,
      "loss": 1.1115,
      "step": 1179
    },
    {
      "epoch": 0.5992889791772473,
      "grad_norm": 4.537458419799805,
      "learning_rate": 6.992304741344511e-06,
      "loss": 1.5648,
      "step": 1180
    },
    {
      "epoch": 0.5997968511934992,
      "grad_norm": 4.677596569061279,
      "learning_rate": 6.977014449868113e-06,
      "loss": 1.5071,
      "step": 1181
    },
    {
      "epoch": 0.6003047232097511,
      "grad_norm": 4.3316850662231445,
      "learning_rate": 6.961731932791169e-06,
      "loss": 1.2314,
      "step": 1182
    },
    {
      "epoch": 0.600812595226003,
      "grad_norm": 4.620197772979736,
      "learning_rate": 6.946457229416688e-06,
      "loss": 1.1893,
      "step": 1183
    },
    {
      "epoch": 0.601320467242255,
      "grad_norm": 4.2192277908325195,
      "learning_rate": 6.931190379027565e-06,
      "loss": 1.407,
      "step": 1184
    },
    {
      "epoch": 0.6018283392585069,
      "grad_norm": 4.571565628051758,
      "learning_rate": 6.9159314208865066e-06,
      "loss": 1.4941,
      "step": 1185
    },
    {
      "epoch": 0.6023362112747588,
      "grad_norm": 4.276565074920654,
      "learning_rate": 6.900680394235925e-06,
      "loss": 1.4298,
      "step": 1186
    },
    {
      "epoch": 0.6028440832910107,
      "grad_norm": 4.913404941558838,
      "learning_rate": 6.885437338297831e-06,
      "loss": 1.5027,
      "step": 1187
    },
    {
      "epoch": 0.6033519553072626,
      "grad_norm": 4.725199222564697,
      "learning_rate": 6.870202292273732e-06,
      "loss": 1.1943,
      "step": 1188
    },
    {
      "epoch": 0.6038598273235145,
      "grad_norm": 4.427515983581543,
      "learning_rate": 6.8549752953445496e-06,
      "loss": 1.2846,
      "step": 1189
    },
    {
      "epoch": 0.6043676993397664,
      "grad_norm": 4.1241254806518555,
      "learning_rate": 6.839756386670491e-06,
      "loss": 1.382,
      "step": 1190
    },
    {
      "epoch": 0.6048755713560183,
      "grad_norm": 4.1516571044921875,
      "learning_rate": 6.824545605390964e-06,
      "loss": 1.2045,
      "step": 1191
    },
    {
      "epoch": 0.6053834433722702,
      "grad_norm": 4.107656002044678,
      "learning_rate": 6.809342990624488e-06,
      "loss": 1.3974,
      "step": 1192
    },
    {
      "epoch": 0.6058913153885221,
      "grad_norm": 4.163454055786133,
      "learning_rate": 6.794148581468562e-06,
      "loss": 1.1562,
      "step": 1193
    },
    {
      "epoch": 0.6063991874047739,
      "grad_norm": 3.9578378200531006,
      "learning_rate": 6.778962416999596e-06,
      "loss": 1.1857,
      "step": 1194
    },
    {
      "epoch": 0.6069070594210259,
      "grad_norm": 3.630510091781616,
      "learning_rate": 6.763784536272785e-06,
      "loss": 1.1776,
      "step": 1195
    },
    {
      "epoch": 0.6074149314372778,
      "grad_norm": 4.183380126953125,
      "learning_rate": 6.748614978322034e-06,
      "loss": 1.3678,
      "step": 1196
    },
    {
      "epoch": 0.6079228034535297,
      "grad_norm": 4.871096134185791,
      "learning_rate": 6.733453782159834e-06,
      "loss": 0.7445,
      "step": 1197
    },
    {
      "epoch": 0.6084306754697816,
      "grad_norm": 4.2791876792907715,
      "learning_rate": 6.718300986777167e-06,
      "loss": 1.2649,
      "step": 1198
    },
    {
      "epoch": 0.6089385474860335,
      "grad_norm": 4.809388637542725,
      "learning_rate": 6.7031566311434225e-06,
      "loss": 1.2088,
      "step": 1199
    },
    {
      "epoch": 0.6094464195022854,
      "grad_norm": 4.801436901092529,
      "learning_rate": 6.688020754206283e-06,
      "loss": 1.0736,
      "step": 1200
    },
    {
      "epoch": 0.6099542915185373,
      "grad_norm": 4.411201000213623,
      "learning_rate": 6.672893394891612e-06,
      "loss": 1.5282,
      "step": 1201
    },
    {
      "epoch": 0.6104621635347892,
      "grad_norm": 5.016659259796143,
      "learning_rate": 6.657774592103389e-06,
      "loss": 1.1121,
      "step": 1202
    },
    {
      "epoch": 0.6109700355510411,
      "grad_norm": 4.217753887176514,
      "learning_rate": 6.6426643847235715e-06,
      "loss": 1.4342,
      "step": 1203
    },
    {
      "epoch": 0.611477907567293,
      "grad_norm": 3.952265501022339,
      "learning_rate": 6.627562811612012e-06,
      "loss": 1.3977,
      "step": 1204
    },
    {
      "epoch": 0.6119857795835449,
      "grad_norm": 3.797877550125122,
      "learning_rate": 6.612469911606374e-06,
      "loss": 1.4424,
      "step": 1205
    },
    {
      "epoch": 0.6124936515997969,
      "grad_norm": 3.7128665447235107,
      "learning_rate": 6.5973857235219965e-06,
      "loss": 1.1818,
      "step": 1206
    },
    {
      "epoch": 0.6130015236160488,
      "grad_norm": 4.827739715576172,
      "learning_rate": 6.582310286151824e-06,
      "loss": 1.4196,
      "step": 1207
    },
    {
      "epoch": 0.6135093956323007,
      "grad_norm": 3.766799211502075,
      "learning_rate": 6.567243638266294e-06,
      "loss": 1.5127,
      "step": 1208
    },
    {
      "epoch": 0.6140172676485526,
      "grad_norm": 4.022468566894531,
      "learning_rate": 6.552185818613241e-06,
      "loss": 1.2584,
      "step": 1209
    },
    {
      "epoch": 0.6145251396648045,
      "grad_norm": 3.6558563709259033,
      "learning_rate": 6.537136865917792e-06,
      "loss": 1.4567,
      "step": 1210
    },
    {
      "epoch": 0.6150330116810564,
      "grad_norm": 4.016808986663818,
      "learning_rate": 6.5220968188822686e-06,
      "loss": 1.5444,
      "step": 1211
    },
    {
      "epoch": 0.6155408836973083,
      "grad_norm": 4.2901153564453125,
      "learning_rate": 6.507065716186102e-06,
      "loss": 1.1884,
      "step": 1212
    },
    {
      "epoch": 0.6160487557135602,
      "grad_norm": 4.470537185668945,
      "learning_rate": 6.492043596485703e-06,
      "loss": 1.1808,
      "step": 1213
    },
    {
      "epoch": 0.6165566277298121,
      "grad_norm": 4.94361686706543,
      "learning_rate": 6.477030498414392e-06,
      "loss": 1.451,
      "step": 1214
    },
    {
      "epoch": 0.617064499746064,
      "grad_norm": 4.530747890472412,
      "learning_rate": 6.462026460582284e-06,
      "loss": 1.333,
      "step": 1215
    },
    {
      "epoch": 0.6175723717623159,
      "grad_norm": 4.878780841827393,
      "learning_rate": 6.447031521576197e-06,
      "loss": 1.3113,
      "step": 1216
    },
    {
      "epoch": 0.6180802437785679,
      "grad_norm": 4.026434898376465,
      "learning_rate": 6.4320457199595375e-06,
      "loss": 1.197,
      "step": 1217
    },
    {
      "epoch": 0.6185881157948198,
      "grad_norm": 4.1217570304870605,
      "learning_rate": 6.417069094272231e-06,
      "loss": 1.3979,
      "step": 1218
    },
    {
      "epoch": 0.6190959878110716,
      "grad_norm": 4.3459930419921875,
      "learning_rate": 6.40210168303059e-06,
      "loss": 1.4266,
      "step": 1219
    },
    {
      "epoch": 0.6196038598273235,
      "grad_norm": 4.097601413726807,
      "learning_rate": 6.387143524727233e-06,
      "loss": 1.4268,
      "step": 1220
    },
    {
      "epoch": 0.6201117318435754,
      "grad_norm": 3.8987221717834473,
      "learning_rate": 6.372194657830986e-06,
      "loss": 1.2839,
      "step": 1221
    },
    {
      "epoch": 0.6206196038598273,
      "grad_norm": 4.2752299308776855,
      "learning_rate": 6.357255120786778e-06,
      "loss": 1.254,
      "step": 1222
    },
    {
      "epoch": 0.6211274758760792,
      "grad_norm": 4.806422233581543,
      "learning_rate": 6.3423249520155415e-06,
      "loss": 1.3237,
      "step": 1223
    },
    {
      "epoch": 0.6216353478923311,
      "grad_norm": 3.7665746212005615,
      "learning_rate": 6.327404189914115e-06,
      "loss": 1.5077,
      "step": 1224
    },
    {
      "epoch": 0.622143219908583,
      "grad_norm": 4.326169490814209,
      "learning_rate": 6.312492872855156e-06,
      "loss": 1.1688,
      "step": 1225
    },
    {
      "epoch": 0.6226510919248349,
      "grad_norm": 4.449665069580078,
      "learning_rate": 6.297591039187016e-06,
      "loss": 1.2381,
      "step": 1226
    },
    {
      "epoch": 0.6231589639410868,
      "grad_norm": 4.6468658447265625,
      "learning_rate": 6.282698727233665e-06,
      "loss": 1.3197,
      "step": 1227
    },
    {
      "epoch": 0.6236668359573387,
      "grad_norm": 4.137972354888916,
      "learning_rate": 6.267815975294591e-06,
      "loss": 1.1778,
      "step": 1228
    },
    {
      "epoch": 0.6241747079735906,
      "grad_norm": 4.701441287994385,
      "learning_rate": 6.252942821644688e-06,
      "loss": 1.2597,
      "step": 1229
    },
    {
      "epoch": 0.6246825799898426,
      "grad_norm": 3.7897799015045166,
      "learning_rate": 6.2380793045341635e-06,
      "loss": 1.3935,
      "step": 1230
    },
    {
      "epoch": 0.6251904520060945,
      "grad_norm": 4.415806293487549,
      "learning_rate": 6.223225462188453e-06,
      "loss": 1.1646,
      "step": 1231
    },
    {
      "epoch": 0.6256983240223464,
      "grad_norm": 5.204166889190674,
      "learning_rate": 6.2083813328081035e-06,
      "loss": 1.3506,
      "step": 1232
    },
    {
      "epoch": 0.6262061960385983,
      "grad_norm": 4.363425254821777,
      "learning_rate": 6.193546954568677e-06,
      "loss": 1.2515,
      "step": 1233
    },
    {
      "epoch": 0.6267140680548502,
      "grad_norm": 4.286894798278809,
      "learning_rate": 6.1787223656206705e-06,
      "loss": 1.2642,
      "step": 1234
    },
    {
      "epoch": 0.6272219400711021,
      "grad_norm": 3.8692104816436768,
      "learning_rate": 6.163907604089397e-06,
      "loss": 1.3592,
      "step": 1235
    },
    {
      "epoch": 0.627729812087354,
      "grad_norm": 5.8255791664123535,
      "learning_rate": 6.149102708074896e-06,
      "loss": 1.0055,
      "step": 1236
    },
    {
      "epoch": 0.6282376841036059,
      "grad_norm": 4.127172946929932,
      "learning_rate": 6.134307715651842e-06,
      "loss": 1.4512,
      "step": 1237
    },
    {
      "epoch": 0.6287455561198578,
      "grad_norm": 4.500064373016357,
      "learning_rate": 6.119522664869431e-06,
      "loss": 1.079,
      "step": 1238
    },
    {
      "epoch": 0.6292534281361097,
      "grad_norm": 5.0319976806640625,
      "learning_rate": 6.104747593751296e-06,
      "loss": 1.4411,
      "step": 1239
    },
    {
      "epoch": 0.6297613001523616,
      "grad_norm": 5.003214359283447,
      "learning_rate": 6.089982540295398e-06,
      "loss": 1.2222,
      "step": 1240
    },
    {
      "epoch": 0.6302691721686136,
      "grad_norm": 4.036032676696777,
      "learning_rate": 6.075227542473954e-06,
      "loss": 1.3097,
      "step": 1241
    },
    {
      "epoch": 0.6307770441848655,
      "grad_norm": 4.442774772644043,
      "learning_rate": 6.060482638233295e-06,
      "loss": 1.1824,
      "step": 1242
    },
    {
      "epoch": 0.6312849162011173,
      "grad_norm": 5.789218902587891,
      "learning_rate": 6.04574786549381e-06,
      "loss": 0.9217,
      "step": 1243
    },
    {
      "epoch": 0.6317927882173692,
      "grad_norm": 3.8436834812164307,
      "learning_rate": 6.031023262149831e-06,
      "loss": 1.3202,
      "step": 1244
    },
    {
      "epoch": 0.6323006602336211,
      "grad_norm": 4.961956977844238,
      "learning_rate": 6.016308866069532e-06,
      "loss": 1.5101,
      "step": 1245
    },
    {
      "epoch": 0.632808532249873,
      "grad_norm": 3.7859721183776855,
      "learning_rate": 6.001604715094834e-06,
      "loss": 1.3134,
      "step": 1246
    },
    {
      "epoch": 0.6333164042661249,
      "grad_norm": 5.033949375152588,
      "learning_rate": 5.986910847041322e-06,
      "loss": 1.2267,
      "step": 1247
    },
    {
      "epoch": 0.6338242762823768,
      "grad_norm": 3.844045400619507,
      "learning_rate": 5.972227299698121e-06,
      "loss": 1.5616,
      "step": 1248
    },
    {
      "epoch": 0.6343321482986287,
      "grad_norm": 5.14638090133667,
      "learning_rate": 5.957554110827823e-06,
      "loss": 1.4153,
      "step": 1249
    },
    {
      "epoch": 0.6348400203148806,
      "grad_norm": 4.653091907501221,
      "learning_rate": 5.94289131816638e-06,
      "loss": 1.0728,
      "step": 1250
    },
    {
      "epoch": 0.6353478923311325,
      "grad_norm": 4.308130264282227,
      "learning_rate": 5.928238959423005e-06,
      "loss": 1.2786,
      "step": 1251
    },
    {
      "epoch": 0.6358557643473844,
      "grad_norm": 4.122445106506348,
      "learning_rate": 5.9135970722800765e-06,
      "loss": 1.3041,
      "step": 1252
    },
    {
      "epoch": 0.6363636363636364,
      "grad_norm": 4.471611976623535,
      "learning_rate": 5.8989656943930394e-06,
      "loss": 1.3083,
      "step": 1253
    },
    {
      "epoch": 0.6368715083798883,
      "grad_norm": 4.624831199645996,
      "learning_rate": 5.884344863390326e-06,
      "loss": 1.4186,
      "step": 1254
    },
    {
      "epoch": 0.6373793803961402,
      "grad_norm": 4.060239315032959,
      "learning_rate": 5.869734616873226e-06,
      "loss": 1.2426,
      "step": 1255
    },
    {
      "epoch": 0.6378872524123921,
      "grad_norm": 3.788975477218628,
      "learning_rate": 5.855134992415819e-06,
      "loss": 1.3551,
      "step": 1256
    },
    {
      "epoch": 0.638395124428644,
      "grad_norm": 4.169189929962158,
      "learning_rate": 5.840546027564866e-06,
      "loss": 1.418,
      "step": 1257
    },
    {
      "epoch": 0.6389029964448959,
      "grad_norm": 4.810233116149902,
      "learning_rate": 5.825967759839715e-06,
      "loss": 1.3121,
      "step": 1258
    },
    {
      "epoch": 0.6394108684611478,
      "grad_norm": 4.231508255004883,
      "learning_rate": 5.811400226732194e-06,
      "loss": 1.4864,
      "step": 1259
    },
    {
      "epoch": 0.6399187404773997,
      "grad_norm": 5.441370487213135,
      "learning_rate": 5.796843465706539e-06,
      "loss": 1.0832,
      "step": 1260
    },
    {
      "epoch": 0.6404266124936516,
      "grad_norm": 5.238120079040527,
      "learning_rate": 5.782297514199274e-06,
      "loss": 1.1778,
      "step": 1261
    },
    {
      "epoch": 0.6409344845099035,
      "grad_norm": 5.060118675231934,
      "learning_rate": 5.767762409619119e-06,
      "loss": 0.9094,
      "step": 1262
    },
    {
      "epoch": 0.6414423565261554,
      "grad_norm": 5.235774040222168,
      "learning_rate": 5.753238189346912e-06,
      "loss": 1.1636,
      "step": 1263
    },
    {
      "epoch": 0.6419502285424074,
      "grad_norm": 4.136366844177246,
      "learning_rate": 5.738724890735487e-06,
      "loss": 1.2629,
      "step": 1264
    },
    {
      "epoch": 0.6424581005586593,
      "grad_norm": 4.454113483428955,
      "learning_rate": 5.724222551109593e-06,
      "loss": 1.3022,
      "step": 1265
    },
    {
      "epoch": 0.6429659725749112,
      "grad_norm": 5.6957879066467285,
      "learning_rate": 5.709731207765802e-06,
      "loss": 1.205,
      "step": 1266
    },
    {
      "epoch": 0.6434738445911631,
      "grad_norm": 4.1256279945373535,
      "learning_rate": 5.695250897972393e-06,
      "loss": 1.1335,
      "step": 1267
    },
    {
      "epoch": 0.6439817166074149,
      "grad_norm": 4.44126558303833,
      "learning_rate": 5.6807816589692836e-06,
      "loss": 1.2609,
      "step": 1268
    },
    {
      "epoch": 0.6444895886236668,
      "grad_norm": 4.811086654663086,
      "learning_rate": 5.666323527967908e-06,
      "loss": 1.1655,
      "step": 1269
    },
    {
      "epoch": 0.6449974606399187,
      "grad_norm": 4.9102396965026855,
      "learning_rate": 5.6518765421511444e-06,
      "loss": 1.2877,
      "step": 1270
    },
    {
      "epoch": 0.6455053326561706,
      "grad_norm": 4.933690547943115,
      "learning_rate": 5.637440738673199e-06,
      "loss": 1.3126,
      "step": 1271
    },
    {
      "epoch": 0.6460132046724225,
      "grad_norm": 4.120951175689697,
      "learning_rate": 5.6230161546595195e-06,
      "loss": 1.2685,
      "step": 1272
    },
    {
      "epoch": 0.6465210766886744,
      "grad_norm": 3.8976340293884277,
      "learning_rate": 5.608602827206713e-06,
      "loss": 1.2991,
      "step": 1273
    },
    {
      "epoch": 0.6470289487049263,
      "grad_norm": 4.237303256988525,
      "learning_rate": 5.59420079338242e-06,
      "loss": 1.1338,
      "step": 1274
    },
    {
      "epoch": 0.6475368207211782,
      "grad_norm": 4.574217319488525,
      "learning_rate": 5.579810090225242e-06,
      "loss": 1.3666,
      "step": 1275
    },
    {
      "epoch": 0.6480446927374302,
      "grad_norm": 3.9830265045166016,
      "learning_rate": 5.565430754744652e-06,
      "loss": 1.3125,
      "step": 1276
    },
    {
      "epoch": 0.6485525647536821,
      "grad_norm": 4.050839424133301,
      "learning_rate": 5.551062823920874e-06,
      "loss": 1.3415,
      "step": 1277
    },
    {
      "epoch": 0.649060436769934,
      "grad_norm": 5.43248987197876,
      "learning_rate": 5.5367063347048044e-06,
