{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1374,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002186987424822307,
      "grad_norm": 0.019816486164927483,
      "learning_rate": 0.0,
      "loss": 0.6513,
      "step": 1
    },
    {
      "epoch": 0.004373974849644614,
      "grad_norm": 0.02367558144032955,
      "learning_rate": 2.1739130434782607e-06,
      "loss": 0.6461,
      "step": 2
    },
    {
      "epoch": 0.0065609622744669215,
      "grad_norm": 0.016746528446674347,
      "learning_rate": 4.347826086956521e-06,
      "loss": 0.5865,
      "step": 3
    },
    {
      "epoch": 0.008747949699289229,
      "grad_norm": 0.028324777260422707,
      "learning_rate": 6.521739130434782e-06,
      "loss": 0.4125,
      "step": 4
    },
    {
      "epoch": 0.010934937124111536,
      "grad_norm": 0.024395287036895752,
      "learning_rate": 8.695652173913043e-06,
      "loss": 0.6942,
      "step": 5
    },
    {
      "epoch": 0.013121924548933843,
      "grad_norm": 0.021644193679094315,
      "learning_rate": 1.0869565217391303e-05,
      "loss": 0.9391,
      "step": 6
    },
    {
      "epoch": 0.01530891197375615,
      "grad_norm": 0.04051661118865013,
      "learning_rate": 1.3043478260869564e-05,
      "loss": 0.6903,
      "step": 7
    },
    {
      "epoch": 0.017495899398578457,
      "grad_norm": 0.02317817509174347,
      "learning_rate": 1.5217391304347826e-05,
      "loss": 0.7229,
      "step": 8
    },
    {
      "epoch": 0.019682886823400764,
      "grad_norm": 0.016089174896478653,
      "learning_rate": 1.7391304347826085e-05,
      "loss": 0.9602,
      "step": 9
    },
    {
      "epoch": 0.02186987424822307,
      "grad_norm": 0.029251758009195328,
      "learning_rate": 1.9565217391304346e-05,
      "loss": 0.9696,
      "step": 10
    },
    {
      "epoch": 0.02405686167304538,
      "grad_norm": 0.02963915839791298,
      "learning_rate": 2.1739130434782607e-05,
      "loss": 0.8149,
      "step": 11
    },
    {
      "epoch": 0.026243849097867686,
      "grad_norm": 0.02249906025826931,
      "learning_rate": 2.3913043478260864e-05,
      "loss": 0.8597,
      "step": 12
    },
    {
      "epoch": 0.028430836522689993,
      "grad_norm": 0.038812343031167984,
      "learning_rate": 2.6086956521739128e-05,
      "loss": 0.6965,
      "step": 13
    },
    {
      "epoch": 0.0306178239475123,
      "grad_norm": 0.1395931988954544,
      "learning_rate": 2.826086956521739e-05,
      "loss": 0.7026,
      "step": 14
    },
    {
      "epoch": 0.03280481137233461,
      "grad_norm": 0.04644119739532471,
      "learning_rate": 3.0434782608695653e-05,
      "loss": 0.5772,
      "step": 15
    },
    {
      "epoch": 0.034991798797156914,
      "grad_norm": 0.038876552134752274,
      "learning_rate": 3.260869565217391e-05,
      "loss": 0.6508,
      "step": 16
    },
    {
      "epoch": 0.037178786221979225,
      "grad_norm": 0.031183617189526558,
      "learning_rate": 3.478260869565217e-05,
      "loss": 0.7206,
      "step": 17
    },
    {
      "epoch": 0.03936577364680153,
      "grad_norm": 0.02312302775681019,
      "learning_rate": 3.695652173913043e-05,
      "loss": 0.6714,
      "step": 18
    },
    {
      "epoch": 0.04155276107162384,
      "grad_norm": 0.022964971140027046,
      "learning_rate": 3.913043478260869e-05,
      "loss": 1.0436,
      "step": 19
    },
    {
      "epoch": 0.04373974849644614,
      "grad_norm": 0.0303932037204504,
      "learning_rate": 4.130434782608695e-05,
      "loss": 0.7183,
      "step": 20
    },
    {
      "epoch": 0.045926735921268454,
      "grad_norm": 0.04829512909054756,
      "learning_rate": 4.3478260869565214e-05,
      "loss": 0.8273,
      "step": 21
    },
    {
      "epoch": 0.04811372334609076,
      "grad_norm": 0.03576242923736572,
      "learning_rate": 4.5652173913043474e-05,
      "loss": 0.7451,
      "step": 22
    },
    {
      "epoch": 0.05030071077091307,
      "grad_norm": 0.0263582281768322,
      "learning_rate": 4.782608695652173e-05,
      "loss": 0.7812,
      "step": 23
    },
    {
      "epoch": 0.05248769819573537,
      "grad_norm": 0.022913772612810135,
      "learning_rate": 4.9999999999999996e-05,
      "loss": 0.9077,
      "step": 24
    },
    {
      "epoch": 0.05467468562055768,
      "grad_norm": 0.026295918971300125,
      "learning_rate": 5.2173913043478256e-05,
      "loss": 0.5756,
      "step": 25
    },
    {
      "epoch": 0.056861673045379986,
      "grad_norm": 0.05335557088255882,
      "learning_rate": 5.434782608695652e-05,
      "loss": 0.6425,
      "step": 26
    },
    {
      "epoch": 0.0590486604702023,
      "grad_norm": 0.03465011343359947,
      "learning_rate": 5.652173913043478e-05,
      "loss": 1.1668,
      "step": 27
    },
    {
      "epoch": 0.0612356478950246,
      "grad_norm": 0.028022659942507744,
      "learning_rate": 5.869565217391304e-05,
      "loss": 0.661,
      "step": 28
    },
    {
      "epoch": 0.0634226353198469,
      "grad_norm": 0.0726475864648819,
      "learning_rate": 6.0869565217391306e-05,
      "loss": 0.6718,
      "step": 29
    },
    {
      "epoch": 0.06560962274466922,
      "grad_norm": 0.03730878606438637,
      "learning_rate": 6.304347826086956e-05,
      "loss": 0.743,
      "step": 30
    },
    {
      "epoch": 0.06779661016949153,
      "grad_norm": 0.08332774043083191,
      "learning_rate": 6.521739130434782e-05,
      "loss": 0.9742,
      "step": 31
    },
    {
      "epoch": 0.06998359759431383,
      "grad_norm": 0.05056551471352577,
      "learning_rate": 6.739130434782608e-05,
      "loss": 1.0044,
      "step": 32
    },
    {
      "epoch": 0.07217058501913615,
      "grad_norm": 0.04123561084270477,
      "learning_rate": 6.956521739130434e-05,
      "loss": 0.7357,
      "step": 33
    },
    {
      "epoch": 0.07435757244395845,
      "grad_norm": 0.03361466899514198,
      "learning_rate": 7.17391304347826e-05,
      "loss": 0.9187,
      "step": 34
    },
    {
      "epoch": 0.07654455986878075,
      "grad_norm": 0.07812534272670746,
      "learning_rate": 7.391304347826086e-05,
      "loss": 0.6135,
      "step": 35
    },
    {
      "epoch": 0.07873154729360306,
      "grad_norm": 0.036363691091537476,
      "learning_rate": 7.608695652173912e-05,
      "loss": 0.6875,
      "step": 36
    },
    {
      "epoch": 0.08091853471842538,
      "grad_norm": 0.07220917195081711,
      "learning_rate": 7.826086956521738e-05,
      "loss": 0.8439,
      "step": 37
    },
    {
      "epoch": 0.08310552214324768,
      "grad_norm": 0.08875200897455215,
      "learning_rate": 8.043478260869566e-05,
      "loss": 0.838,
      "step": 38
    },
    {
      "epoch": 0.08529250956806998,
      "grad_norm": 0.0466715507209301,
      "learning_rate": 8.26086956521739e-05,
      "loss": 0.7609,
      "step": 39
    },
    {
      "epoch": 0.08747949699289229,
      "grad_norm": 0.06119077280163765,
      "learning_rate": 8.478260869565217e-05,
      "loss": 0.5831,
      "step": 40
    },
    {
      "epoch": 0.0896664844177146,
      "grad_norm": 0.033799510449171066,
      "learning_rate": 8.695652173913043e-05,
      "loss": 0.7551,
      "step": 41
    },
    {
      "epoch": 0.09185347184253691,
      "grad_norm": 0.0656280517578125,
      "learning_rate": 8.913043478260869e-05,
      "loss": 0.7214,
      "step": 42
    },
    {
      "epoch": 0.09404045926735921,
      "grad_norm": 0.07249666750431061,
      "learning_rate": 9.130434782608695e-05,
      "loss": 0.7571,
      "step": 43
    },
    {
      "epoch": 0.09622744669218151,
      "grad_norm": 0.05747272074222565,
      "learning_rate": 9.347826086956521e-05,
      "loss": 0.7411,
      "step": 44
    },
    {
      "epoch": 0.09841443411700383,
      "grad_norm": 0.06284407526254654,
      "learning_rate": 9.565217391304346e-05,
      "loss": 0.6705,
      "step": 45
    },
    {
      "epoch": 0.10060142154182614,
      "grad_norm": 0.08808895945549011,
      "learning_rate": 9.782608695652173e-05,
      "loss": 0.7999,
      "step": 46
    },
    {
      "epoch": 0.10278840896664844,
      "grad_norm": 0.04671047255396843,
      "learning_rate": 9.999999999999999e-05,
      "loss": 0.7769,
      "step": 47
    },
    {
      "epoch": 0.10497539639147074,
      "grad_norm": 0.04966476559638977,
      "learning_rate": 0.00010217391304347825,
      "loss": 0.7384,
      "step": 48
    },
    {
      "epoch": 0.10716238381629306,
      "grad_norm": 0.03611300513148308,
      "learning_rate": 0.00010434782608695651,
      "loss": 1.3535,
      "step": 49
    },
    {
      "epoch": 0.10934937124111536,
      "grad_norm": 0.07358794659376144,
      "learning_rate": 0.00010652173913043477,
      "loss": 0.636,
      "step": 50
    },
    {
      "epoch": 0.11153635866593767,
      "grad_norm": 0.0816405862569809,
      "learning_rate": 0.00010869565217391303,
      "loss": 0.9147,
      "step": 51
    },
    {
      "epoch": 0.11372334609075997,
      "grad_norm": 0.027666306123137474,
      "learning_rate": 0.00011086956521739128,
      "loss": 0.8123,
      "step": 52
    },
    {
      "epoch": 0.11591033351558229,
      "grad_norm": 0.05469057708978653,
      "learning_rate": 0.00011304347826086956,
      "loss": 0.5093,
      "step": 53
    },
    {
      "epoch": 0.1180973209404046,
      "grad_norm": 0.06001636013388634,
      "learning_rate": 0.00011521739130434782,
      "loss": 0.5906,
      "step": 54
    },
    {
      "epoch": 0.1202843083652269,
      "grad_norm": 0.11656013131141663,
      "learning_rate": 0.00011739130434782608,
      "loss": 0.7212,
      "step": 55
    },
    {
      "epoch": 0.1224712957900492,
      "grad_norm": 0.05737491697072983,
      "learning_rate": 0.00011956521739130434,
      "loss": 0.9612,
      "step": 56
    },
    {
      "epoch": 0.12465828321487152,
      "grad_norm": 0.08259277790784836,
      "learning_rate": 0.00012173913043478261,
      "loss": 0.607,
      "step": 57
    },
    {
      "epoch": 0.1268452706396938,
      "grad_norm": 0.08760760724544525,
      "learning_rate": 0.00012391304347826086,
      "loss": 0.7577,
      "step": 58
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 0.0668894574046135,
      "learning_rate": 0.00012608695652173912,
      "loss": 0.6312,
      "step": 59
    },
    {
      "epoch": 0.13121924548933844,
      "grad_norm": 0.11188331991434097,
      "learning_rate": 0.00012826086956521738,
      "loss": 0.7738,
      "step": 60
    },
    {
      "epoch": 0.13340623291416073,
      "grad_norm": 0.1832694411277771,
      "learning_rate": 0.00013043478260869564,
      "loss": 0.752,
      "step": 61
    },
    {
      "epoch": 0.13559322033898305,
      "grad_norm": 0.08274513483047485,
      "learning_rate": 0.0001326086956521739,
      "loss": 0.4737,
      "step": 62
    },
    {
      "epoch": 0.13778020776380537,
      "grad_norm": 0.12250404059886932,
      "learning_rate": 0.00013478260869565216,
      "loss": 1.1355,
      "step": 63
    },
    {
      "epoch": 0.13996719518862766,
      "grad_norm": 0.09029451757669449,
      "learning_rate": 0.00013695652173913042,
      "loss": 0.829,
      "step": 64
    },
    {
      "epoch": 0.14215418261344998,
      "grad_norm": 0.07868961989879608,
      "learning_rate": 0.00013913043478260868,
      "loss": 0.6539,
      "step": 65
    },
    {
      "epoch": 0.1443411700382723,
      "grad_norm": 0.05902542173862457,
      "learning_rate": 0.00014130434782608694,
      "loss": 0.7102,
      "step": 66
    },
    {
      "epoch": 0.14652815746309458,
      "grad_norm": 0.056226056069135666,
      "learning_rate": 0.0001434782608695652,
      "loss": 0.4959,
      "step": 67
    },
    {
      "epoch": 0.1487151448879169,
      "grad_norm": 0.0737537145614624,
      "learning_rate": 0.00014565217391304347,
      "loss": 0.6453,
      "step": 68
    },
    {
      "epoch": 0.1509021323127392,
      "grad_norm": 0.06562914699316025,
      "learning_rate": 0.00014782608695652173,
      "loss": 0.5184,
      "step": 69
    },
    {
      "epoch": 0.1530891197375615,
      "grad_norm": 0.047399215400218964,
      "learning_rate": 0.00015,
      "loss": 0.7722,
      "step": 70
    },
    {
      "epoch": 0.15527610716238383,
      "grad_norm": 0.06085558980703354,
      "learning_rate": 0.00015217391304347825,
      "loss": 0.5718,
      "step": 71
    },
    {
      "epoch": 0.15746309458720611,
      "grad_norm": 0.08985075354576111,
      "learning_rate": 0.00015434782608695648,
      "loss": 0.7404,
      "step": 72
    },
    {
      "epoch": 0.15965008201202843,
      "grad_norm": 0.07962161302566528,
      "learning_rate": 0.00015652173913043477,
      "loss": 0.5717,
      "step": 73
    },
    {
      "epoch": 0.16183706943685075,
      "grad_norm": 0.10055044293403625,
      "learning_rate": 0.00015869565217391303,
      "loss": 0.7359,
      "step": 74
    },
    {
      "epoch": 0.16402405686167304,
      "grad_norm": 0.0740840807557106,
      "learning_rate": 0.00016086956521739132,
      "loss": 0.5564,
      "step": 75
    },
    {
      "epoch": 0.16621104428649536,
      "grad_norm": 0.10917298495769501,
      "learning_rate": 0.00016304347826086955,
      "loss": 1.0051,
      "step": 76
    },
    {
      "epoch": 0.16839803171131765,
      "grad_norm": 0.09235462546348572,
      "learning_rate": 0.0001652173913043478,
      "loss": 0.6685,
      "step": 77
    },
    {
      "epoch": 0.17058501913613996,
      "grad_norm": 0.09104294329881668,
      "learning_rate": 0.00016739130434782607,
      "loss": 0.7589,
      "step": 78
    },
    {
      "epoch": 0.17277200656096228,
      "grad_norm": 0.06644225120544434,
      "learning_rate": 0.00016956521739130433,
      "loss": 0.7341,
      "step": 79
    },
    {
      "epoch": 0.17495899398578457,
      "grad_norm": 0.1140175461769104,
      "learning_rate": 0.0001717391304347826,
      "loss": 0.4718,
      "step": 80
    },
    {
      "epoch": 0.1771459814106069,
      "grad_norm": 0.08415459096431732,
      "learning_rate": 0.00017391304347826085,
      "loss": 0.5123,
      "step": 81
    },
    {
      "epoch": 0.1793329688354292,
      "grad_norm": 0.09409467875957489,
      "learning_rate": 0.00017608695652173914,
      "loss": 0.8004,
      "step": 82
    },
    {
      "epoch": 0.1815199562602515,
      "grad_norm": 0.06406699866056442,
      "learning_rate": 0.00017826086956521738,
      "loss": 0.7602,
      "step": 83
    },
    {
      "epoch": 0.18370694368507381,
      "grad_norm": 0.08316787332296371,
      "learning_rate": 0.00018043478260869564,
      "loss": 0.5476,
      "step": 84
    },
    {
      "epoch": 0.1858939311098961,
      "grad_norm": 0.12399126589298248,
      "learning_rate": 0.0001826086956521739,
      "loss": 0.9202,
      "step": 85
    },
    {
      "epoch": 0.18808091853471842,
      "grad_norm": 0.055993784219026566,
      "learning_rate": 0.00018478260869565216,
      "loss": 0.7112,
      "step": 86
    },
    {
      "epoch": 0.19026790595954074,
      "grad_norm": 0.05300869792699814,
      "learning_rate": 0.00018695652173913042,
      "loss": 0.5599,
      "step": 87
    },
    {
      "epoch": 0.19245489338436303,
      "grad_norm": 0.0660649836063385,
      "learning_rate": 0.00018913043478260868,
      "loss": 0.6456,
      "step": 88
    },
    {
      "epoch": 0.19464188080918535,
      "grad_norm": 0.0759139358997345,
      "learning_rate": 0.0001913043478260869,
      "loss": 0.5772,
      "step": 89
    },
    {
      "epoch": 0.19682886823400766,
      "grad_norm": 0.08481380343437195,
      "learning_rate": 0.0001934782608695652,
      "loss": 0.5918,
      "step": 90
    },
    {
      "epoch": 0.19901585565882995,
      "grad_norm": 0.04496153071522713,
      "learning_rate": 0.00019565217391304346,
      "loss": 0.5967,
      "step": 91
    },
    {
      "epoch": 0.20120284308365227,
      "grad_norm": 0.06545284390449524,
      "learning_rate": 0.00019782608695652172,
      "loss": 0.7882,
      "step": 92
    },
    {
      "epoch": 0.2033898305084746,
      "grad_norm": 0.0938073992729187,
      "learning_rate": 0.00019999999999999998,
      "loss": 0.5598,
      "step": 93
    },
    {
      "epoch": 0.20557681793329688,
      "grad_norm": 0.08632558584213257,
      "learning_rate": 0.00020217391304347824,
      "loss": 0.687,
      "step": 94
    },
    {
      "epoch": 0.2077638053581192,
      "grad_norm": 0.053765781223773956,
      "learning_rate": 0.0002043478260869565,
      "loss": 0.6326,
      "step": 95
    },
    {
      "epoch": 0.2099507927829415,
      "grad_norm": 0.3586576282978058,
      "learning_rate": 0.00020652173913043474,
      "loss": 0.6559,
      "step": 96
    },
    {
      "epoch": 0.2121377802077638,
      "grad_norm": 0.10852506011724472,