      "loss": 1.2401,
      "step": 1278
    },
    {
      "epoch": 0.6495683087861859,
      "grad_norm": 4.734487056732178,
      "learning_rate": 5.522361324017922e-06,
      "loss": 1.1555,
      "step": 1279
    },
    {
      "epoch": 0.6500761808024378,
      "grad_norm": 4.565279006958008,
      "learning_rate": 5.508027828752181e-06,
      "loss": 1.3278,
      "step": 1280
    },
    {
      "epoch": 0.6505840528186897,
      "grad_norm": 5.360850811004639,
      "learning_rate": 5.493705885769911e-06,
      "loss": 1.375,
      "step": 1281
    },
    {
      "epoch": 0.6510919248349416,
      "grad_norm": 5.566100120544434,
      "learning_rate": 5.47939553190375e-06,
      "loss": 1.2314,
      "step": 1282
    },
    {
      "epoch": 0.6515997968511935,
      "grad_norm": 4.226099967956543,
      "learning_rate": 5.465096803956525e-06,
      "loss": 1.4235,
      "step": 1283
    },
    {
      "epoch": 0.6521076688674454,
      "grad_norm": 4.158005237579346,
      "learning_rate": 5.450809738701155e-06,
      "loss": 1.2179,
      "step": 1284
    },
    {
      "epoch": 0.6526155408836973,
      "grad_norm": 3.9321670532226562,
      "learning_rate": 5.436534372880572e-06,
      "loss": 1.1007,
      "step": 1285
    },
    {
      "epoch": 0.6531234128999492,
      "grad_norm": 5.232217311859131,
      "learning_rate": 5.4222707432076264e-06,
      "loss": 1.3191,
      "step": 1286
    },
    {
      "epoch": 0.6536312849162011,
      "grad_norm": 4.562361240386963,
      "learning_rate": 5.4080188863649745e-06,
      "loss": 1.1464,
      "step": 1287
    },
    {
      "epoch": 0.6541391569324531,
      "grad_norm": 4.1017069816589355,
      "learning_rate": 5.393778839004997e-06,
      "loss": 1.4151,
      "step": 1288
    },
    {
      "epoch": 0.654647028948705,
      "grad_norm": 4.379333019256592,
      "learning_rate": 5.379550637749718e-06,
      "loss": 1.3093,
      "step": 1289
    },
    {
      "epoch": 0.6551549009649569,
      "grad_norm": 5.013577938079834,
      "learning_rate": 5.365334319190676e-06,
      "loss": 1.271,
      "step": 1290
    },
    {
      "epoch": 0.6556627729812088,
      "grad_norm": 4.2003679275512695,
      "learning_rate": 5.351129919888858e-06,
      "loss": 1.4057,
      "step": 1291
    },
    {
      "epoch": 0.6561706449974607,
      "grad_norm": 4.2302727699279785,
      "learning_rate": 5.336937476374608e-06,
      "loss": 1.1751,
      "step": 1292
    },
    {
      "epoch": 0.6566785170137125,
      "grad_norm": 4.532559394836426,
      "learning_rate": 5.322757025147507e-06,
      "loss": 1.1194,
      "step": 1293
    },
    {
      "epoch": 0.6571863890299644,
      "grad_norm": 5.613903522491455,
      "learning_rate": 5.308588602676299e-06,
      "loss": 0.8151,
      "step": 1294
    },
    {
      "epoch": 0.6576942610462163,
      "grad_norm": 3.8187813758850098,
      "learning_rate": 5.294432245398801e-06,
      "loss": 1.3571,
      "step": 1295
    },
    {
      "epoch": 0.6582021330624682,
      "grad_norm": 4.5164031982421875,
      "learning_rate": 5.2802879897217886e-06,
      "loss": 1.1092,
      "step": 1296
    },
    {
      "epoch": 0.6587100050787201,
      "grad_norm": 4.190432071685791,
      "learning_rate": 5.266155872020931e-06,
      "loss": 1.2389,
      "step": 1297
    },
    {
      "epoch": 0.659217877094972,
      "grad_norm": 4.01939582824707,
      "learning_rate": 5.25203592864066e-06,
      "loss": 1.1223,
      "step": 1298
    },
    {
      "epoch": 0.659725749111224,
      "grad_norm": 3.84106183052063,
      "learning_rate": 5.23792819589412e-06,
      "loss": 1.329,
      "step": 1299
    },
    {
      "epoch": 0.6602336211274759,
      "grad_norm": 5.062501430511475,
      "learning_rate": 5.223832710063039e-06,
      "loss": 1.3425,
      "step": 1300
    },
    {
      "epoch": 0.6607414931437278,
      "grad_norm": 4.87921142578125,
      "learning_rate": 5.209749507397648e-06,
      "loss": 1.1282,
      "step": 1301
    },
    {
      "epoch": 0.6612493651599797,
      "grad_norm": 4.17822265625,
      "learning_rate": 5.1956786241166005e-06,
      "loss": 1.2733,
      "step": 1302
    },
    {
      "epoch": 0.6617572371762316,
      "grad_norm": 4.390413284301758,
      "learning_rate": 5.181620096406857e-06,
      "loss": 1.1431,
      "step": 1303
    },
    {
      "epoch": 0.6622651091924835,
      "grad_norm": 5.3159589767456055,
      "learning_rate": 5.167573960423598e-06,
      "loss": 0.8811,
      "step": 1304
    },
    {
      "epoch": 0.6627729812087354,
      "grad_norm": 4.213495254516602,
      "learning_rate": 5.153540252290154e-06,
      "loss": 1.1306,
      "step": 1305
    },
    {
      "epoch": 0.6632808532249873,
      "grad_norm": 4.159801483154297,
      "learning_rate": 5.139519008097877e-06,
      "loss": 1.0229,
      "step": 1306
    },
    {
      "epoch": 0.6637887252412392,
      "grad_norm": 3.946521282196045,
      "learning_rate": 5.1255102639060686e-06,
      "loss": 1.4018,
      "step": 1307
    },
    {
      "epoch": 0.6642965972574911,
      "grad_norm": 6.075663089752197,
      "learning_rate": 5.111514055741891e-06,
      "loss": 1.3182,
      "step": 1308
    },
    {
      "epoch": 0.664804469273743,
      "grad_norm": 3.9616241455078125,
      "learning_rate": 5.097530419600254e-06,
      "loss": 1.544,
      "step": 1309
    },
    {
      "epoch": 0.665312341289995,
      "grad_norm": 4.886152744293213,
      "learning_rate": 5.083559391443751e-06,
      "loss": 1.2317,
      "step": 1310
    },
    {
      "epoch": 0.6658202133062469,
      "grad_norm": 4.9962897300720215,
      "learning_rate": 5.069601007202531e-06,
      "loss": 1.4778,
      "step": 1311
    },
    {
      "epoch": 0.6663280853224988,
      "grad_norm": 4.1560139656066895,
      "learning_rate": 5.055655302774247e-06,
      "loss": 1.4843,
      "step": 1312
    },
    {
      "epoch": 0.6668359573387507,
      "grad_norm": 6.246951580047607,
      "learning_rate": 5.041722314023927e-06,
      "loss": 1.2443,
      "step": 1313
    },
    {
      "epoch": 0.6673438293550026,
      "grad_norm": 4.550904750823975,
      "learning_rate": 5.027802076783898e-06,
      "loss": 1.207,
      "step": 1314
    },
    {
      "epoch": 0.6678517013712545,
      "grad_norm": 5.027195930480957,
      "learning_rate": 5.013894626853708e-06,
      "loss": 1.1246,
      "step": 1315
    },
    {
      "epoch": 0.6683595733875064,
      "grad_norm": 4.895467758178711,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.9455,
      "step": 1316
    },
    {
      "epoch": 0.6688674454037582,
      "grad_norm": 4.147242546081543,
      "learning_rate": 4.986118231956452e-06,
      "loss": 1.2378,
      "step": 1317
    },
    {
      "epoch": 0.6693753174200101,
      "grad_norm": 4.623603343963623,
      "learning_rate": 4.972249358423668e-06,
      "loss": 1.3515,
      "step": 1318
    },
    {
      "epoch": 0.669883189436262,
      "grad_norm": 4.853110313415527,
      "learning_rate": 4.958393415069089e-06,
      "loss": 1.3405,
      "step": 1319
    },
    {
      "epoch": 0.6703910614525139,
      "grad_norm": 4.2498908042907715,
      "learning_rate": 4.944550437526902e-06,
      "loss": 1.3609,
      "step": 1320
    },
    {
      "epoch": 0.6708989334687658,
      "grad_norm": 4.73717737197876,
      "learning_rate": 4.930720461397956e-06,
      "loss": 1.0625,
      "step": 1321
    },
    {
      "epoch": 0.6714068054850177,
      "grad_norm": 4.079125881195068,
      "learning_rate": 4.916903522249656e-06,
      "loss": 1.3944,
      "step": 1322
    },
    {
      "epoch": 0.6719146775012697,
      "grad_norm": 4.965070724487305,
      "learning_rate": 4.9030996556158775e-06,
      "loss": 1.5302,
      "step": 1323
    },
    {
      "epoch": 0.6724225495175216,
      "grad_norm": 4.086692810058594,
      "learning_rate": 4.889308896996885e-06,
      "loss": 1.0986,
      "step": 1324
    },
    {
      "epoch": 0.6729304215337735,
      "grad_norm": 4.803402423858643,
      "learning_rate": 4.875531281859231e-06,
      "loss": 1.2473,
      "step": 1325
    },
    {
      "epoch": 0.6734382935500254,
      "grad_norm": 4.7739057540893555,
      "learning_rate": 4.861766845635659e-06,
      "loss": 1.4185,
      "step": 1326
    },
    {
      "epoch": 0.6739461655662773,
      "grad_norm": 4.143311977386475,
      "learning_rate": 4.84801562372502e-06,
      "loss": 1.3922,
      "step": 1327
    },
    {
      "epoch": 0.6744540375825292,
      "grad_norm": 4.873647212982178,
      "learning_rate": 4.834277651492192e-06,
      "loss": 1.3831,
      "step": 1328
    },
    {
      "epoch": 0.6749619095987811,
      "grad_norm": 4.5239763259887695,
      "learning_rate": 4.8205529642679675e-06,
      "loss": 1.4986,
      "step": 1329
    },
    {
      "epoch": 0.675469781615033,
      "grad_norm": 4.792907238006592,
      "learning_rate": 4.80684159734897e-06,
      "loss": 1.1446,
      "step": 1330
    },
    {
      "epoch": 0.6759776536312849,
      "grad_norm": 5.070479869842529,
      "learning_rate": 4.793143585997583e-06,
      "loss": 1.1759,
      "step": 1331
    },
    {
      "epoch": 0.6764855256475368,
      "grad_norm": 4.775655746459961,
      "learning_rate": 4.779458965441826e-06,
      "loss": 1.2957,
      "step": 1332
    },
    {
      "epoch": 0.6769933976637887,
      "grad_norm": 5.204723834991455,
      "learning_rate": 4.765787770875282e-06,
      "loss": 1.3555,
      "step": 1333
    },
    {
      "epoch": 0.6775012696800407,
      "grad_norm": 4.274050235748291,
      "learning_rate": 4.75213003745702e-06,
      "loss": 1.2904,
      "step": 1334
    },
    {
      "epoch": 0.6780091416962926,
      "grad_norm": 6.0831170082092285,
      "learning_rate": 4.738485800311475e-06,
      "loss": 1.1394,
      "step": 1335
    },
    {
      "epoch": 0.6785170137125445,
      "grad_norm": 4.034609317779541,
      "learning_rate": 4.724855094528374e-06,
      "loss": 1.367,
      "step": 1336
    },
    {
      "epoch": 0.6790248857287964,
      "grad_norm": 4.201053619384766,
      "learning_rate": 4.711237955162659e-06,
      "loss": 1.3293,
      "step": 1337
    },
    {
      "epoch": 0.6795327577450483,
      "grad_norm": 4.243471622467041,
      "learning_rate": 4.69763441723436e-06,
      "loss": 1.2624,
      "step": 1338
    },
    {
      "epoch": 0.6800406297613002,
      "grad_norm": 4.744422912597656,
      "learning_rate": 4.684044515728549e-06,
      "loss": 1.1923,
      "step": 1339
    },
    {
      "epoch": 0.6805485017775521,
      "grad_norm": 5.497531890869141,
      "learning_rate": 4.670468285595213e-06,
      "loss": 1.0948,
      "step": 1340
    },
    {
      "epoch": 0.681056373793804,
      "grad_norm": 4.373364448547363,
      "learning_rate": 4.656905761749191e-06,
      "loss": 1.3205,
      "step": 1341
    },
    {
      "epoch": 0.6815642458100558,
      "grad_norm": 4.560938358306885,
      "learning_rate": 4.6433569790700625e-06,
      "loss": 1.5145,
      "step": 1342
    },
    {
      "epoch": 0.6820721178263077,
      "grad_norm": 6.305968761444092,
      "learning_rate": 4.62982197240207e-06,
      "loss": 0.8212,
      "step": 1343
    },
    {
      "epoch": 0.6825799898425596,
      "grad_norm": 4.598505973815918,
      "learning_rate": 4.616300776554035e-06,
      "loss": 1.3917,
      "step": 1344
    },
    {
      "epoch": 0.6830878618588115,
      "grad_norm": 6.314585208892822,
      "learning_rate": 4.602793426299255e-06,
      "loss": 1.108,
      "step": 1345
    },
    {
      "epoch": 0.6835957338750634,
      "grad_norm": 5.1614813804626465,
      "learning_rate": 4.589299956375415e-06,
      "loss": 1.3614,
      "step": 1346
    },
    {
      "epoch": 0.6841036058913154,
      "grad_norm": 4.432321071624756,
      "learning_rate": 4.575820401484516e-06,
      "loss": 1.2833,
      "step": 1347
    },
    {
      "epoch": 0.6846114779075673,
      "grad_norm": 5.343625068664551,
      "learning_rate": 4.562354796292761e-06,
      "loss": 1.0535,
      "step": 1348
    },
    {
      "epoch": 0.6851193499238192,
      "grad_norm": 5.0347161293029785,
      "learning_rate": 4.548903175430482e-06,
      "loss": 0.8767,
      "step": 1349
    },
    {
      "epoch": 0.6856272219400711,
      "grad_norm": 5.4508585929870605,
      "learning_rate": 4.5354655734920495e-06,
      "loss": 1.0727,
      "step": 1350
    },
    {
      "epoch": 0.686135093956323,
      "grad_norm": 4.261341094970703,
      "learning_rate": 4.522042025035773e-06,
      "loss": 1.1111,
      "step": 1351
    },
    {
      "epoch": 0.6866429659725749,
      "grad_norm": 4.242602825164795,
      "learning_rate": 4.508632564583832e-06,
      "loss": 1.2206,
      "step": 1352
    },
    {
      "epoch": 0.6871508379888268,
      "grad_norm": 4.068726062774658,
      "learning_rate": 4.495237226622161e-06,
      "loss": 1.2067,
      "step": 1353
    },
    {
      "epoch": 0.6876587100050787,
      "grad_norm": 4.812721252441406,
      "learning_rate": 4.481856045600388e-06,
      "loss": 1.2238,
      "step": 1354
    },
    {
      "epoch": 0.6881665820213306,
      "grad_norm": 4.183948516845703,
      "learning_rate": 4.468489055931723e-06,
      "loss": 1.1906,
      "step": 1355
    },
    {
      "epoch": 0.6886744540375825,
      "grad_norm": 4.936746597290039,
      "learning_rate": 4.455136291992877e-06,
      "loss": 0.9373,
      "step": 1356
    },
    {
      "epoch": 0.6891823260538344,
      "grad_norm": 4.554867267608643,
      "learning_rate": 4.4417977881239925e-06,
      "loss": 1.2939,
      "step": 1357
    },
    {
      "epoch": 0.6896901980700864,
      "grad_norm": 6.861904144287109,
      "learning_rate": 4.428473578628522e-06,
      "loss": 0.9068,
      "step": 1358
    },
    {
      "epoch": 0.6901980700863383,
      "grad_norm": 5.396211624145508,
      "learning_rate": 4.415163697773155e-06,
      "loss": 1.2767,
      "step": 1359
    },
    {
      "epoch": 0.6907059421025902,
      "grad_norm": 4.2031168937683105,
      "learning_rate": 4.401868179787748e-06,
      "loss": 1.3721,
      "step": 1360
    },
    {
      "epoch": 0.6912138141188421,
      "grad_norm": 4.868252277374268,
      "learning_rate": 4.388587058865207e-06,
      "loss": 1.1377,
      "step": 1361
    },
    {
      "epoch": 0.691721686135094,
      "grad_norm": 5.334263801574707,
      "learning_rate": 4.375320369161408e-06,
      "loss": 1.2758,
      "step": 1362
    },
    {
      "epoch": 0.6922295581513459,
      "grad_norm": 4.449823379516602,
      "learning_rate": 4.362068144795127e-06,
      "loss": 1.3147,
      "step": 1363
    },
    {
      "epoch": 0.6927374301675978,
      "grad_norm": 4.9963250160217285,
      "learning_rate": 4.34883041984793e-06,
      "loss": 1.4626,
      "step": 1364
    },
    {
      "epoch": 0.6932453021838497,
      "grad_norm": 4.718398094177246,
      "learning_rate": 4.33560722836409e-06,
      "loss": 1.5417,
      "step": 1365
    },
    {
      "epoch": 0.6937531742001015,
      "grad_norm": 4.16678524017334,
      "learning_rate": 4.322398604350518e-06,
      "loss": 1.3918,
      "step": 1366
    },
    {
      "epoch": 0.6942610462163534,
      "grad_norm": 5.61831521987915,
      "learning_rate": 4.3092045817766426e-06,
      "loss": 1.0232,
      "step": 1367
    },
    {
      "epoch": 0.6947689182326053,
      "grad_norm": 4.456976890563965,
      "learning_rate": 4.296025194574359e-06,
      "loss": 1.3631,
      "step": 1368
    },
    {
      "epoch": 0.6952767902488572,
      "grad_norm": 4.0251359939575195,
      "learning_rate": 4.282860476637906e-06,
      "loss": 1.3043,
      "step": 1369
    },
    {
      "epoch": 0.6957846622651092,
      "grad_norm": 5.202916145324707,
      "learning_rate": 4.269710461823813e-06,
      "loss": 1.3112,
      "step": 1370
    },
    {
      "epoch": 0.6962925342813611,
      "grad_norm": 4.528244495391846,
      "learning_rate": 4.2565751839507855e-06,
      "loss": 1.2985,
      "step": 1371
    },
    {
      "epoch": 0.696800406297613,
      "grad_norm": 6.558594703674316,
      "learning_rate": 4.243454676799628e-06,
      "loss": 0.8046,
      "step": 1372
    },
    {
      "epoch": 0.6973082783138649,
      "grad_norm": 4.963061809539795,
      "learning_rate": 4.230348974113172e-06,
      "loss": 1.2971,
      "step": 1373
    },
    {
      "epoch": 0.6978161503301168,
      "grad_norm": 4.163506031036377,
      "learning_rate": 4.217258109596161e-06,
      "loss": 1.2361,
      "step": 1374
    },
    {
      "epoch": 0.6983240223463687,
      "grad_norm": 5.891390323638916,
      "learning_rate": 4.204182116915179e-06,
      "loss": 1.0048,
      "step": 1375
    },
    {
      "epoch": 0.6988318943626206,
      "grad_norm": 4.587854385375977,
      "learning_rate": 4.191121029698575e-06,
      "loss": 1.3254,
      "step": 1376
    },
    {
      "epoch": 0.6993397663788725,
      "grad_norm": 4.640517711639404,
      "learning_rate": 4.1780748815363545e-06,
      "loss": 1.2664,