      "learning_rate": 0.00020869565217391303,
      "loss": 0.5066,
      "step": 97
    },
    {
      "epoch": 0.21432476763258612,
      "grad_norm": 0.09538215398788452,
      "learning_rate": 0.00021086956521739129,
      "loss": 0.6338,
      "step": 98
    },
    {
      "epoch": 0.2165117550574084,
      "grad_norm": 0.078074149787426,
      "learning_rate": 0.00021304347826086955,
      "loss": 0.7748,
      "step": 99
    },
    {
      "epoch": 0.21869874248223073,
      "grad_norm": 0.07761886715888977,
      "learning_rate": 0.0002152173913043478,
      "loss": 0.6391,
      "step": 100
    },
    {
      "epoch": 0.22088572990705305,
      "grad_norm": 0.08739285171031952,
      "learning_rate": 0.00021739130434782607,
      "loss": 0.621,
      "step": 101
    },
    {
      "epoch": 0.22307271733187534,
      "grad_norm": 0.10503433644771576,
      "learning_rate": 0.00021956521739130433,
      "loss": 0.7355,
      "step": 102
    },
    {
      "epoch": 0.22525970475669765,
      "grad_norm": 0.08183517307043076,
      "learning_rate": 0.00022173913043478256,
      "loss": 0.9783,
      "step": 103
    },
    {
      "epoch": 0.22744669218151994,
      "grad_norm": 0.1349480152130127,
      "learning_rate": 0.00022391304347826085,
      "loss": 0.7544,
      "step": 104
    },
    {
      "epoch": 0.22963367960634226,
      "grad_norm": 0.07376862317323685,
      "learning_rate": 0.0002260869565217391,
      "loss": 0.5361,
      "step": 105
    },
    {
      "epoch": 0.23182066703116458,
      "grad_norm": 0.04610562324523926,
      "learning_rate": 0.0002282608695652174,
      "loss": 0.4957,
      "step": 106
    },
    {
      "epoch": 0.23400765445598687,
      "grad_norm": 0.045114483684301376,
      "learning_rate": 0.00023043478260869563,
      "loss": 0.6601,
      "step": 107
    },
    {
      "epoch": 0.2361946418808092,
      "grad_norm": 0.06456376612186432,
      "learning_rate": 0.0002326086956521739,
      "loss": 0.6509,
      "step": 108
    },
    {
      "epoch": 0.2383816293056315,
      "grad_norm": 0.10435660928487778,
      "learning_rate": 0.00023478260869565215,
      "loss": 0.7368,
      "step": 109
    },
    {
      "epoch": 0.2405686167304538,
      "grad_norm": 0.08688430488109589,
      "learning_rate": 0.00023695652173913041,
      "loss": 0.6227,
      "step": 110
    },
    {
      "epoch": 0.2427556041552761,
      "grad_norm": 0.060819629579782486,
      "learning_rate": 0.00023913043478260867,
      "loss": 0.7475,
      "step": 111
    },
    {
      "epoch": 0.2449425915800984,
      "grad_norm": 0.05878061801195145,
      "learning_rate": 0.00024130434782608694,
      "loss": 0.5565,
      "step": 112
    },
    {
      "epoch": 0.24712957900492072,
      "grad_norm": 0.07892751693725586,
      "learning_rate": 0.00024347826086956522,
      "loss": 0.6662,
      "step": 113
    },
    {
      "epoch": 0.24931656642974304,
      "grad_norm": 0.07645498961210251,
      "learning_rate": 0.00024565217391304343,
      "loss": 0.5861,
      "step": 114
    },
    {
      "epoch": 0.25150355385456535,
      "grad_norm": 0.07400215417146683,
      "learning_rate": 0.0002478260869565217,
      "loss": 0.701,
      "step": 115
    },
    {
      "epoch": 0.2536905412793876,
      "grad_norm": 0.06814702600240707,
      "learning_rate": 0.00025,
      "loss": 0.7172,
      "step": 116
    },
    {
      "epoch": 0.25587752870420993,
      "grad_norm": 0.14431889355182648,
      "learning_rate": 0.00025217391304347824,
      "loss": 0.6619,
      "step": 117
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 0.07565838098526001,
      "learning_rate": 0.00025434782608695647,
      "loss": 0.6557,
      "step": 118
    },
    {
      "epoch": 0.26025150355385457,
      "grad_norm": 0.20756904780864716,
      "learning_rate": 0.00025652173913043476,
      "loss": 0.754,
      "step": 119
    },
    {
      "epoch": 0.2624384909786769,
      "grad_norm": 0.05772192403674126,
      "learning_rate": 0.00025869565217391305,
      "loss": 0.7944,
      "step": 120
    },
    {
      "epoch": 0.2646254784034992,
      "grad_norm": 0.0794505923986435,
      "learning_rate": 0.0002608695652173913,
      "loss": 0.6919,
      "step": 121
    },
    {
      "epoch": 0.26681246582832147,
      "grad_norm": 0.13767485320568085,
      "learning_rate": 0.00026304347826086957,
      "loss": 0.6862,
      "step": 122
    },
    {
      "epoch": 0.2689994532531438,
      "grad_norm": 0.06422579288482666,
      "learning_rate": 0.0002652173913043478,
      "loss": 0.6693,
      "step": 123
    },
    {
      "epoch": 0.2711864406779661,
      "grad_norm": 0.35600975155830383,
      "learning_rate": 0.00026739130434782604,
      "loss": 0.7302,
      "step": 124
    },
    {
      "epoch": 0.2733734281027884,
      "grad_norm": 0.07260554283857346,
      "learning_rate": 0.0002695652173913043,
      "loss": 0.7832,
      "step": 125
    },
    {
      "epoch": 0.27556041552761074,
      "grad_norm": 0.06728994846343994,
      "learning_rate": 0.0002717391304347826,
      "loss": 0.7292,
      "step": 126
    },
    {
      "epoch": 0.277747402952433,
      "grad_norm": 0.06649418920278549,
      "learning_rate": 0.00027391304347826085,
      "loss": 0.5997,
      "step": 127
    },
    {
      "epoch": 0.2799343903772553,
      "grad_norm": 0.04817730188369751,
      "learning_rate": 0.0002760869565217391,
      "loss": 0.4626,
      "step": 128
    },
    {
      "epoch": 0.28212137780207763,
      "grad_norm": 0.08356320858001709,
      "learning_rate": 0.00027826086956521737,
      "loss": 0.6679,
      "step": 129
    },
    {
      "epoch": 0.28430836522689995,
      "grad_norm": 0.06807713955640793,
      "learning_rate": 0.00028043478260869565,
      "loss": 0.9006,
      "step": 130
    },
    {
      "epoch": 0.28649535265172227,
      "grad_norm": 0.04999319463968277,
      "learning_rate": 0.0002826086956521739,
      "loss": 0.5937,
      "step": 131
    },
    {
      "epoch": 0.2886823400765446,
      "grad_norm": 0.04334180802106857,
      "learning_rate": 0.0002847826086956521,
      "loss": 0.6044,
      "step": 132
    },
    {
      "epoch": 0.29086932750136685,
      "grad_norm": 0.06852805614471436,
      "learning_rate": 0.0002869565217391304,
      "loss": 0.683,
      "step": 133
    },
    {
      "epoch": 0.29305631492618917,
      "grad_norm": 0.12273514270782471,
      "learning_rate": 0.00028913043478260864,
      "loss": 0.665,
      "step": 134
    },
    {
      "epoch": 0.2952433023510115,
      "grad_norm": 0.10215126723051071,
      "learning_rate": 0.00029130434782608693,
      "loss": 0.6439,
      "step": 135
    },
    {
      "epoch": 0.2974302897758338,
      "grad_norm": 0.07202117145061493,
      "learning_rate": 0.0002934782608695652,
      "loss": 0.7748,
      "step": 136
    },
    {
      "epoch": 0.2996172772006561,
      "grad_norm": 0.14298571646213531,
      "learning_rate": 0.00029565217391304345,
      "loss": 0.712,
      "step": 137
    },
    {
      "epoch": 0.3018042646254784,
      "grad_norm": 0.046531159430742264,
      "learning_rate": 0.0002978260869565217,
      "loss": 0.6656,
      "step": 138
    },
    {
      "epoch": 0.3039912520503007,
      "grad_norm": 0.049203984439373016,
      "learning_rate": 0.0003,
      "loss": 0.7335,
      "step": 139
    },
    {
      "epoch": 0.306178239475123,
      "grad_norm": 0.0616229847073555,
      "learning_rate": 0.00029975728155339805,
      "loss": 0.7745,
      "step": 140
    },
    {
      "epoch": 0.30836522689994533,
      "grad_norm": 0.06660609692335129,
      "learning_rate": 0.0002995145631067961,
      "loss": 0.5531,
      "step": 141
    },
    {
      "epoch": 0.31055221432476765,
      "grad_norm": 0.037682708352804184,
      "learning_rate": 0.00029927184466019415,
      "loss": 0.6551,
      "step": 142
    },
    {
      "epoch": 0.3127392017495899,
      "grad_norm": 0.07913398742675781,
      "learning_rate": 0.00029902912621359223,
      "loss": 0.6406,
      "step": 143
    },
    {
      "epoch": 0.31492618917441223,
      "grad_norm": 0.09453226625919342,
      "learning_rate": 0.00029878640776699025,
      "loss": 0.7127,
      "step": 144
    },
    {
      "epoch": 0.31711317659923455,
      "grad_norm": 0.07528267800807953,
      "learning_rate": 0.00029854368932038833,
      "loss": 0.7101,
      "step": 145
    },
    {
      "epoch": 0.31930016402405687,
      "grad_norm": 0.07698322087526321,
      "learning_rate": 0.0002983009708737864,
      "loss": 0.658,
      "step": 146
    },
    {
      "epoch": 0.3214871514488792,
      "grad_norm": 0.05745327100157738,
      "learning_rate": 0.00029805825242718443,
      "loss": 0.7634,
      "step": 147
    },
    {
      "epoch": 0.3236741388737015,
      "grad_norm": 0.09420588612556458,
      "learning_rate": 0.0002978155339805825,
      "loss": 0.6379,
      "step": 148
    },
    {
      "epoch": 0.32586112629852376,
      "grad_norm": 0.09700261056423187,
      "learning_rate": 0.0002975728155339806,
      "loss": 0.5634,
      "step": 149
    },
    {
      "epoch": 0.3280481137233461,
      "grad_norm": 0.0776243656873703,
      "learning_rate": 0.0002973300970873786,
      "loss": 0.6162,
      "step": 150
    },
    {
      "epoch": 0.3302351011481684,
      "grad_norm": 0.02564432844519615,
      "learning_rate": 0.0002970873786407767,
      "loss": 0.4994,
      "step": 151
    },
    {
      "epoch": 0.3324220885729907,
      "grad_norm": 0.05717935040593147,
      "learning_rate": 0.00029684466019417477,
      "loss": 0.6405,
      "step": 152
    },
    {
      "epoch": 0.33460907599781303,
      "grad_norm": 0.05358930304646492,
      "learning_rate": 0.0002966019417475728,
      "loss": 0.5294,
      "step": 153
    },
    {
      "epoch": 0.3367960634226353,
      "grad_norm": 0.09334524720907211,
      "learning_rate": 0.00029635922330097087,
      "loss": 0.7158,
      "step": 154
    },
    {
      "epoch": 0.3389830508474576,
      "grad_norm": 0.1500510573387146,
      "learning_rate": 0.00029611650485436894,
      "loss": 0.6285,
      "step": 155
    },
    {
      "epoch": 0.34117003827227993,
      "grad_norm": 0.07156921923160553,
      "learning_rate": 0.00029587378640776697,
      "loss": 0.5904,
      "step": 156
    },
    {
      "epoch": 0.34335702569710225,
      "grad_norm": 0.05520716309547424,
      "learning_rate": 0.00029563106796116505,
      "loss": 0.6419,
      "step": 157
    },
    {
      "epoch": 0.34554401312192456,
      "grad_norm": 0.060137733817100525,
      "learning_rate": 0.00029538834951456307,
      "loss": 0.6339,
      "step": 158
    },
    {
      "epoch": 0.3477310005467469,
      "grad_norm": 0.04418874531984329,
      "learning_rate": 0.00029514563106796115,
      "loss": 0.6582,
      "step": 159
    },
    {
      "epoch": 0.34991798797156914,
      "grad_norm": 0.14283375442028046,
      "learning_rate": 0.0002949029126213592,
      "loss": 0.7756,
      "step": 160
    },
    {
      "epoch": 0.35210497539639146,
      "grad_norm": 0.12531793117523193,
      "learning_rate": 0.00029466019417475725,
      "loss": 0.6434,
      "step": 161
    },
    {
      "epoch": 0.3542919628212138,
      "grad_norm": 0.06906326115131378,
      "learning_rate": 0.0002944174757281553,
      "loss": 0.6828,
      "step": 162
    },
    {
      "epoch": 0.3564789502460361,
      "grad_norm": 0.10274400562047958,
      "learning_rate": 0.00029417475728155335,
      "loss": 0.7356,
      "step": 163
    },
    {
      "epoch": 0.3586659376708584,
      "grad_norm": 0.07805956155061722,
      "learning_rate": 0.0002939320388349514,
      "loss": 0.7114,
      "step": 164
    },
    {
      "epoch": 0.3608529250956807,
      "grad_norm": 0.059253111481666565,
      "learning_rate": 0.0002936893203883495,
      "loss": 0.6082,
      "step": 165
    },
    {
      "epoch": 0.363039912520503,
      "grad_norm": 0.07108813524246216,
      "learning_rate": 0.00029344660194174753,
      "loss": 0.7756,
      "step": 166
    },
    {
      "epoch": 0.3652268999453253,
      "grad_norm": 0.09952510893344879,
      "learning_rate": 0.0002932038834951456,
      "loss": 0.6716,
      "step": 167
    },
    {
      "epoch": 0.36741388737014763,
      "grad_norm": 0.07966083288192749,
      "learning_rate": 0.0002929611650485437,
      "loss": 0.6009,
      "step": 168
    },
    {
      "epoch": 0.36960087479496995,
      "grad_norm": 0.0691203773021698,
      "learning_rate": 0.0002927184466019417,
      "loss": 0.5261,
      "step": 169
    },
    {
      "epoch": 0.3717878622197922,
      "grad_norm": 0.08406491577625275,
      "learning_rate": 0.0002924757281553398,
      "loss": 0.6501,
      "step": 170
    },
    {
      "epoch": 0.3739748496446145,
      "grad_norm": 0.08572190254926682,
      "learning_rate": 0.0002922330097087378,
      "loss": 0.7071,
      "step": 171
    },
    {
      "epoch": 0.37616183706943684,
      "grad_norm": 0.07649155706167221,
      "learning_rate": 0.0002919902912621359,
      "loss": 0.6682,
      "step": 172
    },
    {
      "epoch": 0.37834882449425916,
      "grad_norm": 0.06834893673658371,
      "learning_rate": 0.00029174757281553396,
      "loss": 0.6535,
      "step": 173
    },
    {
      "epoch": 0.3805358119190815,
      "grad_norm": 0.148598775267601,
      "learning_rate": 0.000291504854368932,
      "loss": 0.6417,
      "step": 174
    },
    {
      "epoch": 0.3827227993439038,
      "grad_norm": 0.09686613082885742,
      "learning_rate": 0.00029126213592233006,
      "loss": 0.6217,
      "step": 175
    },
    {
      "epoch": 0.38490978676872606,
      "grad_norm": 0.047472141683101654,
      "learning_rate": 0.00029101941747572814,
      "loss": 0.39,
      "step": 176
    },
    {
      "epoch": 0.3870967741935484,
      "grad_norm": 0.06221065670251846,
      "learning_rate": 0.00029077669902912616,
      "loss": 0.7205,
      "step": 177
    },
    {
      "epoch": 0.3892837616183707,
      "grad_norm": 0.061044104397296906,
      "learning_rate": 0.00029053398058252424,
      "loss": 0.6231,
      "step": 178
    },
    {
      "epoch": 0.391470749043193,
      "grad_norm": 0.23658694326877594,
      "learning_rate": 0.0002902912621359223,
      "loss": 0.7941,
      "step": 179
    },
    {
      "epoch": 0.39365773646801533,
      "grad_norm": 0.09449535608291626,
      "learning_rate": 0.00029004854368932034,
      "loss": 0.7093,
      "step": 180
    },
    {
      "epoch": 0.3958447238928376,
      "grad_norm": 0.07445142418146133,
      "learning_rate": 0.0002898058252427184,
      "loss": 0.5534,
      "step": 181
    },
    {
      "epoch": 0.3980317113176599,
      "grad_norm": 0.0452071838080883,
      "learning_rate": 0.0002895631067961165,
      "loss": 0.6028,
      "step": 182
    },
    {
      "epoch": 0.4002186987424822,
      "grad_norm": 0.0683695524930954,
      "learning_rate": 0.0002893203883495145,
      "loss": 0.3688,
      "step": 183
    },
    {
      "epoch": 0.40240568616730454,
      "grad_norm": 0.048305071890354156,
      "learning_rate": 0.0002890776699029126,
      "loss": 0.8294,
      "step": 184
    },
    {
      "epoch": 0.40459267359212686,
      "grad_norm": 0.08140576630830765,
      "learning_rate": 0.0002888349514563107,
      "loss": 0.7538,
      "step": 185
    },
    {
      "epoch": 0.4067796610169492,
      "grad_norm": 0.13340412080287933,
      "learning_rate": 0.0002885922330097087,
      "loss": 0.7636,
      "step": 186
    },
    {
      "epoch": 0.40896664844177144,
      "grad_norm": 0.04563393443822861,
      "learning_rate": 0.0002883495145631068,
      "loss": 0.6064,
      "step": 187
    },
    {
      "epoch": 0.41115363586659376,
      "grad_norm": 0.08804814517498016,
      "learning_rate": 0.00028810679611650486,
      "loss": 0.8036,
      "step": 188
    },
    {
      "epoch": 0.4133406232914161,
      "grad_norm": 0.08191797137260437,
      "learning_rate": 0.0002878640776699029,
      "loss": 0.7082,
      "step": 189
    },
    {
      "epoch": 0.4155276107162384,
      "grad_norm": 0.09029440581798553,
      "learning_rate": 0.00028762135922330096,
      "loss": 0.6588,
      "step": 190
    },
    {
      "epoch": 0.4177145981410607,
      "grad_norm": 0.09573402255773544,
      "learning_rate": 0.000287378640776699,
      "loss": 0.7273,
      "step": 191
    },
    {
      "epoch": 0.419901585565883,
      "grad_norm": 0.045617956668138504,
      "learning_rate": 0.00028713592233009706,
      "loss": 0.461,
      "step": 192
    },
    {
      "epoch": 0.4220885729907053,
      "grad_norm": 0.05199209973216057,
      "learning_rate": 0.00028689320388349513,
      "loss": 0.6877,
      "step": 193
    },
    {
      "epoch": 0.4242755604155276,