      "step": 1377
    },
    {
      "epoch": 0.6998476383951244,
      "grad_norm": 5.369966983795166,
      "learning_rate": 4.165043705980102e-06,
      "loss": 1.1808,
      "step": 1378
    },
    {
      "epoch": 0.7003555104113763,
      "grad_norm": 4.568151950836182,
      "learning_rate": 4.152027536542908e-06,
      "loss": 0.9268,
      "step": 1379
    },
    {
      "epoch": 0.7008633824276282,
      "grad_norm": 4.4422125816345215,
      "learning_rate": 4.139026406699254e-06,
      "loss": 1.3176,
      "step": 1380
    },
    {
      "epoch": 0.7013712544438802,
      "grad_norm": 5.446008682250977,
      "learning_rate": 4.12604034988496e-06,
      "loss": 1.3481,
      "step": 1381
    },
    {
      "epoch": 0.7018791264601321,
      "grad_norm": 4.825053691864014,
      "learning_rate": 4.113069399497067e-06,
      "loss": 1.3604,
      "step": 1382
    },
    {
      "epoch": 0.702386998476384,
      "grad_norm": 4.3071136474609375,
      "learning_rate": 4.1001135888937795e-06,
      "loss": 1.3605,
      "step": 1383
    },
    {
      "epoch": 0.7028948704926359,
      "grad_norm": 4.9159040451049805,
      "learning_rate": 4.087172951394356e-06,
      "loss": 1.2433,
      "step": 1384
    },
    {
      "epoch": 0.7034027425088878,
      "grad_norm": 4.7212347984313965,
      "learning_rate": 4.074247520279033e-06,
      "loss": 1.2597,
      "step": 1385
    },
    {
      "epoch": 0.7039106145251397,
      "grad_norm": 4.443634986877441,
      "learning_rate": 4.061337328788951e-06,
      "loss": 1.2344,
      "step": 1386
    },
    {
      "epoch": 0.7044184865413916,
      "grad_norm": 4.415527820587158,
      "learning_rate": 4.0484424101260465e-06,
      "loss": 1.3632,
      "step": 1387
    },
    {
      "epoch": 0.7049263585576435,
      "grad_norm": 5.799596309661865,
      "learning_rate": 4.0355627974529775e-06,
      "loss": 1.4293,
      "step": 1388
    },
    {
      "epoch": 0.7054342305738954,
      "grad_norm": 4.581190586090088,
      "learning_rate": 4.022698523893052e-06,
      "loss": 1.2923,
      "step": 1389
    },
    {
      "epoch": 0.7059421025901473,
      "grad_norm": 5.361939907073975,
      "learning_rate": 4.009849622530113e-06,
      "loss": 1.2167,
      "step": 1390
    },
    {
      "epoch": 0.7064499746063991,
      "grad_norm": 4.397521495819092,
      "learning_rate": 3.997016126408477e-06,
      "loss": 1.2955,
      "step": 1391
    },
    {
      "epoch": 0.706957846622651,
      "grad_norm": 4.393995761871338,
      "learning_rate": 3.984198068532848e-06,
      "loss": 1.3823,
      "step": 1392
    },
    {
      "epoch": 0.707465718638903,
      "grad_norm": 2.7322566509246826,
      "learning_rate": 3.971395481868218e-06,
      "loss": 0.3306,
      "step": 1393
    },
    {
      "epoch": 0.7079735906551549,
      "grad_norm": 5.019045352935791,
      "learning_rate": 3.9586083993397874e-06,
      "loss": 1.344,
      "step": 1394
    },
    {
      "epoch": 0.7084814626714068,
      "grad_norm": 4.308985710144043,
      "learning_rate": 3.945836853832895e-06,
      "loss": 1.353,
      "step": 1395
    },
    {
      "epoch": 0.7089893346876587,
      "grad_norm": 4.66213846206665,
      "learning_rate": 3.933080878192919e-06,
      "loss": 1.2769,
      "step": 1396
    },
    {
      "epoch": 0.7094972067039106,
      "grad_norm": 5.00924825668335,
      "learning_rate": 3.920340505225192e-06,
      "loss": 1.1399,
      "step": 1397
    },
    {
      "epoch": 0.7100050787201625,
      "grad_norm": 5.90763521194458,
      "learning_rate": 3.907615767694916e-06,
      "loss": 1.2984,
      "step": 1398
    },
    {
      "epoch": 0.7105129507364144,
      "grad_norm": 7.6582465171813965,
      "learning_rate": 3.894906698327095e-06,
      "loss": 0.9014,
      "step": 1399
    },
    {
      "epoch": 0.7110208227526663,
      "grad_norm": 4.450231552124023,
      "learning_rate": 3.8822133298064264e-06,
      "loss": 1.484,
      "step": 1400
    },
    {
      "epoch": 0.7115286947689182,
      "grad_norm": 5.579190254211426,
      "learning_rate": 3.869535694777232e-06,
      "loss": 1.3797,
      "step": 1401
    },
    {
      "epoch": 0.7120365667851701,
      "grad_norm": 4.927333831787109,
      "learning_rate": 3.856873825843379e-06,
      "loss": 1.3417,
      "step": 1402
    },
    {
      "epoch": 0.712544438801422,
      "grad_norm": 4.601704120635986,
      "learning_rate": 3.844227755568177e-06,
      "loss": 1.2373,
      "step": 1403
    },
    {
      "epoch": 0.713052310817674,
      "grad_norm": 4.791215419769287,
      "learning_rate": 3.831597516474306e-06,
      "loss": 1.3297,
      "step": 1404
    },
    {
      "epoch": 0.7135601828339259,
      "grad_norm": 4.16111421585083,
      "learning_rate": 3.818983141043742e-06,
      "loss": 1.4404,
      "step": 1405
    },
    {
      "epoch": 0.7140680548501778,
      "grad_norm": 4.602897644042969,
      "learning_rate": 3.806384661717655e-06,
      "loss": 1.3687,
      "step": 1406
    },
    {
      "epoch": 0.7145759268664297,
      "grad_norm": 5.141113758087158,
      "learning_rate": 3.7938021108963297e-06,
      "loss": 1.0342,
      "step": 1407
    },
    {
      "epoch": 0.7150837988826816,
      "grad_norm": 4.493198871612549,
      "learning_rate": 3.7812355209391003e-06,
      "loss": 1.2501,
      "step": 1408
    },
    {
      "epoch": 0.7155916708989335,
      "grad_norm": 4.949621677398682,
      "learning_rate": 3.768684924164241e-06,
      "loss": 1.3255,
      "step": 1409
    },
    {
      "epoch": 0.7160995429151854,
      "grad_norm": 3.930460214614868,
      "learning_rate": 3.7561503528489007e-06,
      "loss": 1.3839,
      "step": 1410
    },
    {
      "epoch": 0.7166074149314373,
      "grad_norm": 5.201033115386963,
      "learning_rate": 3.7436318392290182e-06,
      "loss": 1.144,
      "step": 1411
    },
    {
      "epoch": 0.7171152869476892,
      "grad_norm": 4.482243537902832,
      "learning_rate": 3.7311294154992296e-06,
      "loss": 1.2822,
      "step": 1412
    },
    {
      "epoch": 0.7176231589639411,
      "grad_norm": 4.71453332901001,
      "learning_rate": 3.718643113812792e-06,
      "loss": 1.3802,
      "step": 1413
    },
    {
      "epoch": 0.718131030980193,
      "grad_norm": 5.0039472579956055,
      "learning_rate": 3.7061729662814984e-06,
      "loss": 1.1548,
      "step": 1414
    },
    {
      "epoch": 0.718638902996445,
      "grad_norm": 4.845503330230713,
      "learning_rate": 3.693719004975609e-06,
      "loss": 1.2801,
      "step": 1415
    },
    {
      "epoch": 0.7191467750126967,
      "grad_norm": 5.027263164520264,
      "learning_rate": 3.6812812619237435e-06,
      "loss": 1.288,
      "step": 1416
    },
    {
      "epoch": 0.7196546470289487,
      "grad_norm": 4.287696361541748,
      "learning_rate": 3.668859769112815e-06,
      "loss": 1.2276,
      "step": 1417
    },
    {
      "epoch": 0.7201625190452006,
      "grad_norm": 4.9991936683654785,
      "learning_rate": 3.6564545584879543e-06,
      "loss": 1.2035,
      "step": 1418
    },
    {
      "epoch": 0.7206703910614525,
      "grad_norm": 4.15277624130249,
      "learning_rate": 3.644065661952406e-06,
      "loss": 1.316,
      "step": 1419
    },
    {
      "epoch": 0.7211782630777044,
      "grad_norm": 4.639977931976318,
      "learning_rate": 3.6316931113674615e-06,
      "loss": 1.2438,
      "step": 1420
    },
    {
      "epoch": 0.7216861350939563,
      "grad_norm": 5.692741870880127,
      "learning_rate": 3.6193369385523834e-06,
      "loss": 1.2181,
      "step": 1421
    },
    {
      "epoch": 0.7221940071102082,
      "grad_norm": 5.025320053100586,
      "learning_rate": 3.6069971752843015e-06,
      "loss": 1.2688,
      "step": 1422
    },
    {
      "epoch": 0.7227018791264601,
      "grad_norm": 6.0205888748168945,
      "learning_rate": 3.594673853298156e-06,
      "loss": 1.202,
      "step": 1423
    },
    {
      "epoch": 0.723209751142712,
      "grad_norm": 4.887278079986572,
      "learning_rate": 3.582367004286593e-06,
      "loss": 1.2262,
      "step": 1424
    },
    {
      "epoch": 0.7237176231589639,
      "grad_norm": 5.552386283874512,
      "learning_rate": 3.570076659899906e-06,
      "loss": 1.0381,
      "step": 1425
    },
    {
      "epoch": 0.7242254951752158,
      "grad_norm": 4.675529956817627,
      "learning_rate": 3.557802851745932e-06,
      "loss": 1.3177,
      "step": 1426
    },
    {
      "epoch": 0.7247333671914677,
      "grad_norm": 6.0183186531066895,
      "learning_rate": 3.545545611389981e-06,
      "loss": 1.0576,
      "step": 1427
    },
    {
      "epoch": 0.7252412392077197,
      "grad_norm": 3.846447229385376,
      "learning_rate": 3.5333049703547663e-06,
      "loss": 1.4093,
      "step": 1428
    },
    {
      "epoch": 0.7257491112239716,
      "grad_norm": 4.579927921295166,
      "learning_rate": 3.5210809601203e-06,
      "loss": 1.176,
      "step": 1429
    },
    {
      "epoch": 0.7262569832402235,
      "grad_norm": 5.084172248840332,
      "learning_rate": 3.5088736121238243e-06,
      "loss": 1.2232,
      "step": 1430
    },
    {
      "epoch": 0.7267648552564754,
      "grad_norm": 4.2836594581604,
      "learning_rate": 3.4966829577597392e-06,
      "loss": 1.1602,
      "step": 1431
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 4.318619728088379,
      "learning_rate": 3.4845090283795046e-06,
      "loss": 1.2524,
      "step": 1432
    },
    {
      "epoch": 0.7277805992889792,
      "grad_norm": 5.638737201690674,
      "learning_rate": 3.4723518552915656e-06,
      "loss": 0.764,
      "step": 1433
    },
    {
      "epoch": 0.7282884713052311,
      "grad_norm": 5.3186774253845215,
      "learning_rate": 3.4602114697612864e-06,
      "loss": 1.1524,
      "step": 1434
    },
    {
      "epoch": 0.728796343321483,
      "grad_norm": 4.445633888244629,
      "learning_rate": 3.4480879030108436e-06,
      "loss": 1.3725,
      "step": 1435
    },
    {
      "epoch": 0.7293042153377349,
      "grad_norm": 4.598139762878418,
      "learning_rate": 3.435981186219165e-06,
      "loss": 1.3252,
      "step": 1436
    },
    {
      "epoch": 0.7298120873539868,
      "grad_norm": 4.246560573577881,
      "learning_rate": 3.4238913505218508e-06,
      "loss": 1.0359,
      "step": 1437
    },
    {
      "epoch": 0.7303199593702387,
      "grad_norm": 5.461703300476074,
      "learning_rate": 3.4118184270110742e-06,
      "loss": 1.0321,
      "step": 1438
    },
    {
      "epoch": 0.7308278313864907,
      "grad_norm": 4.48328161239624,
      "learning_rate": 3.3997624467355274e-06,
      "loss": 1.3616,
      "step": 1439
    },
    {
      "epoch": 0.7313357034027425,
      "grad_norm": 5.669865131378174,
      "learning_rate": 3.3877234407003223e-06,
      "loss": 1.1658,
      "step": 1440
    },
    {
      "epoch": 0.7318435754189944,
      "grad_norm": 4.731591701507568,
      "learning_rate": 3.375701439866916e-06,
      "loss": 1.2121,
      "step": 1441
    },
    {
      "epoch": 0.7323514474352463,
      "grad_norm": 5.656239986419678,
      "learning_rate": 3.3636964751530352e-06,
      "loss": 1.0572,
      "step": 1442
    },
    {
      "epoch": 0.7328593194514982,
      "grad_norm": 6.066091537475586,
      "learning_rate": 3.351708577432586e-06,
      "loss": 1.0253,
      "step": 1443
    },
    {
      "epoch": 0.7333671914677501,
      "grad_norm": 4.80064058303833,
      "learning_rate": 3.339737777535599e-06,
      "loss": 1.2829,
      "step": 1444
    },
    {
      "epoch": 0.733875063484002,
      "grad_norm": 4.782726287841797,
      "learning_rate": 3.3277841062481166e-06,
      "loss": 1.1591,
      "step": 1445
    },
    {
      "epoch": 0.7343829355002539,
      "grad_norm": 5.189933776855469,
      "learning_rate": 3.315847594312134e-06,
      "loss": 0.8328,
      "step": 1446
    },
    {
      "epoch": 0.7348908075165058,
      "grad_norm": 5.7964911460876465,
      "learning_rate": 3.303928272425526e-06,
      "loss": 1.1676,
      "step": 1447
    },
    {
      "epoch": 0.7353986795327577,
      "grad_norm": 4.1361212730407715,
      "learning_rate": 3.29202617124195e-06,
      "loss": 1.2228,
      "step": 1448
    },
    {
      "epoch": 0.7359065515490096,
      "grad_norm": 4.595512390136719,
      "learning_rate": 3.280141321370772e-06,
      "loss": 0.9714,
      "step": 1449
    },
    {
      "epoch": 0.7364144235652615,
      "grad_norm": 4.6223931312561035,
      "learning_rate": 3.268273753377008e-06,
      "loss": 1.3393,
      "step": 1450
    },
    {
      "epoch": 0.7369222955815135,
      "grad_norm": 5.296187400817871,
      "learning_rate": 3.2564234977812095e-06,
      "loss": 1.3692,
      "step": 1451
    },
    {
      "epoch": 0.7374301675977654,
      "grad_norm": 4.795011043548584,
      "learning_rate": 3.2445905850594205e-06,
      "loss": 1.3872,
      "step": 1452
    },
    {
      "epoch": 0.7379380396140173,
      "grad_norm": 6.799955368041992,
      "learning_rate": 3.2327750456430706e-06,
      "loss": 1.3781,
      "step": 1453
    },
    {
      "epoch": 0.7384459116302692,
      "grad_norm": 4.597665309906006,
      "learning_rate": 3.2209769099189225e-06,
      "loss": 1.1434,
      "step": 1454
    },
    {
      "epoch": 0.7389537836465211,
      "grad_norm": 4.368586540222168,
      "learning_rate": 3.2091962082289707e-06,
      "loss": 1.1225,
      "step": 1455
    },
    {
      "epoch": 0.739461655662773,
      "grad_norm": 4.291135787963867,
      "learning_rate": 3.197432970870372e-06,
      "loss": 1.3276,
      "step": 1456
    },
    {
      "epoch": 0.7399695276790249,
      "grad_norm": 4.883306503295898,
      "learning_rate": 3.185687228095381e-06,
      "loss": 1.4548,
      "step": 1457
    },
    {
      "epoch": 0.7404773996952768,
      "grad_norm": 5.671407222747803,
      "learning_rate": 3.1739590101112495e-06,
      "loss": 1.1628,
      "step": 1458
    },
    {
      "epoch": 0.7409852717115287,
      "grad_norm": 4.768751621246338,
      "learning_rate": 3.162248347080159e-06,
      "loss": 1.2853,
      "step": 1459
    },
    {
      "epoch": 0.7414931437277806,
      "grad_norm": 5.760345935821533,
      "learning_rate": 3.1505552691191566e-06,
      "loss": 1.2817,
      "step": 1460
    },
    {
      "epoch": 0.7420010157440325,
      "grad_norm": 5.223692893981934,
      "learning_rate": 3.138879806300051e-06,
      "loss": 1.2861,
      "step": 1461
    },
    {
      "epoch": 0.7425088877602845,
      "grad_norm": 4.0341033935546875,
      "learning_rate": 3.127221988649353e-06,
      "loss": 1.3493,
      "step": 1462
    },
    {
      "epoch": 0.7430167597765364,
      "grad_norm": 5.6256632804870605,
      "learning_rate": 3.1155818461482e-06,
      "loss": 1.0489,
      "step": 1463
    },
    {
      "epoch": 0.7435246317927883,
      "grad_norm": 5.683304786682129,
      "learning_rate": 3.1039594087322667e-06,
      "loss": 0.7919,
      "step": 1464
    },
    {
      "epoch": 0.7440325038090401,
      "grad_norm": 4.113526821136475,
      "learning_rate": 3.0923547062916915e-06,
      "loss": 1.3307,
      "step": 1465
    },
    {
      "epoch": 0.744540375825292,
      "grad_norm": 5.324737071990967,
      "learning_rate": 3.0807677686710112e-06,
      "loss": 1.3653,
      "step": 1466
    },
    {
      "epoch": 0.7450482478415439,
      "grad_norm": 5.714325904846191,
      "learning_rate": 3.0691986256690774e-06,
      "loss": 1.0556,
      "step": 1467
    },
    {
      "epoch": 0.7455561198577958,
      "grad_norm": 4.550981044769287,
      "learning_rate": 3.0576473070389634e-06,
      "loss": 1.2504,
      "step": 1468
    },
    {
      "epoch": 0.7460639918740477,
      "grad_norm": 5.716010570526123,
      "learning_rate": 3.04611384248792e-06,
      "loss": 0.9282,
      "step": 1469
    },
    {
      "epoch": 0.7465718638902996,
      "grad_norm": 5.312298774719238,
      "learning_rate": 3.0345982616772707e-06,
      "loss": 1.349,
      "step": 1470
    },
    {
      "epoch": 0.7470797359065515,
      "grad_norm": 4.471862316131592,
      "learning_rate": 3.0231005942223467e-06,
      "loss": 1.2788,
      "step": 1471
    },
    {
      "epoch": 0.7475876079228034,
      "grad_norm": 4.880403518676758,
      "learning_rate": 3.011620869692412e-06,
      "loss": 1.2968,
      "step": 1472
    },
    {
      "epoch": 0.7480954799390553,
      "grad_norm": 5.2570648193359375,
      "learning_rate": 3.0001591176105917e-06,
      "loss": 1.1027,
      "step": 1473
    },
    {
      "epoch": 0.7486033519553073,
      "grad_norm": 5.979409217834473,
      "learning_rate": 2.9887153674537795e-06,
      "loss": 1.1802,
      "step": 1474
    },
    {
      "epoch": 0.7491112239715592,
      "grad_norm": 4.316808700561523,
      "learning_rate": 2.9772896486525783e-06,
      "loss": 1.1965,
      "step": 1475
    },
    {
      "epoch": 0.7496190959878111,
      "grad_norm": 5.031011581420898,
      "learning_rate": 2.9658819905912205e-06,
      "loss": 1.3382,
      "step": 1476