      "grad_norm": 0.05869835987687111,
      "learning_rate": 0.00028665048543689316,
      "loss": 0.4181,
      "step": 194
    },
    {
      "epoch": 0.4264625478403499,
      "grad_norm": 0.14079192280769348,
      "learning_rate": 0.00028640776699029124,
      "loss": 0.6978,
      "step": 195
    },
    {
      "epoch": 0.42864953526517224,
      "grad_norm": 0.09773832559585571,
      "learning_rate": 0.0002861650485436893,
      "loss": 0.6019,
      "step": 196
    },
    {
      "epoch": 0.4308365226899945,
      "grad_norm": 0.04800047725439072,
      "learning_rate": 0.00028592233009708734,
      "loss": 0.613,
      "step": 197
    },
    {
      "epoch": 0.4330235101148168,
      "grad_norm": 0.056679654866456985,
      "learning_rate": 0.0002856796116504854,
      "loss": 0.5245,
      "step": 198
    },
    {
      "epoch": 0.43521049753963914,
      "grad_norm": 0.04190812632441521,
      "learning_rate": 0.0002854368932038835,
      "loss": 0.9215,
      "step": 199
    },
    {
      "epoch": 0.43739748496446146,
      "grad_norm": 0.11672720313072205,
      "learning_rate": 0.0002851941747572815,
      "loss": 0.8228,
      "step": 200
    },
    {
      "epoch": 0.4395844723892838,
      "grad_norm": 0.05854491889476776,
      "learning_rate": 0.0002849514563106796,
      "loss": 0.7991,
      "step": 201
    },
    {
      "epoch": 0.4417714598141061,
      "grad_norm": 0.06320830434560776,
      "learning_rate": 0.00028470873786407767,
      "loss": 0.6015,
      "step": 202
    },
    {
      "epoch": 0.44395844723892836,
      "grad_norm": 0.09820743650197983,
      "learning_rate": 0.0002844660194174757,
      "loss": 0.5632,
      "step": 203
    },
    {
      "epoch": 0.4461454346637507,
      "grad_norm": 0.10225464403629303,
      "learning_rate": 0.00028422330097087377,
      "loss": 0.8571,
      "step": 204
    },
    {
      "epoch": 0.448332422088573,
      "grad_norm": 0.0544356070458889,
      "learning_rate": 0.00028398058252427185,
      "loss": 0.804,
      "step": 205
    },
    {
      "epoch": 0.4505194095133953,
      "grad_norm": 0.10255508124828339,
      "learning_rate": 0.0002837378640776699,
      "loss": 0.6108,
      "step": 206
    },
    {
      "epoch": 0.4527063969382176,
      "grad_norm": 0.06616372615098953,
      "learning_rate": 0.00028349514563106795,
      "loss": 0.6481,
      "step": 207
    },
    {
      "epoch": 0.4548933843630399,
      "grad_norm": 0.0479828417301178,
      "learning_rate": 0.00028325242718446603,
      "loss": 0.6919,
      "step": 208
    },
    {
      "epoch": 0.4570803717878622,
      "grad_norm": 0.06924855709075928,
      "learning_rate": 0.00028300970873786405,
      "loss": 0.705,
      "step": 209
    },
    {
      "epoch": 0.4592673592126845,
      "grad_norm": 0.03697100654244423,
      "learning_rate": 0.00028276699029126213,
      "loss": 0.7116,
      "step": 210
    },
    {
      "epoch": 0.46145434663750684,
      "grad_norm": 0.09638485312461853,
      "learning_rate": 0.00028252427184466015,
      "loss": 0.7317,
      "step": 211
    },
    {
      "epoch": 0.46364133406232916,
      "grad_norm": 0.06610197573900223,
      "learning_rate": 0.00028228155339805823,
      "loss": 0.6847,
      "step": 212
    },
    {
      "epoch": 0.4658283214871515,
      "grad_norm": 0.13233628869056702,
      "learning_rate": 0.0002820388349514563,
      "loss": 0.6703,
      "step": 213
    },
    {
      "epoch": 0.46801530891197374,
      "grad_norm": 0.1086718887090683,
      "learning_rate": 0.00028179611650485433,
      "loss": 0.7942,
      "step": 214
    },
    {
      "epoch": 0.47020229633679606,
      "grad_norm": 0.06878714263439178,
      "learning_rate": 0.0002815533980582524,
      "loss": 0.5508,
      "step": 215
    },
    {
      "epoch": 0.4723892837616184,
      "grad_norm": 0.08610357344150543,
      "learning_rate": 0.00028131067961165043,
      "loss": 0.6846,
      "step": 216
    },
    {
      "epoch": 0.4745762711864407,
      "grad_norm": 0.11769182235002518,
      "learning_rate": 0.0002810679611650485,
      "loss": 0.686,
      "step": 217
    },
    {
      "epoch": 0.476763258611263,
      "grad_norm": 0.060380857437849045,
      "learning_rate": 0.0002808252427184466,
      "loss": 0.627,
      "step": 218
    },
    {
      "epoch": 0.47895024603608527,
      "grad_norm": 0.07884469628334045,
      "learning_rate": 0.0002805825242718446,
      "loss": 0.6379,
      "step": 219
    },
    {
      "epoch": 0.4811372334609076,
      "grad_norm": 0.0553802065551281,
      "learning_rate": 0.0002803398058252427,
      "loss": 1.234,
      "step": 220
    },
    {
      "epoch": 0.4833242208857299,
      "grad_norm": 0.053239986300468445,
      "learning_rate": 0.0002800970873786407,
      "loss": 0.6463,
      "step": 221
    },
    {
      "epoch": 0.4855112083105522,
      "grad_norm": 0.09387147426605225,
      "learning_rate": 0.0002798543689320388,
      "loss": 0.8651,
      "step": 222
    },
    {
      "epoch": 0.48769819573537454,
      "grad_norm": 0.08428589254617691,
      "learning_rate": 0.00027961165048543687,
      "loss": 0.4774,
      "step": 223
    },
    {
      "epoch": 0.4898851831601968,
      "grad_norm": 0.09616990387439728,
      "learning_rate": 0.0002793689320388349,
      "loss": 0.5953,
      "step": 224
    },
    {
      "epoch": 0.4920721705850191,
      "grad_norm": 0.12636440992355347,
      "learning_rate": 0.00027912621359223297,
      "loss": 0.5935,
      "step": 225
    },
    {
      "epoch": 0.49425915800984144,
      "grad_norm": 0.04346989095211029,
      "learning_rate": 0.00027888349514563105,
      "loss": 0.6199,
      "step": 226
    },
    {
      "epoch": 0.49644614543466375,
      "grad_norm": 0.07197440415620804,
      "learning_rate": 0.00027864077669902907,
      "loss": 0.6705,
      "step": 227
    },
    {
      "epoch": 0.4986331328594861,
      "grad_norm": 0.08746636658906937,
      "learning_rate": 0.00027839805825242715,
      "loss": 0.8351,
      "step": 228
    },
    {
      "epoch": 0.5008201202843083,
      "grad_norm": 0.1399545520544052,
      "learning_rate": 0.0002781553398058252,
      "loss": 0.7487,
      "step": 229
    },
    {
      "epoch": 0.5030071077091307,
      "grad_norm": 0.0729706659913063,
      "learning_rate": 0.00027791262135922325,
      "loss": 0.7952,
      "step": 230
    },
    {
      "epoch": 0.505194095133953,
      "grad_norm": 0.059242866933345795,
      "learning_rate": 0.0002776699029126213,
      "loss": 0.6364,
      "step": 231
    },
    {
      "epoch": 0.5073810825587752,
      "grad_norm": 0.09419375658035278,
      "learning_rate": 0.0002774271844660194,
      "loss": 0.8293,
      "step": 232
    },
    {
      "epoch": 0.5095680699835976,
      "grad_norm": 0.06085381284356117,
      "learning_rate": 0.0002771844660194174,
      "loss": 0.8756,
      "step": 233
    },
    {
      "epoch": 0.5117550574084199,
      "grad_norm": 0.05615559592843056,
      "learning_rate": 0.0002769417475728155,
      "loss": 0.618,
      "step": 234
    },
    {
      "epoch": 0.5139420448332422,
      "grad_norm": 0.07931798696517944,
      "learning_rate": 0.0002766990291262136,
      "loss": 0.7047,
      "step": 235
    },
    {
      "epoch": 0.5161290322580645,
      "grad_norm": 0.046281419694423676,
      "learning_rate": 0.0002764563106796116,
      "loss": 0.4596,
      "step": 236
    },
    {
      "epoch": 0.5183160196828869,
      "grad_norm": 0.04308830946683884,
      "learning_rate": 0.0002762135922330097,
      "loss": 0.6457,
      "step": 237
    },
    {
      "epoch": 0.5205030071077091,
      "grad_norm": 0.0868159681558609,
      "learning_rate": 0.00027597087378640776,
      "loss": 0.7541,
      "step": 238
    },
    {
      "epoch": 0.5226899945325314,
      "grad_norm": 0.05126407369971275,
      "learning_rate": 0.0002757281553398058,
      "loss": 0.6661,
      "step": 239
    },
    {
      "epoch": 0.5248769819573538,
      "grad_norm": 0.0836605578660965,
      "learning_rate": 0.00027548543689320386,
      "loss": 0.7741,
      "step": 240
    },
    {
      "epoch": 0.527063969382176,
      "grad_norm": 0.10537996888160706,
      "learning_rate": 0.00027524271844660194,
      "loss": 0.5839,
      "step": 241
    },
    {
      "epoch": 0.5292509568069984,
      "grad_norm": 0.06915068626403809,
      "learning_rate": 0.00027499999999999996,
      "loss": 0.7563,
      "step": 242
    },
    {
      "epoch": 0.5314379442318207,
      "grad_norm": 0.05846241116523743,
      "learning_rate": 0.00027475728155339804,
      "loss": 0.6107,
      "step": 243
    },
    {
      "epoch": 0.5336249316566429,
      "grad_norm": 0.07253163307905197,
      "learning_rate": 0.0002745145631067961,
      "loss": 0.6413,
      "step": 244
    },
    {
      "epoch": 0.5358119190814653,
      "grad_norm": 0.0637165755033493,
      "learning_rate": 0.00027427184466019414,
      "loss": 0.6918,
      "step": 245
    },
    {
      "epoch": 0.5379989065062876,
      "grad_norm": 0.04841014742851257,
      "learning_rate": 0.0002740291262135922,
      "loss": 0.4336,
      "step": 246
    },
    {
      "epoch": 0.5401858939311099,
      "grad_norm": 0.07899260520935059,
      "learning_rate": 0.0002737864077669903,
      "loss": 1.1465,
      "step": 247
    },
    {
      "epoch": 0.5423728813559322,
      "grad_norm": 0.128900945186615,
      "learning_rate": 0.0002735436893203883,
      "loss": 0.7688,
      "step": 248
    },
    {
      "epoch": 0.5445598687807545,
      "grad_norm": 0.09204466640949249,
      "learning_rate": 0.0002733009708737864,
      "loss": 0.7298,
      "step": 249
    },
    {
      "epoch": 0.5467468562055768,
      "grad_norm": 0.08942551165819168,
      "learning_rate": 0.0002730582524271845,
      "loss": 0.6038,
      "step": 250
    },
    {
      "epoch": 0.5489338436303991,
      "grad_norm": 0.11410896480083466,
      "learning_rate": 0.0002728155339805825,
      "loss": 0.6325,
      "step": 251
    },
    {
      "epoch": 0.5511208310552215,
      "grad_norm": 0.04977932199835777,
      "learning_rate": 0.0002725728155339806,
      "loss": 0.6421,
      "step": 252
    },
    {
      "epoch": 0.5533078184800437,
      "grad_norm": 0.05651041865348816,
      "learning_rate": 0.00027233009708737865,
      "loss": 0.5886,
      "step": 253
    },
    {
      "epoch": 0.555494805904866,
      "grad_norm": 0.056123483926057816,
      "learning_rate": 0.0002720873786407767,
      "loss": 0.5284,
      "step": 254
    },
    {
      "epoch": 0.5576817933296884,
      "grad_norm": 0.05689515545964241,
      "learning_rate": 0.00027184466019417475,
      "loss": 0.6985,
      "step": 255
    },
    {
      "epoch": 0.5598687807545106,
      "grad_norm": 0.05087985843420029,
      "learning_rate": 0.0002716019417475728,
      "loss": 0.5622,
      "step": 256
    },
    {
      "epoch": 0.562055768179333,
      "grad_norm": 0.03652990981936455,
      "learning_rate": 0.00027135922330097086,
      "loss": 0.3838,
      "step": 257
    },
    {
      "epoch": 0.5642427556041553,
      "grad_norm": 0.20960339903831482,
      "learning_rate": 0.00027111650485436893,
      "loss": 0.7656,
      "step": 258
    },
    {
      "epoch": 0.5664297430289775,
      "grad_norm": 0.05671091377735138,
      "learning_rate": 0.00027087378640776696,
      "loss": 0.6568,
      "step": 259
    },
    {
      "epoch": 0.5686167304537999,
      "grad_norm": 0.07829181104898453,
      "learning_rate": 0.00027063106796116503,
      "loss": 0.6861,
      "step": 260
    },
    {
      "epoch": 0.5708037178786222,
      "grad_norm": 0.10538700968027115,
      "learning_rate": 0.0002703883495145631,
      "loss": 0.911,
      "step": 261
    },
    {
      "epoch": 0.5729907053034445,
      "grad_norm": 0.03851666674017906,
      "learning_rate": 0.00027014563106796114,
      "loss": 0.5833,
      "step": 262
    },
    {
      "epoch": 0.5751776927282668,
      "grad_norm": 0.09947405755519867,
      "learning_rate": 0.0002699029126213592,
      "loss": 0.6924,
      "step": 263
    },
    {
      "epoch": 0.5773646801530892,
      "grad_norm": 0.05998876690864563,
      "learning_rate": 0.00026966019417475724,
      "loss": 0.6582,
      "step": 264
    },
    {
      "epoch": 0.5795516675779114,
      "grad_norm": 0.05930376797914505,
      "learning_rate": 0.0002694174757281553,
      "loss": 0.6135,
      "step": 265
    },
    {
      "epoch": 0.5817386550027337,
      "grad_norm": 0.06882154196500778,
      "learning_rate": 0.0002691747572815534,
      "loss": 0.6843,
      "step": 266
    },
    {
      "epoch": 0.5839256424275561,
      "grad_norm": 0.05111825838685036,
      "learning_rate": 0.0002689320388349514,
      "loss": 0.8738,
      "step": 267
    },
    {
      "epoch": 0.5861126298523783,
      "grad_norm": 0.03514356166124344,
      "learning_rate": 0.0002686893203883495,
      "loss": 0.5535,
      "step": 268
    },
    {
      "epoch": 0.5882996172772007,
      "grad_norm": 0.04661941900849342,
      "learning_rate": 0.0002684466019417475,
      "loss": 0.5437,
      "step": 269
    },
    {
      "epoch": 0.590486604702023,
      "grad_norm": 0.11490173637866974,
      "learning_rate": 0.0002682038834951456,
      "loss": 0.7309,
      "step": 270
    },
    {
      "epoch": 0.5926735921268452,
      "grad_norm": 0.06036648526787758,
      "learning_rate": 0.00026796116504854367,
      "loss": 0.7171,
      "step": 271
    },
    {
      "epoch": 0.5948605795516676,
      "grad_norm": 0.07145757973194122,
      "learning_rate": 0.0002677184466019417,
      "loss": 0.8972,
      "step": 272
    },
    {
      "epoch": 0.5970475669764899,
      "grad_norm": 0.03046463429927826,
      "learning_rate": 0.00026747572815533977,
      "loss": 0.6225,
      "step": 273
    },
    {
      "epoch": 0.5992345544013122,
      "grad_norm": 0.058935679495334625,
      "learning_rate": 0.00026723300970873785,
      "loss": 0.7951,
      "step": 274
    },
    {
      "epoch": 0.6014215418261345,
      "grad_norm": 0.08104430884122849,
      "learning_rate": 0.0002669902912621359,
      "loss": 0.7906,
      "step": 275
    },
    {
      "epoch": 0.6036085292509568,
      "grad_norm": 0.037242889404296875,
      "learning_rate": 0.00026674757281553395,
      "loss": 0.4116,
      "step": 276
    },
    {
      "epoch": 0.6057955166757791,
      "grad_norm": 0.06395541876554489,
      "learning_rate": 0.00026650485436893203,
      "loss": 0.5387,
      "step": 277
    },
    {
      "epoch": 0.6079825041006014,
      "grad_norm": 0.09105798602104187,
      "learning_rate": 0.00026626213592233005,
      "loss": 0.76,
      "step": 278
    },
    {
      "epoch": 0.6101694915254238,
      "grad_norm": 0.0888744667172432,
      "learning_rate": 0.00026601941747572813,
      "loss": 0.5783,
      "step": 279
    },
    {
      "epoch": 0.612356478950246,
      "grad_norm": 0.0881604477763176,
      "learning_rate": 0.0002657766990291262,
      "loss": 0.6175,
      "step": 280
    },
    {
      "epoch": 0.6145434663750683,
      "grad_norm": 0.0212717242538929,
      "learning_rate": 0.00026553398058252423,
      "loss": 0.6384,
      "step": 281
    },
    {
      "epoch": 0.6167304537998907,
      "grad_norm": 0.08296003937721252,
      "learning_rate": 0.0002652912621359223,
      "loss": 0.7919,
      "step": 282
    },
    {
      "epoch": 0.6189174412247129,
      "grad_norm": 0.05302100628614426,
      "learning_rate": 0.0002650485436893204,
      "loss": 0.7996,
      "step": 283
    },
    {
      "epoch": 0.6211044286495353,
      "grad_norm": 0.05207206308841705,
      "learning_rate": 0.0002648058252427184,
      "loss": 0.5176,
      "step": 284
    },
    {
      "epoch": 0.6232914160743576,
      "grad_norm": 0.08915773034095764,
      "learning_rate": 0.0002645631067961165,
      "loss": 0.7059,
      "step": 285
    },
    {
      "epoch": 0.6254784034991798,
      "grad_norm": 0.12500520050525665,
      "learning_rate": 0.00026432038834951456,
      "loss": 0.8228,
      "step": 286
    },
    {
      "epoch": 0.6276653909240022,
      "grad_norm": 0.04379987716674805,
      "learning_rate": 0.0002640776699029126,
      "loss": 0.5246,
      "step": 287
    },
    {
      "epoch": 0.6298523783488245,
      "grad_norm": 0.10752689838409424,
      "learning_rate": 0.00026383495145631067,
      "loss": 0.7095,
      "step": 288
    },
    {
      "epoch": 0.6320393657736468,
      "grad_norm": 0.12701405584812164,
      "learning_rate": 0.00026359223300970874,
      "loss": 0.7316,
      "step": 289
    },
    {
      "epoch": 0.6342263531984691,
      "grad_norm": 0.11033506691455841,
      "learning_rate": 0.00026334951456310677,
      "loss": 0.7736,
      "step": 290