    },
    {
      "epoch": 0.750126968004063,
      "grad_norm": 4.343713760375977,
      "learning_rate": 2.9544924226074866e-06,
      "loss": 1.205,
      "step": 1477
    },
    {
      "epoch": 0.7506348400203149,
      "grad_norm": 4.553668022155762,
      "learning_rate": 2.9431209739926324e-06,
      "loss": 1.6361,
      "step": 1478
    },
    {
      "epoch": 0.7511427120365668,
      "grad_norm": 5.322806358337402,
      "learning_rate": 2.9317676739913224e-06,
      "loss": 1.1267,
      "step": 1479
    },
    {
      "epoch": 0.7511427120365668,
      "eval_loss": 1.273730754852295,
      "eval_runtime": 168.1921,
      "eval_samples_per_second": 5.85,
      "eval_steps_per_second": 1.463,
      "step": 1479
    },
    {
      "epoch": 0.7516505840528187,
      "grad_norm": 4.639729022979736,
      "learning_rate": 2.9204325518015384e-06,
      "loss": 1.238,
      "step": 1480
    },
    {
      "epoch": 0.7521584560690706,
      "grad_norm": 5.131244659423828,
      "learning_rate": 2.909115636574519e-06,
      "loss": 1.1402,
      "step": 1481
    },
    {
      "epoch": 0.7526663280853225,
      "grad_norm": 4.832932949066162,
      "learning_rate": 2.89781695741468e-06,
      "loss": 1.3281,
      "step": 1482
    },
    {
      "epoch": 0.7531742001015744,
      "grad_norm": 5.1387104988098145,
      "learning_rate": 2.886536543379532e-06,
      "loss": 1.0876,
      "step": 1483
    },
    {
      "epoch": 0.7536820721178263,
      "grad_norm": 4.41108512878418,
      "learning_rate": 2.875274423479617e-06,
      "loss": 1.2291,
      "step": 1484
    },
    {
      "epoch": 0.7541899441340782,
      "grad_norm": 5.090763568878174,
      "learning_rate": 2.8640306266784225e-06,
      "loss": 1.1907,
      "step": 1485
    },
    {
      "epoch": 0.7546978161503302,
      "grad_norm": 5.384039402008057,
      "learning_rate": 2.8528051818923264e-06,
      "loss": 1.1933,
      "step": 1486
    },
    {
      "epoch": 0.7552056881665821,
      "grad_norm": 5.645175933837891,
      "learning_rate": 2.8415981179904973e-06,
      "loss": 1.1926,
      "step": 1487
    },
    {
      "epoch": 0.755713560182834,
      "grad_norm": 4.961467742919922,
      "learning_rate": 2.830409463794833e-06,
      "loss": 1.055,
      "step": 1488
    },
    {
      "epoch": 0.7562214321990859,
      "grad_norm": 5.414806365966797,
      "learning_rate": 2.8192392480798957e-06,
      "loss": 1.2816,
      "step": 1489
    },
    {
      "epoch": 0.7567293042153377,
      "grad_norm": 4.7053351402282715,
      "learning_rate": 2.8080874995728204e-06,
      "loss": 1.225,
      "step": 1490
    },
    {
      "epoch": 0.7572371762315896,
      "grad_norm": 4.617583751678467,
      "learning_rate": 2.796954246953246e-06,
      "loss": 1.1683,
      "step": 1491
    },
    {
      "epoch": 0.7577450482478415,
      "grad_norm": 4.665943622589111,
      "learning_rate": 2.785839518853256e-06,
      "loss": 1.2315,
      "step": 1492
    },
    {
      "epoch": 0.7582529202640934,
      "grad_norm": 5.0149760246276855,
      "learning_rate": 2.77474334385728e-06,
      "loss": 1.2449,
      "step": 1493
    },
    {
      "epoch": 0.7587607922803453,
      "grad_norm": 4.400003433227539,
      "learning_rate": 2.763665750502045e-06,
      "loss": 1.0575,
      "step": 1494
    },
    {
      "epoch": 0.7592686642965972,
      "grad_norm": 5.59690523147583,
      "learning_rate": 2.752606767276479e-06,
      "loss": 1.1692,
      "step": 1495
    },
    {
      "epoch": 0.7597765363128491,
      "grad_norm": 4.624378204345703,
      "learning_rate": 2.741566422621661e-06,
      "loss": 1.3387,
      "step": 1496
    },
    {
      "epoch": 0.760284408329101,
      "grad_norm": 4.425624370574951,
      "learning_rate": 2.7305447449307245e-06,
      "loss": 1.6304,
      "step": 1497
    },
    {
      "epoch": 0.760792280345353,
      "grad_norm": 4.48796272277832,
      "learning_rate": 2.7195417625488075e-06,
      "loss": 1.3156,
      "step": 1498
    },
    {
      "epoch": 0.7613001523616049,
      "grad_norm": 4.57307243347168,
      "learning_rate": 2.7085575037729607e-06,
      "loss": 1.1777,
      "step": 1499
    },
    {
      "epoch": 0.7618080243778568,
      "grad_norm": 4.739150524139404,
      "learning_rate": 2.6975919968520813e-06,
      "loss": 1.1394,
      "step": 1500
    },
    {
      "epoch": 0.7623158963941087,
      "grad_norm": 5.048216819763184,
      "learning_rate": 2.686645269986843e-06,
      "loss": 1.1932,
      "step": 1501
    },
    {
      "epoch": 0.7628237684103606,
      "grad_norm": 4.460635185241699,
      "learning_rate": 2.6757173513296275e-06,
      "loss": 1.4603,
      "step": 1502
    },
    {
      "epoch": 0.7633316404266125,
      "grad_norm": 3.7929909229278564,
      "learning_rate": 2.6648082689844367e-06,
      "loss": 1.0992,
      "step": 1503
    },
    {
      "epoch": 0.7638395124428644,
      "grad_norm": 4.791143894195557,
      "learning_rate": 2.653918051006833e-06,
      "loss": 1.1181,
      "step": 1504
    },
    {
      "epoch": 0.7643473844591163,
      "grad_norm": 4.762235641479492,
      "learning_rate": 2.6430467254038716e-06,
      "loss": 1.2734,
      "step": 1505
    },
    {
      "epoch": 0.7648552564753682,
      "grad_norm": 4.421921253204346,
      "learning_rate": 2.6321943201340117e-06,
      "loss": 1.2535,
      "step": 1506
    },
    {
      "epoch": 0.7653631284916201,
      "grad_norm": 5.441786766052246,
      "learning_rate": 2.621360863107053e-06,
      "loss": 1.1429,
      "step": 1507
    },
    {
      "epoch": 0.765871000507872,
      "grad_norm": 4.744370460510254,
      "learning_rate": 2.6105463821840727e-06,
      "loss": 1.1283,
      "step": 1508
    },
    {
      "epoch": 0.766378872524124,
      "grad_norm": 1.064754843711853,
      "learning_rate": 2.599750905177346e-06,
      "loss": 0.2659,
      "step": 1509
    },
    {
      "epoch": 0.7668867445403759,
      "grad_norm": 4.2103800773620605,
      "learning_rate": 2.5889744598502643e-06,
      "loss": 1.3094,
      "step": 1510
    },
    {
      "epoch": 0.7673946165566278,
      "grad_norm": 4.509934902191162,
      "learning_rate": 2.578217073917285e-06,
      "loss": 1.2055,
      "step": 1511
    },
    {
      "epoch": 0.7679024885728797,
      "grad_norm": 4.744308948516846,
      "learning_rate": 2.5674787750438447e-06,
      "loss": 1.1876,
      "step": 1512
    },
    {
      "epoch": 0.7684103605891316,
      "grad_norm": 5.503872871398926,
      "learning_rate": 2.5567595908462905e-06,
      "loss": 1.3198,
      "step": 1513
    },
    {
      "epoch": 0.7689182326053834,
      "grad_norm": 4.523368835449219,
      "learning_rate": 2.5460595488918104e-06,
      "loss": 1.2507,
      "step": 1514
    },
    {
      "epoch": 0.7694261046216353,
      "grad_norm": 5.009969234466553,
      "learning_rate": 2.5353786766983736e-06,
      "loss": 0.9893,
      "step": 1515
    },
    {
      "epoch": 0.7699339766378872,
      "grad_norm": 5.002704620361328,
      "learning_rate": 2.5247170017346366e-06,
      "loss": 1.2085,
      "step": 1516
    },
    {
      "epoch": 0.7704418486541391,
      "grad_norm": 5.431525230407715,
      "learning_rate": 2.5140745514198894e-06,
      "loss": 1.2576,
      "step": 1517
    },
    {
      "epoch": 0.770949720670391,
      "grad_norm": 4.792313098907471,
      "learning_rate": 2.503451353123987e-06,
      "loss": 1.2376,
      "step": 1518
    },
    {
      "epoch": 0.7714575926866429,
      "grad_norm": 4.87385892868042,
      "learning_rate": 2.4928474341672627e-06,
      "loss": 1.3085,
      "step": 1519
    },
    {
      "epoch": 0.7719654647028948,
      "grad_norm": 5.436407089233398,
      "learning_rate": 2.4822628218204726e-06,
      "loss": 1.3114,
      "step": 1520
    },
    {
      "epoch": 0.7724733367191468,
      "grad_norm": 4.732920169830322,
      "learning_rate": 2.471697543304724e-06,
      "loss": 1.3637,
      "step": 1521
    },
    {
      "epoch": 0.7729812087353987,
      "grad_norm": 4.354372501373291,
      "learning_rate": 2.4611516257913947e-06,
      "loss": 1.2134,
      "step": 1522
    },
    {
      "epoch": 0.7734890807516506,
      "grad_norm": 5.263430118560791,
      "learning_rate": 2.4506250964020805e-06,
      "loss": 1.1194,
      "step": 1523
    },
    {
      "epoch": 0.7739969527679025,
      "grad_norm": 4.511394023895264,
      "learning_rate": 2.4401179822085043e-06,
      "loss": 1.2076,
      "step": 1524
    },
    {
      "epoch": 0.7745048247841544,
      "grad_norm": 4.439097881317139,
      "learning_rate": 2.429630310232469e-06,
      "loss": 1.2132,
      "step": 1525
    },
    {
      "epoch": 0.7750126968004063,
      "grad_norm": 4.4272894859313965,
      "learning_rate": 2.4191621074457663e-06,
      "loss": 1.0654,
      "step": 1526
    },
    {
      "epoch": 0.7755205688166582,
      "grad_norm": 4.477302074432373,
      "learning_rate": 2.4087134007701264e-06,
      "loss": 1.1918,
      "step": 1527
    },
    {
      "epoch": 0.7760284408329101,
      "grad_norm": 4.18808126449585,
      "learning_rate": 2.398284217077135e-06,
      "loss": 1.1018,
      "step": 1528
    },
    {
      "epoch": 0.776536312849162,
      "grad_norm": 4.593721866607666,
      "learning_rate": 2.387874583188171e-06,
      "loss": 1.2392,
      "step": 1529
    },
    {
      "epoch": 0.7770441848654139,
      "grad_norm": 6.527272701263428,
      "learning_rate": 2.3774845258743317e-06,
      "loss": 1.083,
      "step": 1530
    },
    {
      "epoch": 0.7775520568816658,
      "grad_norm": 4.023992538452148,
      "learning_rate": 2.367114071856379e-06,
      "loss": 1.3593,
      "step": 1531
    },
    {
      "epoch": 0.7780599288979178,
      "grad_norm": 4.405642986297607,
      "learning_rate": 2.356763247804649e-06,
      "loss": 1.3145,
      "step": 1532
    },
    {
      "epoch": 0.7785678009141697,
      "grad_norm": 4.339470386505127,
      "learning_rate": 2.346432080338995e-06,
      "loss": 1.0931,
      "step": 1533
    },
    {
      "epoch": 0.7790756729304216,
      "grad_norm": 5.390122890472412,
      "learning_rate": 2.336120596028728e-06,
      "loss": 1.193,
      "step": 1534
    },
    {
      "epoch": 0.7795835449466735,
      "grad_norm": 5.478177070617676,
      "learning_rate": 2.3258288213925283e-06,
      "loss": 1.4469,
      "step": 1535
    },
    {
      "epoch": 0.7800914169629254,
      "grad_norm": 4.124152183532715,
      "learning_rate": 2.315556782898388e-06,
      "loss": 1.3121,
      "step": 1536
    },
    {
      "epoch": 0.7805992889791773,
      "grad_norm": 5.337630748748779,
      "learning_rate": 2.305304506963548e-06,
      "loss": 1.3499,
      "step": 1537
    },
    {
      "epoch": 0.7811071609954292,
      "grad_norm": 4.505415439605713,
      "learning_rate": 2.2950720199544275e-06,
      "loss": 1.1125,
      "step": 1538
    },
    {
      "epoch": 0.781615033011681,
      "grad_norm": 5.707019805908203,
      "learning_rate": 2.28485934818654e-06,
      "loss": 1.1868,
      "step": 1539
    },
    {
      "epoch": 0.7821229050279329,
      "grad_norm": 5.610849380493164,
      "learning_rate": 2.2746665179244543e-06,
      "loss": 1.1175,
      "step": 1540
    },
    {
      "epoch": 0.7826307770441848,
      "grad_norm": 6.6965413093566895,
      "learning_rate": 2.2644935553816994e-06,
      "loss": 1.1509,
      "step": 1541
    },
    {
      "epoch": 0.7831386490604367,
      "grad_norm": 5.4784135818481445,
      "learning_rate": 2.2543404867207165e-06,
      "loss": 1.2848,
      "step": 1542
    },
    {
      "epoch": 0.7836465210766886,
      "grad_norm": 4.223880767822266,
      "learning_rate": 2.2442073380527775e-06,
      "loss": 1.1961,
      "step": 1543
    },
    {
      "epoch": 0.7841543930929405,
      "grad_norm": 4.37587308883667,
      "learning_rate": 2.2340941354379363e-06,
      "loss": 1.2925,
      "step": 1544
    },
    {
      "epoch": 0.7846622651091925,
      "grad_norm": 4.4966511726379395,
      "learning_rate": 2.22400090488494e-06,
      "loss": 1.0749,
      "step": 1545
    },
    {
      "epoch": 0.7851701371254444,
      "grad_norm": 5.9518046379089355,
      "learning_rate": 2.2139276723511715e-06,
      "loss": 1.2663,
      "step": 1546
    },
    {
      "epoch": 0.7856780091416963,
      "grad_norm": 5.12912130355835,
      "learning_rate": 2.2038744637425967e-06,
      "loss": 1.3729,
      "step": 1547
    },
    {
      "epoch": 0.7861858811579482,
      "grad_norm": 4.480926513671875,
      "learning_rate": 2.19384130491367e-06,
      "loss": 1.3531,
      "step": 1548
    },
    {
      "epoch": 0.7866937531742001,
      "grad_norm": 4.151101589202881,
      "learning_rate": 2.183828221667287e-06,
      "loss": 1.3121,
      "step": 1549
    },
    {
      "epoch": 0.787201625190452,
      "grad_norm": 5.753885269165039,
      "learning_rate": 2.173835239754719e-06,
      "loss": 1.2623,
      "step": 1550
    },
    {
      "epoch": 0.7877094972067039,
      "grad_norm": 4.274275779724121,
      "learning_rate": 2.163862384875535e-06,
      "loss": 1.4674,
      "step": 1551
    },
    {
      "epoch": 0.7882173692229558,
      "grad_norm": 6.411306858062744,
      "learning_rate": 2.153909682677544e-06,
      "loss": 1.1709,
      "step": 1552
    },
    {
      "epoch": 0.7887252412392077,
      "grad_norm": 5.627938270568848,
      "learning_rate": 2.1439771587567325e-06,
      "loss": 1.1385,
      "step": 1553
    },
    {
      "epoch": 0.7892331132554596,
      "grad_norm": 5.276611804962158,
      "learning_rate": 2.134064838657185e-06,
      "loss": 1.1439,
      "step": 1554
    },
    {
      "epoch": 0.7897409852717115,
      "grad_norm": 5.1386213302612305,
      "learning_rate": 2.124172747871027e-06,
      "loss": 1.2375,
      "step": 1555
    },
    {
      "epoch": 0.7902488572879635,
      "grad_norm": 4.349878311157227,
      "learning_rate": 2.1143009118383672e-06,
      "loss": 1.4328,
      "step": 1556
    },
    {
      "epoch": 0.7907567293042154,
      "grad_norm": 5.146805286407471,
      "learning_rate": 2.1044493559472167e-06,
      "loss": 1.3282,
      "step": 1557
    },
    {
      "epoch": 0.7912646013204673,
      "grad_norm": 4.454208850860596,
      "learning_rate": 2.0946181055334336e-06,
      "loss": 1.1992,
      "step": 1558
    },
    {
      "epoch": 0.7917724733367192,
      "grad_norm": 6.001837730407715,
      "learning_rate": 2.0848071858806496e-06,
      "loss": 0.9643,
      "step": 1559
    },
    {
      "epoch": 0.7922803453529711,
      "grad_norm": 4.391412734985352,
      "learning_rate": 2.0750166222202218e-06,
      "loss": 1.2218,
      "step": 1560
    },
    {
      "epoch": 0.792788217369223,
      "grad_norm": 5.964014053344727,
      "learning_rate": 2.065246439731148e-06,
      "loss": 1.24,
      "step": 1561
    },
    {
      "epoch": 0.7932960893854749,
      "grad_norm": 4.850198268890381,
      "learning_rate": 2.055496663540009e-06,
      "loss": 1.5233,
      "step": 1562
    },
    {
      "epoch": 0.7938039614017267,
      "grad_norm": 4.668525695800781,
      "learning_rate": 2.045767318720915e-06,
      "loss": 1.0346,
      "step": 1563
    },
    {
      "epoch": 0.7943118334179786,
      "grad_norm": 5.233807563781738,
      "learning_rate": 2.03605843029542e-06,
      "loss": 1.2894,
      "step": 1564
    },
    {
      "epoch": 0.7948197054342305,
      "grad_norm": 4.899875640869141,
      "learning_rate": 2.02637002323248e-06,
      "loss": 1.2911,
      "step": 1565
    },
    {
      "epoch": 0.7953275774504824,
      "grad_norm": 5.5429606437683105,
      "learning_rate": 2.0167021224483675e-06,
      "loss": 1.2636,
      "step": 1566
    },
    {
      "epoch": 0.7958354494667343,
      "grad_norm": 4.837125301361084,
      "learning_rate": 2.0070547528066265e-06,
      "loss": 1.4016,
      "step": 1567
    },
    {
      "epoch": 0.7963433214829863,
      "grad_norm": 5.405340671539307,
      "learning_rate": 1.997427939117993e-06,
      "loss": 1.0387,
      "step": 1568
    },
    {
      "epoch": 0.7968511934992382,
      "grad_norm": 4.169713020324707,
      "learning_rate": 1.9878217061403436e-06,
      "loss": 1.4287,
      "step": 1569
    },
    {
      "epoch": 0.7973590655154901,
      "grad_norm": 4.973928451538086,
      "learning_rate": 1.9782360785786236e-06,
      "loss": 1.1843,
      "step": 1570
    },
    {
      "epoch": 0.797866937531742,
      "grad_norm": 5.256599426269531,
      "learning_rate": 1.9686710810847832e-06,
      "loss": 1.3887,
      "step": 1571
    },
    {
      "epoch": 0.7983748095479939,
      "grad_norm": 6.420433044433594,
      "learning_rate": 1.9591267382577193e-06,
      "loss": 1.0757,
      "step": 1572
    },
    {
      "epoch": 0.7988826815642458,
      "grad_norm": 5.725488662719727,
      "learning_rate": 1.949603074643216e-06,
      "loss": 1.0758,
      "step": 1573
    },
    {
      "epoch": 0.7993905535804977,
      "grad_norm": 5.522940158843994,
      "learning_rate": 1.9401001147338647e-06,