    },
    {
      "epoch": 0.6364133406232915,
      "grad_norm": 0.04730546101927757,
      "learning_rate": 0.00026310679611650484,
      "loss": 0.5268,
      "step": 291
    },
    {
      "epoch": 0.6386003280481137,
      "grad_norm": 0.10472558438777924,
      "learning_rate": 0.0002628640776699029,
      "loss": 0.6933,
      "step": 292
    },
    {
      "epoch": 0.640787315472936,
      "grad_norm": 0.056678298860788345,
      "learning_rate": 0.00026262135922330094,
      "loss": 0.4571,
      "step": 293
    },
    {
      "epoch": 0.6429743028977584,
      "grad_norm": 0.07412777096033096,
      "learning_rate": 0.000262378640776699,
      "loss": 0.6422,
      "step": 294
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 0.08613819628953934,
      "learning_rate": 0.00026213592233009705,
      "loss": 0.7245,
      "step": 295
    },
    {
      "epoch": 0.647348277747403,
      "grad_norm": 0.0964793786406517,
      "learning_rate": 0.0002618932038834951,
      "loss": 0.7397,
      "step": 296
    },
    {
      "epoch": 0.6495352651722253,
      "grad_norm": 0.08231832832098007,
      "learning_rate": 0.0002616504854368932,
      "loss": 0.4,
      "step": 297
    },
    {
      "epoch": 0.6517222525970475,
      "grad_norm": 0.09483752399682999,
      "learning_rate": 0.0002614077669902912,
      "loss": 0.7464,
      "step": 298
    },
    {
      "epoch": 0.6539092400218699,
      "grad_norm": 0.041885074228048325,
      "learning_rate": 0.0002611650485436893,
      "loss": 0.4597,
      "step": 299
    },
    {
      "epoch": 0.6560962274466922,
      "grad_norm": 0.07590518891811371,
      "learning_rate": 0.0002609223300970874,
      "loss": 0.9338,
      "step": 300
    },
    {
      "epoch": 0.6582832148715145,
      "grad_norm": 0.06022942438721657,
      "learning_rate": 0.0002606796116504854,
      "loss": 0.5732,
      "step": 301
    },
    {
      "epoch": 0.6604702022963368,
      "grad_norm": 0.0570467971265316,
      "learning_rate": 0.0002604368932038835,
      "loss": 0.5997,
      "step": 302
    },
    {
      "epoch": 0.6626571897211591,
      "grad_norm": 0.028766008093953133,
      "learning_rate": 0.00026019417475728156,
      "loss": 0.7863,
      "step": 303
    },
    {
      "epoch": 0.6648441771459814,
      "grad_norm": 0.03608011081814766,
      "learning_rate": 0.0002599514563106796,
      "loss": 0.703,
      "step": 304
    },
    {
      "epoch": 0.6670311645708037,
      "grad_norm": 0.07903197407722473,
      "learning_rate": 0.00025970873786407766,
      "loss": 0.6879,
      "step": 305
    },
    {
      "epoch": 0.6692181519956261,
      "grad_norm": 0.04391651973128319,
      "learning_rate": 0.00025946601941747574,
      "loss": 0.6,
      "step": 306
    },
    {
      "epoch": 0.6714051394204483,
      "grad_norm": 0.0842578336596489,
      "learning_rate": 0.00025922330097087376,
      "loss": 0.7932,
      "step": 307
    },
    {
      "epoch": 0.6735921268452706,
      "grad_norm": 0.061180584132671356,
      "learning_rate": 0.00025898058252427184,
      "loss": 0.8001,
      "step": 308
    },
    {
      "epoch": 0.675779114270093,
      "grad_norm": 0.08094706386327744,
      "learning_rate": 0.00025873786407766986,
      "loss": 0.604,
      "step": 309
    },
    {
      "epoch": 0.6779661016949152,
      "grad_norm": 0.05069729685783386,
      "learning_rate": 0.00025849514563106794,
      "loss": 0.5355,
      "step": 310
    },
    {
      "epoch": 0.6801530891197376,
      "grad_norm": 0.0906616821885109,
      "learning_rate": 0.000258252427184466,
      "loss": 0.5386,
      "step": 311
    },
    {
      "epoch": 0.6823400765445599,
      "grad_norm": 0.02921673282980919,
      "learning_rate": 0.00025800970873786404,
      "loss": 0.4077,
      "step": 312
    },
    {
      "epoch": 0.6845270639693821,
      "grad_norm": 0.14991872012615204,
      "learning_rate": 0.0002577669902912621,
      "loss": 0.7154,
      "step": 313
    },
    {
      "epoch": 0.6867140513942045,
      "grad_norm": 0.08407143503427505,
      "learning_rate": 0.00025752427184466014,
      "loss": 0.7446,
      "step": 314
    },
    {
      "epoch": 0.6889010388190268,
      "grad_norm": 0.0366005077958107,
      "learning_rate": 0.0002572815533980582,
      "loss": 0.6288,
      "step": 315
    },
    {
      "epoch": 0.6910880262438491,
      "grad_norm": 0.07749815285205841,
      "learning_rate": 0.0002570388349514563,
      "loss": 0.7508,
      "step": 316
    },
    {
      "epoch": 0.6932750136686714,
      "grad_norm": 0.080388642847538,
      "learning_rate": 0.0002567961165048543,
      "loss": 0.6253,
      "step": 317
    },
    {
      "epoch": 0.6954620010934938,
      "grad_norm": 0.07014317810535431,
      "learning_rate": 0.0002565533980582524,
      "loss": 0.6952,
      "step": 318
    },
    {
      "epoch": 0.697648988518316,
      "grad_norm": 0.0520801767706871,
      "learning_rate": 0.0002563106796116505,
      "loss": 0.4726,
      "step": 319
    },
    {
      "epoch": 0.6998359759431383,
      "grad_norm": 0.07832513004541397,
      "learning_rate": 0.0002560679611650485,
      "loss": 0.7236,
      "step": 320
    },
    {
      "epoch": 0.7020229633679607,
      "grad_norm": 0.14095266163349152,
      "learning_rate": 0.0002558252427184466,
      "loss": 0.8119,
      "step": 321
    },
    {
      "epoch": 0.7042099507927829,
      "grad_norm": 0.06780201941728592,
      "learning_rate": 0.0002555825242718446,
      "loss": 0.5814,
      "step": 322
    },
    {
      "epoch": 0.7063969382176053,
      "grad_norm": 0.11610683053731918,
      "learning_rate": 0.0002553398058252427,
      "loss": 0.679,
      "step": 323
    },
    {
      "epoch": 0.7085839256424276,
      "grad_norm": 0.10824164003133774,
      "learning_rate": 0.00025509708737864075,
      "loss": 0.6732,
      "step": 324
    },
    {
      "epoch": 0.7107709130672498,
      "grad_norm": 0.02975596860051155,
      "learning_rate": 0.0002548543689320388,
      "loss": 0.5066,
      "step": 325
    },
    {
      "epoch": 0.7129579004920722,
      "grad_norm": 0.060648877173662186,
      "learning_rate": 0.00025461165048543686,
      "loss": 0.6598,
      "step": 326
    },
    {
      "epoch": 0.7151448879168945,
      "grad_norm": 0.06371823698282242,
      "learning_rate": 0.00025436893203883493,
      "loss": 0.8467,
      "step": 327
    },
    {
      "epoch": 0.7173318753417168,
      "grad_norm": 0.585040271282196,
      "learning_rate": 0.00025412621359223296,
      "loss": 0.6203,
      "step": 328
    },
    {
      "epoch": 0.7195188627665391,
      "grad_norm": 0.094622403383255,
      "learning_rate": 0.00025388349514563103,
      "loss": 0.7434,
      "step": 329
    },
    {
      "epoch": 0.7217058501913614,
      "grad_norm": 0.11974699050188065,
      "learning_rate": 0.0002536407766990291,
      "loss": 0.9787,
      "step": 330
    },
    {
      "epoch": 0.7238928376161837,
      "grad_norm": 0.06584473699331284,
      "learning_rate": 0.00025339805825242714,
      "loss": 0.5138,
      "step": 331
    },
    {
      "epoch": 0.726079825041006,
      "grad_norm": 0.0335596539080143,
      "learning_rate": 0.0002531553398058252,
      "loss": 0.5124,
      "step": 332
    },
    {
      "epoch": 0.7282668124658284,
      "grad_norm": 0.08706491440534592,
      "learning_rate": 0.0002529126213592233,
      "loss": 0.7463,
      "step": 333
    },
    {
      "epoch": 0.7304537998906506,
      "grad_norm": 0.06929214298725128,
      "learning_rate": 0.0002526699029126213,
      "loss": 0.4162,
      "step": 334
    },
    {
      "epoch": 0.7326407873154729,
      "grad_norm": 0.08074366301298141,
      "learning_rate": 0.0002524271844660194,
      "loss": 0.4847,
      "step": 335
    },
    {
      "epoch": 0.7348277747402953,
      "grad_norm": 0.04467281326651573,
      "learning_rate": 0.00025218446601941747,
      "loss": 0.8289,
      "step": 336
    },
    {
      "epoch": 0.7370147621651175,
      "grad_norm": 0.0674748346209526,
      "learning_rate": 0.0002519417475728155,
      "loss": 0.7378,
      "step": 337
    },
    {
      "epoch": 0.7392017495899399,
      "grad_norm": 0.05383702740073204,
      "learning_rate": 0.00025169902912621357,
      "loss": 0.6848,
      "step": 338
    },
    {
      "epoch": 0.7413887370147622,
      "grad_norm": 0.08543651551008224,