      "loss": 1.0674,
      "step": 1574
    },
    {
      "epoch": 0.7998984255967496,
      "grad_norm": 6.481788158416748,
      "learning_rate": 1.930617882969016e-06,
      "loss": 1.117,
      "step": 1575
    },
    {
      "epoch": 0.8004062976130015,
      "grad_norm": 4.595775127410889,
      "learning_rate": 1.9211564037347188e-06,
      "loss": 1.2688,
      "step": 1576
    },
    {
      "epoch": 0.8009141696292534,
      "grad_norm": 5.821315765380859,
      "learning_rate": 1.9117157013636435e-06,
      "loss": 1.2299,
      "step": 1577
    },
    {
      "epoch": 0.8014220416455053,
      "grad_norm": 4.451550483703613,
      "learning_rate": 1.9022958001350255e-06,
      "loss": 1.2717,
      "step": 1578
    },
    {
      "epoch": 0.8019299136617573,
      "grad_norm": 5.8209357261657715,
      "learning_rate": 1.8928967242746154e-06,
      "loss": 1.0486,
      "step": 1579
    },
    {
      "epoch": 0.8024377856780092,
      "grad_norm": 5.02223539352417,
      "learning_rate": 1.8835184979545984e-06,
      "loss": 1.2014,
      "step": 1580
    },
    {
      "epoch": 0.8029456576942611,
      "grad_norm": 4.6913228034973145,
      "learning_rate": 1.8741611452935382e-06,
      "loss": 1.2134,
      "step": 1581
    },
    {
      "epoch": 0.803453529710513,
      "grad_norm": 4.663290023803711,
      "learning_rate": 1.8648246903563238e-06,
      "loss": 1.2812,
      "step": 1582
    },
    {
      "epoch": 0.8039614017267649,
      "grad_norm": 4.683165550231934,
      "learning_rate": 1.8555091571540928e-06,
      "loss": 1.3909,
      "step": 1583
    },
    {
      "epoch": 0.8044692737430168,
      "grad_norm": 5.630870342254639,
      "learning_rate": 1.8462145696441768e-06,
      "loss": 1.5263,
      "step": 1584
    },
    {
      "epoch": 0.8049771457592687,
      "grad_norm": 4.374763488769531,
      "learning_rate": 1.8369409517300485e-06,
      "loss": 1.3124,
      "step": 1585
    },
    {
      "epoch": 0.8054850177755206,
      "grad_norm": 6.090631008148193,
      "learning_rate": 1.8276883272612434e-06,
      "loss": 1.1145,
      "step": 1586
    },
    {
      "epoch": 0.8059928897917725,
      "grad_norm": 4.912288665771484,
      "learning_rate": 1.8184567200333104e-06,
      "loss": 1.1028,
      "step": 1587
    },
    {
      "epoch": 0.8065007618080243,
      "grad_norm": 4.933427333831787,
      "learning_rate": 1.8092461537877436e-06,
      "loss": 1.1397,
      "step": 1588
    },
    {
      "epoch": 0.8070086338242762,
      "grad_norm": 5.186468601226807,
      "learning_rate": 1.8000566522119323e-06,
      "loss": 1.3007,
      "step": 1589
    },
    {
      "epoch": 0.8075165058405281,
      "grad_norm": 4.519622325897217,
      "learning_rate": 1.7908882389390857e-06,
      "loss": 1.3,
      "step": 1590
    },
    {
      "epoch": 0.80802437785678,
      "grad_norm": 4.897008895874023,
      "learning_rate": 1.7817409375481798e-06,
      "loss": 1.3872,
      "step": 1591
    },
    {
      "epoch": 0.808532249873032,
      "grad_norm": 5.396913051605225,
      "learning_rate": 1.7726147715638985e-06,
      "loss": 1.1085,
      "step": 1592
    },
    {
      "epoch": 0.8090401218892839,
      "grad_norm": 5.167842388153076,
      "learning_rate": 1.763509764456568e-06,
      "loss": 1.1642,
      "step": 1593
    },
    {
      "epoch": 0.8095479939055358,
      "grad_norm": 4.8681559562683105,
      "learning_rate": 1.7544259396420993e-06,
      "loss": 1.3848,
      "step": 1594
    },
    {
      "epoch": 0.8100558659217877,
      "grad_norm": 4.275599956512451,
      "learning_rate": 1.745363320481932e-06,
      "loss": 1.2506,
      "step": 1595
    },
    {
      "epoch": 0.8105637379380396,
      "grad_norm": 5.7443742752075195,
      "learning_rate": 1.7363219302829625e-06,
      "loss": 1.1711,
      "step": 1596
    },
    {
      "epoch": 0.8110716099542915,
      "grad_norm": 5.762200832366943,
      "learning_rate": 1.7273017922974944e-06,
      "loss": 1.1643,
      "step": 1597
    },
    {
      "epoch": 0.8115794819705434,
      "grad_norm": 5.273171901702881,
      "learning_rate": 1.718302929723179e-06,
      "loss": 1.3189,
      "step": 1598
    },
    {
      "epoch": 0.8120873539867953,
      "grad_norm": 4.075494289398193,
      "learning_rate": 1.7093253657029485e-06,
      "loss": 1.349,
      "step": 1599
    },
    {
      "epoch": 0.8125952260030472,
      "grad_norm": 4.767585754394531,
      "learning_rate": 1.7003691233249597e-06,
      "loss": 1.4168,
      "step": 1600
    },
    {
      "epoch": 0.8131030980192991,
      "grad_norm": 5.139211654663086,
      "learning_rate": 1.6914342256225347e-06,
      "loss": 1.3283,
      "step": 1601
    },
    {
      "epoch": 0.813610970035551,
      "grad_norm": 4.2492241859436035,
      "learning_rate": 1.6825206955741092e-06,
      "loss": 1.4171,
      "step": 1602
    },
    {
      "epoch": 0.814118842051803,
      "grad_norm": 5.196702003479004,
      "learning_rate": 1.6736285561031595e-06,
      "loss": 0.9668,
      "step": 1603
    },
    {
      "epoch": 0.8146267140680549,
      "grad_norm": 6.883477210998535,
      "learning_rate": 1.6647578300781486e-06,
      "loss": 1.0839,
      "step": 1604
    },
    {
      "epoch": 0.8151345860843068,
      "grad_norm": 4.714962005615234,
      "learning_rate": 1.6559085403124797e-06,
      "loss": 1.1373,
      "step": 1605
    },
    {
      "epoch": 0.8156424581005587,
      "grad_norm": 4.462029933929443,
      "learning_rate": 1.6470807095644127e-06,
      "loss": 1.3212,
      "step": 1606
    },
    {
      "epoch": 0.8161503301168106,
      "grad_norm": 5.117851734161377,
      "learning_rate": 1.6382743605370344e-06,
      "loss": 1.2225,
      "step": 1607
    },
    {
      "epoch": 0.8166582021330625,
      "grad_norm": 4.902977466583252,
      "learning_rate": 1.6294895158781743e-06,
      "loss": 1.3309,
      "step": 1608
    },
    {
      "epoch": 0.8171660741493144,
      "grad_norm": 4.548398017883301,
      "learning_rate": 1.6207261981803657e-06,
      "loss": 1.4154,
      "step": 1609
    },
    {
      "epoch": 0.8176739461655663,
      "grad_norm": 5.7301506996154785,
      "learning_rate": 1.611984429980772e-06,
      "loss": 1.2463,
      "step": 1610
    },
    {
      "epoch": 0.8181818181818182,
      "grad_norm": 4.687128067016602,
      "learning_rate": 1.6032642337611458e-06,
      "loss": 1.1388,
      "step": 1611
    },
    {
      "epoch": 0.8186896901980701,
      "grad_norm": 5.663663864135742,
      "learning_rate": 1.594565631947753e-06,
      "loss": 1.1684,
      "step": 1612
    },
    {
      "epoch": 0.8191975622143219,
      "grad_norm": 4.085853099822998,
      "learning_rate": 1.5858886469113277e-06,
      "loss": 1.3194,
      "step": 1613
    },
    {
      "epoch": 0.8197054342305738,
      "grad_norm": 4.943175792694092,
      "learning_rate": 1.5772333009670137e-06,
      "loss": 1.2014,
      "step": 1614
    },
    {
      "epoch": 0.8202133062468258,
      "grad_norm": 5.281403541564941,
      "learning_rate": 1.568599616374299e-06,
      "loss": 0.8753,
      "step": 1615
    },
    {
      "epoch": 0.8207211782630777,
      "grad_norm": 5.390008926391602,
      "learning_rate": 1.5599876153369653e-06,
      "loss": 1.3108,
      "step": 1616
    },
    {
      "epoch": 0.8212290502793296,
      "grad_norm": 4.513533592224121,
      "learning_rate": 1.5513973200030274e-06,
      "loss": 1.2118,
      "step": 1617
    },
    {
      "epoch": 0.8217369222955815,
      "grad_norm": 7.519176483154297,
      "learning_rate": 1.5428287524646867e-06,
      "loss": 0.88,
      "step": 1618
    },
    {
      "epoch": 0.8222447943118334,
      "grad_norm": 4.701818466186523,
      "learning_rate": 1.534281934758256e-06,
      "loss": 1.3704,
      "step": 1619
    },
    {
      "epoch": 0.8227526663280853,
      "grad_norm": 6.736048221588135,
      "learning_rate": 1.525756888864115e-06,
      "loss": 1.211,
      "step": 1620
    },
    {
      "epoch": 0.8232605383443372,
      "grad_norm": 5.540770053863525,
      "learning_rate": 1.5172536367066582e-06,
      "loss": 1.2807,
      "step": 1621
    },
    {
      "epoch": 0.8237684103605891,
      "grad_norm": 4.7243781089782715,
      "learning_rate": 1.5087722001542216e-06,
      "loss": 1.1751,
      "step": 1622
    },
    {
      "epoch": 0.824276282376841,
      "grad_norm": 4.625030517578125,
      "learning_rate": 1.5003126010190427e-06,
      "loss": 1.0707,
      "step": 1623
    },
    {
      "epoch": 0.8247841543930929,
      "grad_norm": 5.2486796379089355,
      "learning_rate": 1.491874861057202e-06,
      "loss": 1.1341,
      "step": 1624
    },
    {
      "epoch": 0.8252920264093448,
      "grad_norm": 5.326663017272949,
      "learning_rate": 1.4834590019685545e-06,
      "loss": 1.2813,
      "step": 1625
    },
    {
      "epoch": 0.8257998984255968,
      "grad_norm": 4.928109169006348,
      "learning_rate": 1.4750650453966863e-06,
      "loss": 1.3705,
      "step": 1626
    },
    {
      "epoch": 0.8263077704418487,
      "grad_norm": 4.358314514160156,
      "learning_rate": 1.4666930129288603e-06,
      "loss": 1.3322,
      "step": 1627
    },
    {
      "epoch": 0.8268156424581006,
      "grad_norm": 4.756168365478516,
      "learning_rate": 1.458342926095948e-06,
      "loss": 1.247,
      "step": 1628
    },
    {
      "epoch": 0.8273235144743525,
      "grad_norm": 5.003471851348877,
      "learning_rate": 1.4500148063723885e-06,
      "loss": 1.4165,
      "step": 1629
    },
    {
      "epoch": 0.8278313864906044,
      "grad_norm": 5.815068244934082,
      "learning_rate": 1.4417086751761188e-06,
      "loss": 1.2575,
      "step": 1630
    },
    {
      "epoch": 0.8283392585068563,
      "grad_norm": 5.7446746826171875,
      "learning_rate": 1.4334245538685376e-06,
      "loss": 1.3561,
      "step": 1631
    },
    {
      "epoch": 0.8288471305231082,
      "grad_norm": 5.561966896057129,
      "learning_rate": 1.4251624637544315e-06,
      "loss": 1.2635,
      "step": 1632
    },
    {
      "epoch": 0.8293550025393601,
      "grad_norm": 3.895627737045288,
      "learning_rate": 1.416922426081927e-06,
      "loss": 1.3344,
      "step": 1633
    },
    {
      "epoch": 0.829862874555612,
      "grad_norm": 4.815524101257324,
      "learning_rate": 1.4087044620424461e-06,
      "loss": 1.2994,
      "step": 1634
    },
    {
      "epoch": 0.8303707465718639,
      "grad_norm": 4.807487964630127,
      "learning_rate": 1.4005085927706308e-06,
      "loss": 1.385,
      "step": 1635
    },
    {
      "epoch": 0.8308786185881158,
      "grad_norm": 5.106080055236816,
      "learning_rate": 1.3923348393443115e-06,
      "loss": 1.2108,
      "step": 1636
    },
    {
      "epoch": 0.8313864906043676,
      "grad_norm": 4.974977016448975,
      "learning_rate": 1.3841832227844331e-06,
      "loss": 0.9978,
      "step": 1637
    },
    {
      "epoch": 0.8318943626206196,
      "grad_norm": 4.848039627075195,
      "learning_rate": 1.3760537640550165e-06,
      "loss": 1.1942,
      "step": 1638
    },
    {
      "epoch": 0.8324022346368715,
      "grad_norm": 5.152993679046631,
      "learning_rate": 1.3679464840630918e-06,
      "loss": 1.0863,
      "step": 1639
    },
    {
      "epoch": 0.8329101066531234,
      "grad_norm": 4.386135578155518,
      "learning_rate": 1.3598614036586576e-06,
      "loss": 1.1307,
      "step": 1640
    },
    {
      "epoch": 0.8334179786693753,
      "grad_norm": 5.556447982788086,
      "learning_rate": 1.351798543634615e-06,
      "loss": 1.0966,
      "step": 1641
    },
    {
      "epoch": 0.8339258506856272,
      "grad_norm": 4.602492332458496,
      "learning_rate": 1.3437579247267195e-06,
      "loss": 1.2305,
      "step": 1642
    },
    {
      "epoch": 0.8344337227018791,
      "grad_norm": 5.0979390144348145,
      "learning_rate": 1.335739567613532e-06,
      "loss": 1.2751,
      "step": 1643
    },
    {
      "epoch": 0.834941594718131,
      "grad_norm": 4.603364944458008,
      "learning_rate": 1.327743492916359e-06,
      "loss": 1.1891,
      "step": 1644
    },
    {
      "epoch": 0.8354494667343829,
      "grad_norm": 4.802546977996826,
      "learning_rate": 1.3197697211992012e-06,
      "loss": 1.2852,
      "step": 1645
    },
    {
      "epoch": 0.8359573387506348,
      "grad_norm": 4.88373327255249,
      "learning_rate": 1.311818272968699e-06,
      "loss": 1.2016,
      "step": 1646
    },
    {
      "epoch": 0.8364652107668867,
      "grad_norm": 5.029901504516602,
      "learning_rate": 1.3038891686740896e-06,
      "loss": 1.3481,
      "step": 1647
    },
    {
      "epoch": 0.8369730827831386,
      "grad_norm": 4.834127902984619,
      "learning_rate": 1.29598242870714e-06,
      "loss": 1.3121,
      "step": 1648
    },
    {
      "epoch": 0.8374809547993906,
      "grad_norm": 6.290788650512695,
      "learning_rate": 1.288098073402102e-06,
      "loss": 1.2261,
      "step": 1649
    },
    {
      "epoch": 0.8379888268156425,
      "grad_norm": 4.68792200088501,
      "learning_rate": 1.2802361230356642e-06,
      "loss": 1.5772,
      "step": 1650
    },
    {
      "epoch": 0.8384966988318944,
      "grad_norm": 5.027810573577881,
      "learning_rate": 1.2723965978268938e-06,
      "loss": 1.1969,
      "step": 1651
    },
    {
      "epoch": 0.8390045708481463,
      "grad_norm": 5.2663798332214355,
      "learning_rate": 1.264579517937179e-06,
      "loss": 1.0489,
      "step": 1652
    },
    {
      "epoch": 0.8395124428643982,
      "grad_norm": 4.718344211578369,
      "learning_rate": 1.2567849034701928e-06,
      "loss": 1.2672,
      "step": 1653
    },
    {
      "epoch": 0.8400203148806501,
      "grad_norm": 5.486889839172363,
      "learning_rate": 1.2490127744718283e-06,
      "loss": 1.1353,
      "step": 1654
    },
    {
      "epoch": 0.840528186896902,
      "grad_norm": 6.703730583190918,
      "learning_rate": 1.2412631509301487e-06,
      "loss": 1.1024,
      "step": 1655
    },
    {
      "epoch": 0.8410360589131539,
      "grad_norm": 4.949845314025879,
      "learning_rate": 1.2335360527753471e-06,
      "loss": 1.0375,
      "step": 1656
    },
    {
      "epoch": 0.8415439309294058,
      "grad_norm": 4.4931254386901855,
      "learning_rate": 1.2258314998796772e-06,
      "loss": 1.2449,
      "step": 1657
    },
    {
      "epoch": 0.8420518029456577,
      "grad_norm": 5.583803653717041,
      "learning_rate": 1.218149512057416e-06,
      "loss": 1.1586,
      "step": 1658
    },
    {
      "epoch": 0.8425596749619096,
      "grad_norm": 5.380489349365234,
      "learning_rate": 1.2104901090648068e-06,
      "loss": 1.3036,
      "step": 1659
    },
    {
      "epoch": 0.8430675469781616,
      "grad_norm": 6.198934555053711,
      "learning_rate": 1.202853310600015e-06,
      "loss": 0.9664,
      "step": 1660
    },
    {
      "epoch": 0.8435754189944135,
      "grad_norm": 4.49476432800293,
      "learning_rate": 1.195239136303068e-06,
      "loss": 1.2589,
      "step": 1661
    },
    {
      "epoch": 0.8440832910106653,
      "grad_norm": 4.489984512329102,
      "learning_rate": 1.187647605755806e-06,
      "loss": 1.4972,
      "step": 1662
    },
    {
      "epoch": 0.8445911630269172,
      "grad_norm": 5.477954864501953,
      "learning_rate": 1.1800787384818445e-06,
      "loss": 1.0707,
      "step": 1663
    },
    {
      "epoch": 0.8450990350431691,
      "grad_norm": 4.977619647979736,
      "learning_rate": 1.172532553946505e-06,
      "loss": 1.1271,
      "step": 1664
    },
    {
      "epoch": 0.845606907059421,
      "grad_norm": 5.6147332191467285,
      "learning_rate": 1.1650090715567796e-06,
      "loss": 1.1355,
      "step": 1665
    },
    {
      "epoch": 0.8461147790756729,
      "grad_norm": 4.383717060089111,
      "learning_rate": 1.1575083106612771e-06,
      "loss": 1.127,
      "step": 1666
    },
    {
      "epoch": 0.8466226510919248,
      "grad_norm": 5.167008399963379,
      "learning_rate": 1.1500302905501682e-06,
      "loss": 1.0018,
      "step": 1667
    },
    {
      "epoch": 0.8471305231081767,
      "grad_norm": 4.774293899536133,
      "learning_rate": 1.1425750304551387e-06,
      "loss": 1.3005,
      "step": 1668
    },
    {
      "epoch": 0.8476383951244286,
      "grad_norm": 6.064826488494873,
      "learning_rate": 1.1351425495493462e-06,
      "loss": 1.3613,
      "step": 1669
    },
    {
      "epoch": 0.8481462671406805,
      "grad_norm": 4.394528388977051,
      "learning_rate": 1.127732866947362e-06,
      "loss": 1.2643,
      "step": 1670
    },
    {
      "epoch": 0.8486541391569324,
      "grad_norm": 4.836899757385254,
      "learning_rate": 1.1203460017051249e-06,
      "loss": 1.4117,
      "step": 1671
    },
    {
      "epoch": 0.8491620111731844,
      "grad_norm": 4.757767200469971,
      "learning_rate": 1.1129819728198987e-06,
      "loss": 1.4844,
      "step": 1672
    },
    {
      "epoch": 0.8496698831894363,
      "grad_norm": 4.990601062774658,
      "learning_rate": 1.10564079923021e-06,
      "loss": 1.1944,
      "step": 1673
    },
    {
      "epoch": 0.8501777552056882,
      "grad_norm": 5.517089366912842,
      "learning_rate": 1.0983224998158104e-06,
      "loss": 0.8689,
      "step": 1674
    },
    {
      "epoch": 0.8506856272219401,
      "grad_norm": 5.221260070800781,
      "learning_rate": 1.091027093397623e-06,
      "loss": 1.2486,
      "step": 1675
    },
    {
      "epoch": 0.851193499238192,
      "grad_norm": 4.8471245765686035,
      "learning_rate": 1.083754598737702e-06,
      "loss": 1.1717,
      "step": 1676
    },
    {
      "epoch": 0.8517013712544439,
      "grad_norm": 6.431856155395508,
      "learning_rate": 1.076505034539167e-06,
      "loss": 0.9817,
      "step": 1677
    },
    {
      "epoch": 0.8522092432706958,
      "grad_norm": 5.298643589019775,
      "learning_rate": 1.069278419446178e-06,
      "loss": 1.3423,
      "step": 1678
    },
    {
      "epoch": 0.8527171152869477,
      "grad_norm": 4.67987060546875,
      "learning_rate": 1.0620747720438629e-06,
      "loss": 1.2656,
      "step": 1679
    },
    {
      "epoch": 0.8532249873031996,
      "grad_norm": 5.228744029998779,
      "learning_rate": 1.0548941108582956e-06,
      "loss": 1.364,
      "step": 1680
    },
    {
      "epoch": 0.8537328593194515,
      "grad_norm": 4.668889045715332,
      "learning_rate": 1.0477364543564205e-06,
      "loss": 1.3017,
      "step": 1681
    },
    {
      "epoch": 0.8542407313357034,
      "grad_norm": 5.059421539306641,
      "learning_rate": 1.0406018209460344e-06,
      "loss": 0.942,
      "step": 1682
    },
    {
      "epoch": 0.8547486033519553,
      "grad_norm": 4.647306442260742,
      "learning_rate": 1.0334902289757121e-06,
      "loss": 1.1538,
      "step": 1683
    },
    {
      "epoch": 0.8552564753682073,
      "grad_norm": 5.314273834228516,
      "learning_rate": 1.0264016967347744e-06,
      "loss": 1.2407,
      "step": 1684
    },
    {
      "epoch": 0.8557643473844592,
      "grad_norm": 4.523673057556152,
      "learning_rate": 1.0193362424532428e-06,
      "loss": 1.1304,
      "step": 1685
    },
    {
      "epoch": 0.856272219400711,
      "grad_norm": 4.930280685424805,
      "learning_rate": 1.0122938843017837e-06,
      "loss": 1.3051,
      "step": 1686
    },
    {
      "epoch": 0.8567800914169629,
      "grad_norm": 4.712639331817627,
      "learning_rate": 1.0052746403916668e-06,
      "loss": 1.4619,
      "step": 1687
    },
    {
      "epoch": 0.8572879634332148,
      "grad_norm": 4.836647033691406,
      "learning_rate": 9.982785287747131e-07,
      "loss": 1.4634,
      "step": 1688
    },
    {
      "epoch": 0.8577958354494667,
      "grad_norm": 5.0175700187683105,
      "learning_rate": 9.913055674432614e-07,
      "loss": 1.1327,
      "step": 1689
    },
    {
      "epoch": 0.8583037074657186,
      "grad_norm": 4.9350175857543945,
      "learning_rate": 9.84355774330108e-07,
      "loss": 1.053,
      "step": 1690
    },
    {
      "epoch": 0.8588115794819705,
      "grad_norm": 7.005141258239746,
      "learning_rate": 9.774291673084646e-07,
      "loss": 1.2266,
      "step": 1691
    },
    {
      "epoch": 0.8593194514982224,
      "grad_norm": 4.391099452972412,
      "learning_rate": 9.705257641919185e-07,
      "loss": 1.1779,
      "step": 1692
    },
    {
      "epoch": 0.8598273235144743,
      "grad_norm": 4.599768161773682,
      "learning_rate": 9.636455827343784e-07,
      "loss": 1.2445,
      "step": 1693
    },
    {
      "epoch": 0.8603351955307262,
      "grad_norm": 4.420525550842285,
      "learning_rate": 9.56788640630033e-07,
      "loss": 1.3395,
      "step": 1694
    },
    {
      "epoch": 0.8608430675469781,
      "grad_norm": 4.860838413238525,
      "learning_rate": 9.499549555133103e-07,
      "loss": 1.3235,
      "step": 1695
    },
    {
      "epoch": 0.8613509395632301,
      "grad_norm": 5.740962505340576,
      "learning_rate": 9.431445449588184e-07,
      "loss": 1.1896,
      "step": 1696
    },
    {
      "epoch": 0.861858811579482,
      "grad_norm": 5.737368583679199,
      "learning_rate": 9.363574264813113e-07,
      "loss": 1.1083,
      "step": 1697
    },
    {
      "epoch": 0.8623666835957339,
      "grad_norm": 5.145242691040039,
      "learning_rate": 9.295936175356457e-07,
      "loss": 1.0518,
      "step": 1698
    },
    {
      "epoch": 0.8628745556119858,
      "grad_norm": 6.466031551361084,
      "learning_rate": 9.22853135516728e-07,
      "loss": 0.9954,
      "step": 1699
    },
    {
      "epoch": 0.8633824276282377,
      "grad_norm": 4.695497035980225,
      "learning_rate": 9.161359977594719e-07,
      "loss": 1.3734,
      "step": 1700
    },
    {
      "epoch": 0.8638902996444896,
      "grad_norm": 4.825030326843262,
      "learning_rate": 9.094422215387599e-07,
      "loss": 1.1216,
      "step": 1701
    },
    {
      "epoch": 0.8643981716607415,
      "grad_norm": 4.423190116882324,
      "learning_rate": 9.027718240693895e-07,
      "loss": 1.3598,
      "step": 1702
    },
    {
      "epoch": 0.8649060436769934,
      "grad_norm": 5.499927997589111,
      "learning_rate": 8.961248225060382e-07,
      "loss": 1.0977,
      "step": 1703
    },
    {
      "epoch": 0.8654139156932453,
      "grad_norm": 4.7551703453063965,
      "learning_rate": 8.895012339432074e-07,
      "loss": 1.2789,
      "step": 1704
    },
    {
      "epoch": 0.8659217877094972,
      "grad_norm": 4.84812068939209,
      "learning_rate": 8.829010754151957e-07,
      "loss": 1.2925,
      "step": 1705
    },
    {
      "epoch": 0.8664296597257491,
      "grad_norm": 4.788599967956543,
      "learning_rate": 8.763243638960362e-07,
      "loss": 1.3916,
      "step": 1706
    },
    {
      "epoch": 0.866937531742001,
      "grad_norm": 4.135072231292725,
      "learning_rate": 8.697711162994705e-07,
      "loss": 1.3201,
      "step": 1707
    },
    {
      "epoch": 0.867445403758253,
      "grad_norm": 4.580753326416016,
      "learning_rate": 8.632413494788871e-07,
      "loss": 1.4539,
      "step": 1708
    },
    {
      "epoch": 0.8679532757745049,
      "grad_norm": 4.9030022621154785,
      "learning_rate": 8.567350802272967e-07,
      "loss": 1.3565,
      "step": 1709
    },
    {
      "epoch": 0.8684611477907568,
      "grad_norm": 4.669422149658203,
      "learning_rate": 8.502523252772721e-07,
      "loss": 1.1809,
      "step": 1710
    },
    {
      "epoch": 0.8689690198070086,
      "grad_norm": 4.870145797729492,
      "learning_rate": 8.437931013009193e-07,
      "loss": 1.3639,
      "step": 1711
    },
    {
      "epoch": 0.8694768918232605,
      "grad_norm": 4.493689060211182,
      "learning_rate": 8.373574249098238e-07,
      "loss": 1.3479,
      "step": 1712
    },
    {
      "epoch": 0.8699847638395124,
      "grad_norm": 4.5029497146606445,
      "learning_rate": 8.309453126550115e-07,
      "loss": 1.2695,
      "step": 1713
    },
    {
      "epoch": 0.8704926358557643,
      "grad_norm": 4.815980434417725,
      "learning_rate": 8.245567810269128e-07,
      "loss": 1.2276,
      "step": 1714
    },
    {
      "epoch": 0.8710005078720162,
      "grad_norm": 4.96169376373291,
      "learning_rate": 8.181918464553107e-07,
      "loss": 1.2781,
      "step": 1715
    },
    {
      "epoch": 0.8715083798882681,
      "grad_norm": 5.211386680603027,
      "learning_rate": 8.118505253093001e-07,
      "loss": 1.0092,
      "step": 1716
    },
    {
      "epoch": 0.87201625190452,
      "grad_norm": 4.784932613372803,
      "learning_rate": 8.055328338972501e-07,
      "loss": 1.3213,
      "step": 1717
    },
    {
      "epoch": 0.8725241239207719,
      "grad_norm": 5.034736156463623,
      "learning_rate": 7.992387884667607e-07,
      "loss": 1.38,
      "step": 1718
    },
    {
      "epoch": 0.8730319959370239,
      "grad_norm": 5.30827522277832,
      "learning_rate": 7.929684052046194e-07,
      "loss": 1.2445,
      "step": 1719
    },
    {
      "epoch": 0.8735398679532758,
      "grad_norm": 5.4127888679504395,
      "learning_rate": 7.867217002367577e-07,
      "loss": 1.1274,
      "step": 1720
    },
    {
      "epoch": 0.8740477399695277,
      "grad_norm": 5.677814483642578,
      "learning_rate": 7.804986896282152e-07,
      "loss": 1.1436,
      "step": 1721
    },
    {
      "epoch": 0.8745556119857796,
      "grad_norm": 4.909538745880127,
      "learning_rate": 7.742993893830963e-07,
      "loss": 1.3524,
      "step": 1722
    },
    {
      "epoch": 0.8750634840020315,
      "grad_norm": 5.164133548736572,
      "learning_rate": 7.681238154445226e-07,
      "loss": 1.2597,
      "step": 1723
    },
    {
      "epoch": 0.8755713560182834,
      "grad_norm": 5.629814624786377,
      "learning_rate": 7.619719836946049e-07,
      "loss": 1.3796,
      "step": 1724
    },
    {
      "epoch": 0.8760792280345353,
      "grad_norm": 6.742391586303711,
      "learning_rate": 7.55843909954388e-07,
      "loss": 0.7936,
      "step": 1725
    },
    {
      "epoch": 0.8765871000507872,
      "grad_norm": 6.896872043609619,
      "learning_rate": 7.497396099838162e-07,
      "loss": 1.1554,
      "step": 1726
    },
    {
      "epoch": 0.8770949720670391,
      "grad_norm": 6.02463436126709,
      "learning_rate": 7.436590994817028e-07,
      "loss": 1.1979,
      "step": 1727
    },
    {
      "epoch": 0.877602844083291,
      "grad_norm": 4.152164936065674,
      "learning_rate": 7.376023940856702e-07,
      "loss": 1.3851,
      "step": 1728
    },
    {
      "epoch": 0.8781107160995429,
      "grad_norm": 4.744118690490723,
      "learning_rate": 7.315695093721221e-07,
      "loss": 1.2747,
      "step": 1729
    },
    {
      "epoch": 0.8786185881157949,
      "grad_norm": 4.65793514251709,
      "learning_rate": 7.255604608562051e-07,
      "loss": 1.3909,
      "step": 1730
    },
    {
      "epoch": 0.8791264601320468,
      "grad_norm": 5.222963333129883,
      "learning_rate": 7.195752639917619e-07,
      "loss": 1.3076,
      "step": 1731
    },
    {
      "epoch": 0.8796343321482987,
      "grad_norm": 5.764087200164795,
      "learning_rate": 7.136139341712933e-07,
      "loss": 1.278,
      "step": 1732
    },
    {
      "epoch": 0.8801422041645506,
      "grad_norm": 4.756399154663086,
      "learning_rate": 7.076764867259189e-07,
      "loss": 1.3099,
      "step": 1733
    },
    {
      "epoch": 0.8806500761808025,
      "grad_norm": 4.776485919952393,
      "learning_rate": 7.017629369253454e-07,
      "loss": 1.5061,
      "step": 1734
    },
    {
      "epoch": 0.8811579481970544,
      "grad_norm": 4.794582366943359,
      "learning_rate": 6.958732999778118e-07,
      "loss": 1.0798,
      "step": 1735
    },
    {
      "epoch": 0.8816658202133062,
      "grad_norm": 4.570837020874023,
      "learning_rate": 6.90007591030063e-07,
      "loss": 1.2282,
      "step": 1736
    },
    {
      "epoch": 0.8821736922295581,
      "grad_norm": 5.90083122253418,
      "learning_rate": 6.841658251673111e-07,
      "loss": 1.181,
      "step": 1737
    },
    {
      "epoch": 0.88268156424581,
      "grad_norm": 5.166455268859863,
      "learning_rate": 6.783480174131829e-07,
      "loss": 1.4458,
      "step": 1738
    },
    {
      "epoch": 0.8831894362620619,
      "grad_norm": 5.3771257400512695,
      "learning_rate": 6.725541827296966e-07,
      "loss": 1.2551,
      "step": 1739
    },
    {
      "epoch": 0.8836973082783138,
      "grad_norm": 4.8618483543396,
      "learning_rate": 6.667843360172177e-07,
      "loss": 1.1116,
      "step": 1740
    },
    {
      "epoch": 0.8842051802945657,
      "grad_norm": 4.8128838539123535,
      "learning_rate": 6.610384921144175e-07,
      "loss": 1.2609,
      "step": 1741
    },
    {
      "epoch": 0.8847130523108176,
      "grad_norm": 5.699490547180176,
      "learning_rate": 6.553166657982391e-07,
      "loss": 1.1375,
      "step": 1742
    },
    {
      "epoch": 0.8852209243270696,
      "grad_norm": 5.422540664672852,
      "learning_rate": 6.496188717838592e-07,
      "loss": 1.2577,
      "step": 1743
    },
    {
      "epoch": 0.8857287963433215,
      "grad_norm": 5.66526460647583,
      "learning_rate": 6.439451247246486e-07,
      "loss": 1.26,
      "step": 1744
    },
    {
      "epoch": 0.8862366683595734,
      "grad_norm": 5.978145122528076,
      "learning_rate": 6.382954392121321e-07,
      "loss": 1.0481,
      "step": 1745
    },
    {
      "epoch": 0.8867445403758253,
      "grad_norm": 5.229310989379883,
      "learning_rate": 6.326698297759571e-07,
      "loss": 1.2828,
      "step": 1746
    },
    {
      "epoch": 0.8872524123920772,
      "grad_norm": 4.935593605041504,
      "learning_rate": 6.270683108838549e-07,
      "loss": 1.3433,
      "step": 1747
    },
    {
      "epoch": 0.8877602844083291,
      "grad_norm": 6.317234039306641,
      "learning_rate": 6.214908969415956e-07,
      "loss": 1.4458,
      "step": 1748
    },
    {
      "epoch": 0.888268156424581,
      "grad_norm": 6.326529026031494,
      "learning_rate": 6.159376022929631e-07,
      "loss": 1.0901,
      "step": 1749
    },
    {
      "epoch": 0.8887760284408329,
      "grad_norm": 4.908727169036865,
      "learning_rate": 6.104084412197076e-07,
      "loss": 1.2956,
      "step": 1750
    },
    {
      "epoch": 0.8892839004570848,
      "grad_norm": 8.57208251953125,
      "learning_rate": 6.049034279415189e-07,
      "loss": 0.6276,
      "step": 1751
    },
    {
      "epoch": 0.8897917724733367,
      "grad_norm": 4.645028114318848,
      "learning_rate": 5.994225766159778e-07,
      "loss": 1.3687,
      "step": 1752
    },
    {
      "epoch": 0.8902996444895886,
      "grad_norm": 4.887918949127197,
      "learning_rate": 5.939659013385324e-07,
      "loss": 1.3089,
      "step": 1753
    },
    {
      "epoch": 0.8908075165058406,
      "grad_norm": 4.582350730895996,
      "learning_rate": 5.885334161424516e-07,
      "loss": 1.3622,
      "step": 1754
    },
    {
      "epoch": 0.8913153885220925,
      "grad_norm": 4.830961227416992,
      "learning_rate": 5.831251349987921e-07,
      "loss": 1.261,
      "step": 1755
    },
    {
      "epoch": 0.8918232605383444,
      "grad_norm": 4.80549955368042,
      "learning_rate": 5.777410718163712e-07,
      "loss": 1.2435,
      "step": 1756
    },
    {
      "epoch": 0.8923311325545963,
      "grad_norm": 4.790905952453613,
      "learning_rate": 5.72381240441715e-07,
      "loss": 1.3054,
      "step": 1757
    },
    {
      "epoch": 0.8928390045708482,
      "grad_norm": 4.888479709625244,
      "learning_rate": 5.670456546590331e-07,
      "loss": 1.035,
      "step": 1758
    },
    {
      "epoch": 0.8933468765871001,
      "grad_norm": 5.3526787757873535,
      "learning_rate": 5.617343281901854e-07,
      "loss": 1.3651,
      "step": 1759
    },
    {
      "epoch": 0.8938547486033519,
      "grad_norm": 4.392794609069824,
      "learning_rate": 5.564472746946392e-07,
      "loss": 1.2984,
      "step": 1760
    },
    {
      "epoch": 0.8943626206196038,
      "grad_norm": 4.773922920227051,
      "learning_rate": 5.511845077694366e-07,
      "loss": 1.2163,
      "step": 1761
    },
    {
      "epoch": 0.8948704926358557,
      "grad_norm": 5.451786994934082,
      "learning_rate": 5.459460409491613e-07,
      "loss": 1.2309,
      "step": 1762
    },
    {
      "epoch": 0.8953783646521076,
      "grad_norm": 6.1389689445495605,
      "learning_rate": 5.407318877059031e-07,
      "loss": 1.3086,
      "step": 1763
    },
    {
      "epoch": 0.8958862366683595,
      "grad_norm": 5.596880912780762,
      "learning_rate": 5.355420614492268e-07,
      "loss": 1.2208,
      "step": 1764
    },
    {
      "epoch": 0.8963941086846114,
      "grad_norm": 4.918369293212891,
      "learning_rate": 5.30376575526127e-07,
      "loss": 1.2087,
      "step": 1765
    },
    {
      "epoch": 0.8969019807008634,
      "grad_norm": 5.364236831665039,
      "learning_rate": 5.252354432210072e-07,
      "loss": 1.2595,
      "step": 1766
    },
    {
      "epoch": 0.8974098527171153,
      "grad_norm": 5.538005828857422,
      "learning_rate": 5.201186777556333e-07,
      "loss": 1.429,
      "step": 1767
    },
    {
      "epoch": 0.8979177247333672,
      "grad_norm": 4.762514114379883,
      "learning_rate": 5.150262922891092e-07,
      "loss": 1.1968,
      "step": 1768
    },
    {
      "epoch": 0.8984255967496191,
      "grad_norm": 5.0200958251953125,
      "learning_rate": 5.09958299917841e-07,
      "loss": 1.0847,
      "step": 1769
    },
    {
      "epoch": 0.898933468765871,
      "grad_norm": 4.925847053527832,
      "learning_rate": 5.049147136754962e-07,
      "loss": 1.3489,
      "step": 1770
    },
    {
      "epoch": 0.8994413407821229,
      "grad_norm": 6.314319133758545,
      "learning_rate": 4.998955465329791e-07,
      "loss": 1.1123,
      "step": 1771
    },
    {
      "epoch": 0.8999492127983748,
      "grad_norm": 5.687747955322266,
      "learning_rate": 4.949008113983933e-07,
      "loss": 1.2315,
      "step": 1772
    },
    {
      "epoch": 0.9004570848146267,
      "grad_norm": 6.498187065124512,
      "learning_rate": 4.899305211170102e-07,
      "loss": 1.1394,
      "step": 1773
    },
    {
      "epoch": 0.9009649568308786,
      "grad_norm": 5.5723772048950195,
      "learning_rate": 4.849846884712317e-07,
      "loss": 1.1502,
      "step": 1774
    },
    {
      "epoch": 0.9014728288471305,
      "grad_norm": 4.490263938903809,
      "learning_rate": 4.800633261805632e-07,
      "loss": 1.3187,
      "step": 1775
    },
    {
      "epoch": 0.9019807008633824,
      "grad_norm": 4.024411201477051,
      "learning_rate": 4.751664469015771e-07,
      "loss": 1.2368,
      "step": 1776
    },
    {
      "epoch": 0.9024885728796344,
      "grad_norm": 5.656371593475342,
      "learning_rate": 4.7029406322788186e-07,
      "loss": 1.2759,
      "step": 1777
    },
    {
      "epoch": 0.9029964448958863,
      "grad_norm": 4.788447380065918,
      "learning_rate": 4.654461876900895e-07,
      "loss": 1.2423,
      "step": 1778
    },
    {
      "epoch": 0.9035043169121382,
      "grad_norm": 4.719852447509766,
      "learning_rate": 4.6062283275578045e-07,
      "loss": 1.2496,
      "step": 1779
    },
    {
      "epoch": 0.9040121889283901,
      "grad_norm": 5.475706577301025,
      "learning_rate": 4.5582401082947867e-07,
      "loss": 1.3311,
      "step": 1780
    },
    {
      "epoch": 0.904520060944642,
      "grad_norm": 5.003331661224365,
      "learning_rate": 4.510497342526088e-07,
      "loss": 1.291,
      "step": 1781
    },
    {
      "epoch": 0.9050279329608939,
      "grad_norm": 4.8336381912231445,
      "learning_rate": 4.463000153034769e-07,
      "loss": 1.3825,
      "step": 1782
    },
    {
      "epoch": 0.9055358049771458,
      "grad_norm": 4.646341800689697,
      "learning_rate": 4.415748661972297e-07,
      "loss": 1.3562,
      "step": 1783
    },
    {
      "epoch": 0.9060436769933977,
      "grad_norm": 5.7394914627075195,
      "learning_rate": 4.3687429908582323e-07,
      "loss": 1.3305,
      "step": 1784
    },
    {
      "epoch": 0.9065515490096495,
      "grad_norm": 4.842685699462891,
      "learning_rate": 4.3219832605800093e-07,
      "loss": 1.4796,
      "step": 1785
    },
    {
      "epoch": 0.9070594210259014,
      "grad_norm": 5.051614284515381,
      "learning_rate": 4.275469591392511e-07,
      "loss": 1.2763,
      "step": 1786
    },
    {
      "epoch": 0.9075672930421533,
      "grad_norm": 4.732386589050293,
      "learning_rate": 4.229202102917818e-07,
      "loss": 1.3074,
      "step": 1787
    },
    {
      "epoch": 0.9080751650584052,
      "grad_norm": 4.5038251876831055,
      "learning_rate": 4.183180914144924e-07,
      "loss": 1.2547,
      "step": 1788
    },
    {
      "epoch": 0.9085830370746572,
      "grad_norm": 6.258203506469727,
      "learning_rate": 4.137406143429357e-07,
      "loss": 1.232,
      "step": 1789
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 4.272420883178711,
      "learning_rate": 4.0918779084929363e-07,
      "loss": 1.1927,
      "step": 1790
    },
    {
      "epoch": 0.909598781107161,
      "grad_norm": 4.761685371398926,
      "learning_rate": 4.0465963264234464e-07,
      "loss": 1.144,
      "step": 1791
    },
    {
      "epoch": 0.9101066531234129,
      "grad_norm": 6.755512237548828,
      "learning_rate": 4.001561513674346e-07,
      "loss": 1.0257,
      "step": 1792
    },
    {
      "epoch": 0.9106145251396648,
      "grad_norm": 4.225400924682617,
      "learning_rate": 3.956773586064455e-07,
      "loss": 1.2539,
      "step": 1793
    },
    {
      "epoch": 0.9111223971559167,
      "grad_norm": 6.154450416564941,
      "learning_rate": 3.912232658777659e-07,
      "loss": 1.2094,
      "step": 1794
    },
    {
      "epoch": 0.9116302691721686,
      "grad_norm": 5.3774333000183105,
      "learning_rate": 3.867938846362651e-07,
      "loss": 1.0695,
      "step": 1795
    },
    {
      "epoch": 0.9121381411884205,
      "grad_norm": 4.787917613983154,
      "learning_rate": 3.823892262732543e-07,
      "loss": 1.282,
      "step": 1796
    },
    {
      "epoch": 0.9126460132046724,
      "grad_norm": 5.000186920166016,
      "learning_rate": 3.780093021164677e-07,
      "loss": 1.3229,
      "step": 1797
    },
    {
      "epoch": 0.9131538852209243,
      "grad_norm": 5.02613639831543,
      "learning_rate": 3.736541234300284e-07,
      "loss": 1.3228,
      "step": 1798
    },
    {
      "epoch": 0.9136617572371762,
      "grad_norm": 5.472193717956543,
      "learning_rate": 3.693237014144191e-07,
      "loss": 0.9133,
      "step": 1799
    },
    {
      "epoch": 0.9141696292534282,
      "grad_norm": 4.462389945983887,
      "learning_rate": 3.650180472064546e-07,
      "loss": 1.2428,
      "step": 1800
    },
    {
      "epoch": 0.9146775012696801,
      "grad_norm": 5.741103649139404,
      "learning_rate": 3.607371718792541e-07,
      "loss": 1.1856,
      "step": 1801
    },
    {
      "epoch": 0.915185373285932,
      "grad_norm": 4.52320671081543,
      "learning_rate": 3.564810864422097e-07,
      "loss": 1.2286,
      "step": 1802
    },
    {
      "epoch": 0.9156932453021839,
      "grad_norm": 5.195258617401123,
      "learning_rate": 3.5224980184095926e-07,
      "loss": 1.2727,
      "step": 1803
    },
    {
      "epoch": 0.9162011173184358,
      "grad_norm": 5.747450351715088,
      "learning_rate": 3.4804332895735924e-07,
      "loss": 1.0976,
      "step": 1804
    },
    {
      "epoch": 0.9167089893346877,
      "grad_norm": 4.624321460723877,
      "learning_rate": 3.4386167860946065e-07,
      "loss": 1.0627,
      "step": 1805
    },
    {
      "epoch": 0.9172168613509396,
      "grad_norm": 4.598133087158203,
      "learning_rate": 3.3970486155146864e-07,
      "loss": 1.1602,
      "step": 1806
    },
    {
      "epoch": 0.9177247333671915,
      "grad_norm": 4.223515033721924,
      "learning_rate": 3.355728884737297e-07,
      "loss": 1.4767,
      "step": 1807
    },
    {
      "epoch": 0.9182326053834434,
      "grad_norm": 4.547367572784424,
      "learning_rate": 3.3146577000269573e-07,
      "loss": 0.9494,
      "step": 1808
    },
    {
      "epoch": 0.9187404773996953,
      "grad_norm": 4.53663969039917,
      "learning_rate": 3.2738351670089764e-07,
      "loss": 1.1722,
      "step": 1809
    },
    {
      "epoch": 0.9192483494159471,
      "grad_norm": 5.274203777313232,
      "learning_rate": 3.2332613906691846e-07,
      "loss": 1.0009,
      "step": 1810
    },
    {
      "epoch": 0.919756221432199,
      "grad_norm": 5.340970039367676,
      "learning_rate": 3.192936475353703e-07,
      "loss": 1.221,
      "step": 1811
    },
    {
      "epoch": 0.920264093448451,
      "grad_norm": 4.922206401824951,
      "learning_rate": 3.1528605247685974e-07,
      "loss": 1.3697,
      "step": 1812
    },
    {
      "epoch": 0.9207719654647029,
      "grad_norm": 4.887393951416016,
      "learning_rate": 3.113033641979679e-07,
      "loss": 1.3847,
      "step": 1813
    },
    {
      "epoch": 0.9212798374809548,
      "grad_norm": 5.808061599731445,
      "learning_rate": 3.073455929412217e-07,
      "loss": 1.1314,
      "step": 1814
    },
    {
      "epoch": 0.9217877094972067,
      "grad_norm": 5.183882713317871,
      "learning_rate": 3.034127488850669e-07,
      "loss": 1.2961,
      "step": 1815
    },
    {
      "epoch": 0.9222955815134586,
      "grad_norm": 5.175242900848389,
      "learning_rate": 2.995048421438407e-07,
      "loss": 1.4633,
      "step": 1816
    },
    {
      "epoch": 0.9228034535297105,
      "grad_norm": 5.085374355316162,
      "learning_rate": 2.956218827677526e-07,
      "loss": 1.1137,
      "step": 1817
    },
    {
      "epoch": 0.9233113255459624,
      "grad_norm": 4.4305267333984375,
      "learning_rate": 2.917638807428491e-07,
      "loss": 1.3338,
      "step": 1818
    },
    {
      "epoch": 0.9238191975622143,
      "grad_norm": 4.831226825714111,
      "learning_rate": 2.879308459909913e-07,
      "loss": 1.2896,
      "step": 1819
    },
    {
      "epoch": 0.9243270695784662,
      "grad_norm": 4.863157749176025,
      "learning_rate": 2.84122788369835e-07,
      "loss": 1.1717,
      "step": 1820
    },
    {
      "epoch": 0.9248349415947181,
      "grad_norm": 4.739955425262451,
      "learning_rate": 2.803397176727951e-07,
      "loss": 1.2834,
      "step": 1821
    },
    {
      "epoch": 0.92534281361097,
      "grad_norm": 5.42057466506958,
      "learning_rate": 2.765816436290325e-07,
      "loss": 1.1681,
      "step": 1822
    },
    {
      "epoch": 0.925850685627222,
      "grad_norm": 4.562961578369141,
      "learning_rate": 2.7284857590341715e-07,
      "loss": 1.1497,
      "step": 1823
    },
    {
      "epoch": 0.9263585576434739,
      "grad_norm": 4.59310245513916,
      "learning_rate": 2.691405240965128e-07,
      "loss": 1.2458,
      "step": 1824
    },
    {
      "epoch": 0.9268664296597258,
      "grad_norm": 5.141499042510986,
      "learning_rate": 2.6545749774454564e-07,
      "loss": 1.2726,
      "step": 1825
    },
    {
      "epoch": 0.9273743016759777,
      "grad_norm": 4.8967766761779785,
      "learning_rate": 2.617995063193812e-07,
      "loss": 1.1715,
      "step": 1826
    },
    {
      "epoch": 0.9278821736922296,
      "grad_norm": 5.117003440856934,
      "learning_rate": 2.581665592285054e-07,
      "loss": 1.1912,
      "step": 1827
    },
    {
      "epoch": 0.9283900457084815,
      "grad_norm": 4.560568809509277,
      "learning_rate": 2.5455866581499234e-07,
      "loss": 1.2921,
      "step": 1828
    },
    {
      "epoch": 0.9288979177247334,
      "grad_norm": 5.413818359375,
      "learning_rate": 2.5097583535748424e-07,
      "loss": 1.374,
      "step": 1829
    },
    {
      "epoch": 0.9294057897409853,
      "grad_norm": 5.509709358215332,
      "learning_rate": 2.4741807707017174e-07,
      "loss": 1.1772,
      "step": 1830
    },
    {
      "epoch": 0.9299136617572372,
      "grad_norm": 5.931797027587891,
      "learning_rate": 2.4388540010275906e-07,
      "loss": 1.05,
      "step": 1831
    },
    {
      "epoch": 0.9304215337734891,
      "grad_norm": 5.164069175720215,
      "learning_rate": 2.403778135404533e-07,
      "loss": 1.3111,
      "step": 1832
    },
    {
      "epoch": 0.930929405789741,
      "grad_norm": 4.883148670196533,
      "learning_rate": 2.3689532640392976e-07,
      "loss": 1.2181,
      "step": 1833
    },
    {
      "epoch": 0.9314372778059928,
      "grad_norm": 4.419644832611084,
      "learning_rate": 2.3343794764931759e-07,
      "loss": 1.327,
      "step": 1834
    },
    {
      "epoch": 0.9319451498222447,
      "grad_norm": 4.260985851287842,
      "learning_rate": 2.3000568616817432e-07,
      "loss": 1.3791,
      "step": 1835
    },
    {
      "epoch": 0.9324530218384967,
      "grad_norm": 4.370043754577637,
      "learning_rate": 2.2659855078745796e-07,
      "loss": 1.1643,
      "step": 1836
    },
    {
      "epoch": 0.9329608938547486,
      "grad_norm": 5.112608432769775,
      "learning_rate": 2.2321655026950938e-07,
      "loss": 1.4418,
      "step": 1837
    },
    {
      "epoch": 0.9334687658710005,
      "grad_norm": 5.211599349975586,
      "learning_rate": 2.1985969331203117e-07,
      "loss": 1.3613,
      "step": 1838
    },
    {
      "epoch": 0.9339766378872524,
      "grad_norm": 4.372023582458496,
      "learning_rate": 2.1652798854805756e-07,
      "loss": 1.4388,
      "step": 1839
    },
    {
      "epoch": 0.9344845099035043,
      "grad_norm": 4.861570835113525,
      "learning_rate": 2.132214445459424e-07,
      "loss": 1.3352,
      "step": 1840
    },
    {
      "epoch": 0.9349923819197562,
      "grad_norm": 4.848053455352783,
      "learning_rate": 2.0994006980933124e-07,
      "loss": 1.0425,
      "step": 1841
    },
    {
      "epoch": 0.9355002539360081,
      "grad_norm": 5.172409534454346,
      "learning_rate": 2.0668387277713474e-07,
      "loss": 1.0898,
      "step": 1842
    },
    {
      "epoch": 0.93600812595226,
      "grad_norm": 5.695395469665527,
      "learning_rate": 2.0345286182352208e-07,
      "loss": 1.1312,
      "step": 1843
    },
    {
      "epoch": 0.9365159979685119,
      "grad_norm": 4.555364608764648,
      "learning_rate": 2.0024704525788086e-07,
      "loss": 1.3331,
      "step": 1844
    },
    {
      "epoch": 0.9370238699847638,
      "grad_norm": 4.197620391845703,
      "learning_rate": 1.9706643132480828e-07,
      "loss": 1.3017,
      "step": 1845
    },
    {
      "epoch": 0.9375317420010157,
      "grad_norm": 4.748517036437988,
      "learning_rate": 1.9391102820409014e-07,
      "loss": 1.032,
      "step": 1846
    },
    {
      "epoch": 0.9380396140172677,
      "grad_norm": 4.709668159484863,
      "learning_rate": 1.9078084401066844e-07,
      "loss": 1.4075,
      "step": 1847
    },
    {
      "epoch": 0.9385474860335196,
      "grad_norm": 5.260051727294922,
      "learning_rate": 1.8767588679463377e-07,
      "loss": 1.0771,
      "step": 1848
    },
    {
      "epoch": 0.9390553580497715,
      "grad_norm": 4.284242630004883,
      "learning_rate": 1.8459616454119755e-07,
      "loss": 1.1278,
      "step": 1849
    },
    {
      "epoch": 0.9395632300660234,
      "grad_norm": 4.960768699645996,
      "learning_rate": 1.8154168517067083e-07,
      "loss": 1.238,
      "step": 1850
    },
    {
      "epoch": 0.9400711020822753,
      "grad_norm": 4.591121673583984,
      "learning_rate": 1.7851245653844994e-07,
      "loss": 1.2979,
      "step": 1851
    },
    {
      "epoch": 0.9405789740985272,
      "grad_norm": 4.424968242645264,
      "learning_rate": 1.7550848643498763e-07,
      "loss": 1.2654,
      "step": 1852
    },
    {
      "epoch": 0.9410868461147791,
      "grad_norm": 5.394791126251221,
      "learning_rate": 1.72529782585783e-07,
      "loss": 1.3486,
      "step": 1853
    },
    {
      "epoch": 0.941594718131031,
      "grad_norm": 6.6211137771606445,
      "learning_rate": 1.6957635265135165e-07,
      "loss": 0.9716,
      "step": 1854
    },
    {
      "epoch": 0.9421025901472829,
      "grad_norm": 5.488575458526611,
      "learning_rate": 1.6664820422721216e-07,
      "loss": 1.2914,
      "step": 1855
    },
    {
      "epoch": 0.9426104621635348,
      "grad_norm": 4.948568344116211,
      "learning_rate": 1.6374534484386639e-07,
      "loss": 1.2524,
      "step": 1856
    },
    {
      "epoch": 0.9431183341797867,
      "grad_norm": 5.84658145904541,
      "learning_rate": 1.6086778196677922e-07,
      "loss": 1.374,
      "step": 1857
    },
    {
      "epoch": 0.9436262061960387,
      "grad_norm": 4.914611339569092,
      "learning_rate": 1.5801552299635426e-07,
      "loss": 1.4766,
      "step": 1858
    },
    {
      "epoch": 0.9441340782122905,
      "grad_norm": 5.559576988220215,
      "learning_rate": 1.5518857526792496e-07,
      "loss": 1.2248,
      "step": 1859
    },
    {
      "epoch": 0.9446419502285424,
      "grad_norm": 5.238275051116943,
      "learning_rate": 1.523869460517291e-07,
      "loss": 1.3467,
      "step": 1860
    },
    {
      "epoch": 0.9451498222447943,
      "grad_norm": 5.570143699645996,
      "learning_rate": 1.496106425528887e-07,
      "loss": 1.4734,
      "step": 1861
    },
    {
      "epoch": 0.9456576942610462,
      "grad_norm": 4.6389594078063965,
      "learning_rate": 1.4685967191139462e-07,
      "loss": 1.4379,
      "step": 1862
    },
    {
      "epoch": 0.9461655662772981,
      "grad_norm": 5.283039569854736,
      "learning_rate": 1.4413404120209084e-07,
      "loss": 1.3353,
      "step": 1863
    },
    {
      "epoch": 0.94667343829355,
      "grad_norm": 4.294154644012451,
      "learning_rate": 1.4143375743465026e-07,
      "loss": 1.2644,
      "step": 1864
    },
    {
      "epoch": 0.9471813103098019,
      "grad_norm": 4.243343353271484,
      "learning_rate": 1.3875882755356008e-07,
      "loss": 1.4211,
      "step": 1865
    },
    {
      "epoch": 0.9476891823260538,
      "grad_norm": 6.04444694519043,
      "learning_rate": 1.3610925843810408e-07,
      "loss": 1.1202,
      "step": 1866
    },
    {
      "epoch": 0.9481970543423057,
      "grad_norm": 5.779665946960449,
      "learning_rate": 1.334850569023438e-07,
      "loss": 1.0796,
      "step": 1867
    },
    {
      "epoch": 0.9487049263585576,
      "grad_norm": 5.141113758087158,
      "learning_rate": 1.3088622969509856e-07,
      "loss": 1.4375,
      "step": 1868
    },
    {
      "epoch": 0.9492127983748095,
      "grad_norm": 5.844571113586426,
      "learning_rate": 1.2831278349993536e-07,
      "loss": 1.1614,
      "step": 1869
    },
    {
      "epoch": 0.9497206703910615,
      "grad_norm": 4.52020263671875,
      "learning_rate": 1.2576472493514567e-07,
      "loss": 1.4625,
      "step": 1870
    },
    {
      "epoch": 0.9502285424073134,
      "grad_norm": 6.979033470153809,
      "learning_rate": 1.2324206055372878e-07,
      "loss": 1.1936,
      "step": 1871
    },
    {
      "epoch": 0.9507364144235653,
      "grad_norm": 5.365548133850098,
      "learning_rate": 1.2074479684337836e-07,
      "loss": 1.4185,
      "step": 1872
    },
    {
      "epoch": 0.9512442864398172,
      "grad_norm": 4.7730021476745605,
      "learning_rate": 1.1827294022646152e-07,
      "loss": 1.1636,
      "step": 1873
    },
    {
      "epoch": 0.9517521584560691,
      "grad_norm": 5.19934606552124,
      "learning_rate": 1.1582649706000537e-07,
      "loss": 1.1086,
      "step": 1874
    },
    {
      "epoch": 0.952260030472321,
      "grad_norm": 4.595972061157227,
      "learning_rate": 1.1340547363567932e-07,
      "loss": 1.1832,
      "step": 1875
    },
    {
      "epoch": 0.9527679024885729,
      "grad_norm": 5.226284980773926,
      "learning_rate": 1.1100987617978065e-07,
      "loss": 1.1639,
      "step": 1876
    },
    {
      "epoch": 0.9532757745048248,
      "grad_norm": 5.403947830200195,
      "learning_rate": 1.0863971085321334e-07,
      "loss": 1.3828,
      "step": 1877
    },
    {
      "epoch": 0.9537836465210767,
      "grad_norm": 4.61500358581543,
      "learning_rate": 1.0629498375148151e-07,
      "loss": 1.2726,
      "step": 1878
    },
    {
      "epoch": 0.9542915185373286,
      "grad_norm": 4.211429595947266,
      "learning_rate": 1.0397570090466491e-07,
      "loss": 1.3503,
      "step": 1879
    },
    {
      "epoch": 0.9547993905535805,
      "grad_norm": 5.15771484375,
      "learning_rate": 1.0168186827740567e-07,
      "loss": 1.2994,
      "step": 1880
    },
    {
      "epoch": 0.9553072625698324,
      "grad_norm": 4.26914644241333,
      "learning_rate": 9.941349176889492e-08,
      "loss": 1.3755,
      "step": 1881
    },
    {
      "epoch": 0.9558151345860844,
      "grad_norm": 4.972912311553955,
      "learning_rate": 9.71705772128606e-08,
      "loss": 1.2771,
      "step": 1882
    },
    {
      "epoch": 0.9563230066023362,
      "grad_norm": 7.042515754699707,
      "learning_rate": 9.495313037754195e-08,
      "loss": 1.2627,
      "step": 1883
    },
    {
      "epoch": 0.9568308786185881,
      "grad_norm": 6.592982769012451,
      "learning_rate": 9.2761156965685e-08,
      "loss": 1.3186,
      "step": 1884
    },
    {
      "epoch": 0.95733875063484,
      "grad_norm": 7.000082969665527,
      "learning_rate": 9.059466261452598e-08,
      "loss": 0.9216,
      "step": 1885
    },
    {
      "epoch": 0.9578466226510919,
      "grad_norm": 4.61784029006958,
      "learning_rate": 8.845365289577023e-08,
      "loss": 1.0449,
      "step": 1886
    },
    {
      "epoch": 0.9583544946673438,
      "grad_norm": 4.5391740798950195,
      "learning_rate": 8.633813331558549e-08,
      "loss": 1.383,
      "step": 1887
    },
    {
      "epoch": 0.9588623666835957,
      "grad_norm": 6.534933090209961,
      "learning_rate": 8.424810931458638e-08,
      "loss": 0.813,
      "step": 1888
    },
    {
      "epoch": 0.9593702386998476,
      "grad_norm": 5.744078636169434,
      "learning_rate": 8.218358626781663e-08,
      "loss": 1.2413,
      "step": 1889
    },
    {
      "epoch": 0.9598781107160995,
      "grad_norm": 5.97611141204834,
      "learning_rate": 8.014456948473692e-08,
      "loss": 1.0117,
      "step": 1890
    },
    {
      "epoch": 0.9603859827323514,
      "grad_norm": 6.729101657867432,
      "learning_rate": 7.81310642092159e-08,
      "loss": 1.0957,
      "step": 1891
    },
    {
      "epoch": 0.9608938547486033,
      "grad_norm": 4.487152576446533,
      "learning_rate": 7.614307561950585e-08,
      "loss": 1.4766,
      "step": 1892
    },
    {
      "epoch": 0.9614017267648552,
      "grad_norm": 5.25182580947876,
      "learning_rate": 7.418060882824485e-08,
      "loss": 1.3268,
      "step": 1893
    },
    {
      "epoch": 0.9619095987811072,
      "grad_norm": 5.395699501037598,
      "learning_rate": 7.224366888242685e-08,
      "loss": 1.2018,
      "step": 1894
    },
    {
      "epoch": 0.9624174707973591,
      "grad_norm": 6.704184055328369,
      "learning_rate": 7.033226076340382e-08,
      "loss": 1.1289,
      "step": 1895
    },
    {
      "epoch": 0.962925342813611,
      "grad_norm": 4.314540386199951,
      "learning_rate": 6.844638938685921e-08,
      "loss": 1.3382,
      "step": 1896
    },
    {
      "epoch": 0.9634332148298629,
      "grad_norm": 5.0664381980896,
      "learning_rate": 6.658605960280785e-08,
      "loss": 1.1835,
      "step": 1897
    },
    {
      "epoch": 0.9639410868461148,
      "grad_norm": 4.805373191833496,
      "learning_rate": 6.475127619557376e-08,
      "loss": 1.0936,
      "step": 1898
    },
    {
      "epoch": 0.9644489588623667,
      "grad_norm": 4.490488529205322,
      "learning_rate": 6.294204388378467e-08,
      "loss": 1.2735,
      "step": 1899
    },
    {
      "epoch": 0.9649568308786186,
      "grad_norm": 5.072920799255371,
      "learning_rate": 6.115836732035418e-08,
      "loss": 1.2059,
      "step": 1900
    },
    {
      "epoch": 0.9654647028948705,
      "grad_norm": 4.6457390785217285,
      "learning_rate": 5.940025109247516e-08,
      "loss": 1.1965,
      "step": 1901
    },
    {
      "epoch": 0.9659725749111224,
      "grad_norm": 5.324655055999756,
      "learning_rate": 5.766769972160302e-08,
      "loss": 1.3515,
      "step": 1902
    },
    {
      "epoch": 0.9664804469273743,
      "grad_norm": 4.400649547576904,
      "learning_rate": 5.596071766344802e-08,
      "loss": 1.2527,
      "step": 1903
    },
    {
      "epoch": 0.9669883189436262,
      "grad_norm": 4.642717361450195,
      "learning_rate": 5.427930930796188e-08,
      "loss": 1.2362,
      "step": 1904
    },
    {
      "epoch": 0.9674961909598782,
      "grad_norm": 5.073624134063721,
      "learning_rate": 5.2623478979326736e-08,
      "loss": 1.2601,
      "step": 1905
    },
    {
      "epoch": 0.9680040629761301,
      "grad_norm": 4.679571151733398,
      "learning_rate": 5.09932309359451e-08,
      "loss": 0.9734,
      "step": 1906
    },
    {
      "epoch": 0.968511934992382,
      "grad_norm": 4.934381484985352,
      "learning_rate": 4.9388569370424354e-08,
      "loss": 1.1064,
      "step": 1907
    },
    {
      "epoch": 0.9690198070086338,
      "grad_norm": 5.23956298828125,
      "learning_rate": 4.78094984095756e-08,
      "loss": 1.1123,
      "step": 1908
    },
    {
      "epoch": 0.9695276790248857,
      "grad_norm": 4.792849540710449,
      "learning_rate": 4.625602211439151e-08,
      "loss": 1.2325,
      "step": 1909
    },
    {
      "epoch": 0.9700355510411376,
      "grad_norm": 3.939530372619629,
      "learning_rate": 4.472814448004403e-08,
      "loss": 1.4759,
      "step": 1910
    },
    {
      "epoch": 0.9705434230573895,
      "grad_norm": 4.985993385314941,
      "learning_rate": 4.322586943587004e-08,
      "loss": 1.0928,
      "step": 1911
    },
    {
      "epoch": 0.9710512950736414,
      "grad_norm": 4.750078201293945,
      "learning_rate": 4.17492008453646e-08,
      "loss": 1.2729,
      "step": 1912
    },
    {
      "epoch": 0.9715591670898933,
      "grad_norm": 5.804688930511475,
      "learning_rate": 4.0298142506166594e-08,
      "loss": 1.2137,
      "step": 1913
    },
    {
      "epoch": 0.9720670391061452,
      "grad_norm": 4.8215131759643555,
      "learning_rate": 3.887269815005423e-08,
      "loss": 1.1932,
      "step": 1914
    },
    {
      "epoch": 0.9725749111223971,
      "grad_norm": 5.2581868171691895,
      "learning_rate": 3.747287144293066e-08,
      "loss": 1.3886,
      "step": 1915
    },
    {
      "epoch": 0.973082783138649,
      "grad_norm": 5.5822930335998535,
      "learning_rate": 3.609866598481726e-08,
      "loss": 1.0071,
      "step": 1916
    },
    {
      "epoch": 0.973590655154901,
      "grad_norm": 4.685520648956299,
      "learning_rate": 3.475008530984259e-08,
      "loss": 1.1638,
      "step": 1917
    },
    {
      "epoch": 0.9740985271711529,
      "grad_norm": 4.754659175872803,
      "learning_rate": 3.342713288623567e-08,
      "loss": 1.1589,
      "step": 1918
    },
    {
      "epoch": 0.9746063991874048,
      "grad_norm": 4.848739147186279,
      "learning_rate": 3.212981211631494e-08,
      "loss": 1.2732,
      "step": 1919
    },
    {
      "epoch": 0.9751142712036567,
      "grad_norm": 5.595327854156494,
      "learning_rate": 3.085812633648155e-08,
      "loss": 1.1356,
      "step": 1920
    },
    {
      "epoch": 0.9756221432199086,
      "grad_norm": 4.891111373901367,
      "learning_rate": 2.96120788172094e-08,
      "loss": 1.3382,
      "step": 1921
    },
    {
      "epoch": 0.9761300152361605,
      "grad_norm": 5.571377754211426,
      "learning_rate": 2.8391672763034005e-08,
      "loss": 1.2959,
      "step": 1922
    },
    {
      "epoch": 0.9766378872524124,
      "grad_norm": 4.298384189605713,
      "learning_rate": 2.719691131255031e-08,
      "loss": 1.3944,
      "step": 1923
    },
    {
      "epoch": 0.9771457592686643,
      "grad_norm": 4.778082847595215,
      "learning_rate": 2.6027797538401565e-08,
      "loss": 1.1293,
      "step": 1924
    },
    {
      "epoch": 0.9776536312849162,
      "grad_norm": 6.3270111083984375,
      "learning_rate": 2.4884334447268234e-08,
      "loss": 1.1693,
      "step": 1925
    },
    {
      "epoch": 0.9781615033011681,
      "grad_norm": 6.495800018310547,
      "learning_rate": 2.3766524979865757e-08,
      "loss": 1.218,
      "step": 1926
    },
    {
      "epoch": 0.97866937531742,
      "grad_norm": 4.57327127456665,
      "learning_rate": 2.2674372010935696e-08,
      "loss": 1.399,
      "step": 1927
    },
    {
      "epoch": 0.979177247333672,
      "grad_norm": 4.237867832183838,
      "learning_rate": 2.1607878349234614e-08,
      "loss": 1.151,
      "step": 1928
    },
    {
      "epoch": 0.9796851193499239,
      "grad_norm": 5.3051838874816895,
      "learning_rate": 2.056704673752963e-08,
      "loss": 0.9802,
      "step": 1929
    },
    {
      "epoch": 0.9801929913661758,
      "grad_norm": 6.228868007659912,
      "learning_rate": 1.955187985259288e-08,
      "loss": 1.3583,
      "step": 1930
    },
    {
      "epoch": 0.9807008633824277,
      "grad_norm": 5.250242233276367,
      "learning_rate": 1.856238030519153e-08,
      "loss": 1.231,
      "step": 1931
    },
    {
      "epoch": 0.9812087353986796,
      "grad_norm": 5.9555158615112305,
      "learning_rate": 1.75985506400822e-08,
      "loss": 0.95,
      "step": 1932
    },
    {
      "epoch": 0.9817166074149314,
      "grad_norm": 5.453784942626953,
      "learning_rate": 1.6660393336007665e-08,
      "loss": 1.1742,
      "step": 1933
    },
    {
      "epoch": 0.9822244794311833,
      "grad_norm": 4.261223793029785,
      "learning_rate": 1.5747910805683496e-08,
      "loss": 1.2285,
      "step": 1934
    },
    {
      "epoch": 0.9827323514474352,
      "grad_norm": 5.803184986114502,
      "learning_rate": 1.4861105395798103e-08,
      "loss": 1.2187,
      "step": 1935
    },
    {
      "epoch": 0.9832402234636871,
      "grad_norm": 4.059924125671387,
      "learning_rate": 1.3999979387003814e-08,
      "loss": 1.4354,
      "step": 1936
    },
    {
      "epoch": 0.983748095479939,
      "grad_norm": 4.6581010818481445,
      "learning_rate": 1.3164534993912458e-08,
      "loss": 1.3204,
      "step": 1937
    },
    {
      "epoch": 0.9842559674961909,
      "grad_norm": 4.7350172996521,
      "learning_rate": 1.2354774365086475e-08,
      "loss": 1.2787,
      "step": 1938
    },
    {
      "epoch": 0.9847638395124428,
      "grad_norm": 5.925784587860107,
      "learning_rate": 1.1570699583040023e-08,
      "loss": 1.3501,
      "step": 1939
    },
    {
      "epoch": 0.9852717115286947,
      "grad_norm": 4.791852951049805,
      "learning_rate": 1.0812312664224556e-08,
      "loss": 1.4943,
      "step": 1940
    },
    {
      "epoch": 0.9857795835449467,
      "grad_norm": 4.5251970291137695,
      "learning_rate": 1.0079615559033251e-08,
      "loss": 1.513,
      "step": 1941
    },
    {
      "epoch": 0.9862874555611986,
      "grad_norm": 5.248088359832764,
      "learning_rate": 9.372610151785477e-09,
      "loss": 1.335,
      "step": 1942
    },
    {
      "epoch": 0.9867953275774505,
      "grad_norm": 4.421478748321533,
      "learning_rate": 8.691298260733449e-09,
      "loss": 1.3018,
      "step": 1943
    },
    {
      "epoch": 0.9873031995937024,
      "grad_norm": 6.706859588623047,
      "learning_rate": 8.03568163804891e-09,
      "loss": 1.2363,
      "step": 1944
    },
    {
      "epoch": 0.9878110716099543,
      "grad_norm": 4.88031005859375,
      "learning_rate": 7.4057619698197914e-09,
      "loss": 1.103,
      "step": 1945
    },
    {
      "epoch": 0.9883189436262062,
      "grad_norm": 4.547206401824951,
      "learning_rate": 6.801540876052448e-09,
      "loss": 1.6001,
      "step": 1946
    },
    {
      "epoch": 0.9888268156424581,
      "grad_norm": 5.272502422332764,
      "learning_rate": 6.223019910657213e-09,
      "loss": 1.3305,
      "step": 1947
    },
    {
      "epoch": 0.98933468765871,
      "grad_norm": 4.947421073913574,
      "learning_rate": 5.670200561453953e-09,
      "loss": 0.9495,
      "step": 1948
    },
    {
      "epoch": 0.9898425596749619,
      "grad_norm": 4.808927536010742,
      "learning_rate": 5.143084250162078e-09,
      "loss": 1.215,
      "step": 1949
    },
    {
      "epoch": 0.9903504316912138,
      "grad_norm": 5.339200973510742,
      "learning_rate": 4.641672332399427e-09,
      "loss": 0.9683,
      "step": 1950
    },
    {
      "epoch": 0.9908583037074657,
      "grad_norm": 4.209288597106934,
      "learning_rate": 4.165966097677832e-09,
      "loss": 1.3954,
      "step": 1951
    },
    {
      "epoch": 0.9913661757237177,
      "grad_norm": 6.088907718658447,
      "learning_rate": 3.7159667694008915e-09,
      "loss": 1.0968,
      "step": 1952
    },
    {
      "epoch": 0.9918740477399696,
      "grad_norm": 4.938228130340576,
      "learning_rate": 3.2916755048595374e-09,
      "loss": 1.1232,
      "step": 1953
    },
    {
      "epoch": 0.9923819197562215,
      "grad_norm": 4.1907525062561035,
      "learning_rate": 2.893093395229807e-09,
      "loss": 1.3027,
      "step": 1954
    },
    {
      "epoch": 0.9928897917724734,
      "grad_norm": 4.614527225494385,
      "learning_rate": 2.5202214655695168e-09,
      "loss": 1.1737,
      "step": 1955
    },
    {
      "epoch": 0.9933976637887253,
      "grad_norm": 5.144586086273193,
      "learning_rate": 2.1730606748182615e-09,
      "loss": 1.1848,
      "step": 1956
    },
    {
      "epoch": 0.9939055358049771,
      "grad_norm": 4.373411655426025,
      "learning_rate": 1.8516119157896416e-09,
      "loss": 1.313,
      "step": 1957
    },
    {
      "epoch": 0.994413407821229,
      "grad_norm": 5.278648376464844,
      "learning_rate": 1.5558760151734854e-09,
      "loss": 1.3557,
      "step": 1958
    },
    {
      "epoch": 0.9949212798374809,
      "grad_norm": 4.67554235458374,
      "learning_rate": 1.2858537335314058e-09,
      "loss": 1.2227,
      "step": 1959
    },
    {
      "epoch": 0.9954291518537328,
      "grad_norm": 5.464794635772705,
      "learning_rate": 1.0415457652990236e-09,
      "loss": 1.2804,
      "step": 1960
    },
    {
      "epoch": 0.9959370238699847,
      "grad_norm": 5.86653995513916,
      "learning_rate": 8.229527387748626e-10,
      "loss": 1.2311,
      "step": 1961
    },
    {
      "epoch": 0.9964448958862366,
      "grad_norm": 4.377933502197266,
      "learning_rate": 6.300752161314539e-10,
      "loss": 1.3215,
      "step": 1962
    },
    {
      "epoch": 0.9969527679024885,
      "grad_norm": 4.7609782218933105,
      "learning_rate": 4.6291369340201173e-10,
      "loss": 1.273,
      "step": 1963
    },
    {
      "epoch": 0.9974606399187405,
      "grad_norm": 4.997765064239502,
      "learning_rate": 3.214686004859857e-10,
      "loss": 1.3889,
      "step": 1964
    },
    {
      "epoch": 0.9979685119349924,
      "grad_norm": 4.37820291519165,
      "learning_rate": 2.0574030114794975e-10,
      "loss": 1.0622,
      "step": 1965
    },
    {
      "epoch": 0.9984763839512443,
      "grad_norm": 5.484806060791016,
      "learning_rate": 1.1572909301205137e-10,
      "loss": 1.4135,
      "step": 1966
    },
    {
      "epoch": 0.9989842559674962,
      "grad_norm": 4.66874361038208,
      "learning_rate": 5.143520756534237e-11,
      "loss": 1.3501,
      "step": 1967
    },
    {
      "epoch": 0.9994921279837481,
      "grad_norm": 4.561672687530518,
      "learning_rate": 1.2858810158888902e-11,
      "loss": 1.2979,
      "step": 1968
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.26925802230835,
      "learning_rate": 0.0,
      "loss": 1.2135,
      "step": 1969
    }
  ],
  "logging_steps": 1,
  "max_steps": 1969,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.689577521152e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}