      "learning_rate": 0.00025145631067961165,
      "loss": 0.8121,
      "step": 339
    },
    {
      "epoch": 0.7435757244395844,
      "grad_norm": 0.06835243105888367,
      "learning_rate": 0.00025121359223300967,
      "loss": 0.7358,
      "step": 340
    },
    {
      "epoch": 0.7457627118644068,
      "grad_norm": 0.0346585176885128,
      "learning_rate": 0.00025097087378640775,
      "loss": 0.5873,
      "step": 341
    },
    {
      "epoch": 0.747949699289229,
      "grad_norm": 0.05856507271528244,
      "learning_rate": 0.0002507281553398058,
      "loss": 0.5853,
      "step": 342
    },
    {
      "epoch": 0.7501366867140514,
      "grad_norm": 0.06167592853307724,
      "learning_rate": 0.00025048543689320385,
      "loss": 0.7931,
      "step": 343
    },
    {
      "epoch": 0.7523236741388737,
      "grad_norm": 0.050117284059524536,
      "learning_rate": 0.00025024271844660193,
      "loss": 0.5862,
      "step": 344
    },
    {
      "epoch": 0.7545106615636961,
      "grad_norm": 0.09028457850217819,
      "learning_rate": 0.00025,
      "loss": 0.8125,
      "step": 345
    },
    {
      "epoch": 0.7566976489885183,
      "grad_norm": 0.07750166952610016,
      "learning_rate": 0.00024975728155339803,
      "loss": 0.8384,
      "step": 346
    },
    {
      "epoch": 0.7588846364133406,
      "grad_norm": 0.11774775385856628,
      "learning_rate": 0.0002495145631067961,
      "loss": 0.5295,
      "step": 347
    },
    {
      "epoch": 0.761071623838163,
      "grad_norm": 0.07647069543600082,
      "learning_rate": 0.0002492718446601942,
      "loss": 0.7767,
      "step": 348
    },
    {
      "epoch": 0.7632586112629852,
      "grad_norm": 0.05447927862405777,
      "learning_rate": 0.0002490291262135922,
      "loss": 0.5385,
      "step": 349
    },
    {
      "epoch": 0.7654455986878076,
      "grad_norm": 0.07965091615915298,
      "learning_rate": 0.0002487864077669903,
      "loss": 0.6024,
      "step": 350
    },
    {
      "epoch": 0.7676325861126299,
      "grad_norm": 0.09056472778320312,
      "learning_rate": 0.00024854368932038836,
      "loss": 0.7804,
      "step": 351
    },
    {
      "epoch": 0.7698195735374521,
      "grad_norm": 0.07823322713375092,
      "learning_rate": 0.0002483009708737864,
      "loss": 0.7167,
      "step": 352
    },
    {
      "epoch": 0.7720065609622745,
      "grad_norm": 0.09253627061843872,
      "learning_rate": 0.00024805825242718446,
      "loss": 0.8845,
      "step": 353
    },
    {
      "epoch": 0.7741935483870968,
      "grad_norm": 0.09817380458116531,
      "learning_rate": 0.00024781553398058254,
      "loss": 0.6951,
      "step": 354
    },
    {
      "epoch": 0.7763805358119191,
      "grad_norm": 0.1380646675825119,
      "learning_rate": 0.00024757281553398056,
      "loss": 0.6207,
      "step": 355
    },
    {
      "epoch": 0.7785675232367414,
      "grad_norm": 0.07183520495891571,
      "learning_rate": 0.00024733009708737864,
      "loss": 0.7533,
      "step": 356
    },
    {
      "epoch": 0.7807545106615636,
      "grad_norm": 0.06690984219312668,
      "learning_rate": 0.00024708737864077667,
      "loss": 0.5925,
      "step": 357
    },
    {
      "epoch": 0.782941498086386,
      "grad_norm": 0.06885010749101639,
      "learning_rate": 0.00024684466019417474,
      "loss": 0.6954,
      "step": 358
    },
    {
      "epoch": 0.7851284855112083,
      "grad_norm": 0.12178938090801239,
      "learning_rate": 0.0002466019417475728,
      "loss": 0.7202,
      "step": 359
    },
    {
      "epoch": 0.7873154729360307,
      "grad_norm": 0.05695571005344391,
      "learning_rate": 0.00024635922330097084,
      "loss": 0.4058,
      "step": 360
    },
    {
      "epoch": 0.7895024603608529,
      "grad_norm": 0.0663481280207634,
      "learning_rate": 0.0002461165048543689,
      "loss": 0.611,
      "step": 361
    },
    {
      "epoch": 0.7916894477856752,
      "grad_norm": 0.08184046298265457,
      "learning_rate": 0.00024587378640776695,
      "loss": 0.91,
      "step": 362
    },
    {
      "epoch": 0.7938764352104976,
      "grad_norm": 0.09832902252674103,
      "learning_rate": 0.000245631067961165,
      "loss": 0.8614,
      "step": 363
    },
    {
      "epoch": 0.7960634226353198,
      "grad_norm": 0.09554405510425568,
      "learning_rate": 0.0002453883495145631,
      "loss": 0.738,
      "step": 364
    },
    {
      "epoch": 0.7982504100601422,
      "grad_norm": 0.042675334960222244,
      "learning_rate": 0.0002451456310679611,
      "loss": 0.5809,
      "step": 365
    },
    {
      "epoch": 0.8004373974849645,
      "grad_norm": 0.17304395139217377,
      "learning_rate": 0.0002449029126213592,
      "loss": 0.7967,
      "step": 366
    },
    {
      "epoch": 0.8026243849097867,
      "grad_norm": 0.10548993200063705,
      "learning_rate": 0.0002446601941747572,
      "loss": 0.673,
      "step": 367
    },
    {
      "epoch": 0.8048113723346091,
      "grad_norm": 0.07851067185401917,
      "learning_rate": 0.0002444174757281553,
      "loss": 0.6461,
      "step": 368
    },
    {
      "epoch": 0.8069983597594313,
      "grad_norm": 0.042184650897979736,
      "learning_rate": 0.0002441747572815534,
      "loss": 0.7225,
      "step": 369
    },
    {
      "epoch": 0.8091853471842537,
      "grad_norm": 0.08041860163211823,
      "learning_rate": 0.00024393203883495143,
      "loss": 0.7501,
      "step": 370
    },
    {
      "epoch": 0.811372334609076,
      "grad_norm": 0.11619728803634644,
      "learning_rate": 0.0002436893203883495,
      "loss": 0.5616,
      "step": 371
    },
    {
      "epoch": 0.8135593220338984,
      "grad_norm": 0.07489950209856033,
      "learning_rate": 0.00024344660194174756,
      "loss": 0.4979,
      "step": 372
    },
    {
      "epoch": 0.8157463094587206,
      "grad_norm": 0.1000688374042511,
      "learning_rate": 0.0002432038834951456,
      "loss": 0.8427,
      "step": 373
    },
    {
      "epoch": 0.8179332968835429,
      "grad_norm": 0.12430839985609055,
      "learning_rate": 0.00024296116504854366,
      "loss": 0.6974,
      "step": 374
    },
    {
      "epoch": 0.8201202843083653,
      "grad_norm": 0.04190218076109886,
      "learning_rate": 0.00024271844660194174,
      "loss": 0.6709,
      "step": 375
    },
    {
      "epoch": 0.8223072717331875,
      "grad_norm": 0.04408605396747589,
      "learning_rate": 0.0002424757281553398,
      "loss": 0.7654,
      "step": 376
    },
    {
      "epoch": 0.8244942591580099,
      "grad_norm": 0.09234176576137543,
      "learning_rate": 0.00024223300970873784,
      "loss": 0.6825,
      "step": 377
    },
    {
      "epoch": 0.8266812465828322,
      "grad_norm": 0.04161737486720085,
      "learning_rate": 0.00024199029126213592,
      "loss": 0.5359,
      "step": 378
    },
    {
      "epoch": 0.8288682340076544,
      "grad_norm": 0.0515722818672657,
      "learning_rate": 0.00024174757281553394,
      "loss": 0.7162,
      "step": 379
    },
    {
      "epoch": 0.8310552214324768,
      "grad_norm": 0.028478797525167465,
      "learning_rate": 0.00024150485436893202,
      "loss": 0.531,
      "step": 380
    },
    {
      "epoch": 0.833242208857299,
      "grad_norm": 0.06542545557022095,
      "learning_rate": 0.0002412621359223301,
      "loss": 0.726,
      "step": 381
    },
    {
      "epoch": 0.8354291962821214,
      "grad_norm": 0.04673764482140541,
      "learning_rate": 0.00024101941747572812,
      "loss": 0.5919,
      "step": 382
    },
    {
      "epoch": 0.8376161837069437,
      "grad_norm": 0.09057341516017914,
      "learning_rate": 0.0002407766990291262,
      "loss": 0.8387,
      "step": 383
    },
    {
      "epoch": 0.839803171131766,
      "grad_norm": 0.06963503360748291,
      "learning_rate": 0.00024053398058252427,
      "loss": 0.5094,
      "step": 384
    },
    {
      "epoch": 0.8419901585565883,
      "grad_norm": 0.10514622181653976,
      "learning_rate": 0.0002402912621359223,
      "loss": 0.7377,
      "step": 385
    },
    {
      "epoch": 0.8441771459814106,
      "grad_norm": 0.15959374606609344,
      "learning_rate": 0.00024004854368932037,
      "loss": 0.5388,
      "step": 386
    },
    {
      "epoch": 0.846364133406233,
      "grad_norm": 0.058087654411792755,
      "learning_rate": 0.00023980582524271845,
      "loss": 0.6291,
      "step": 387
    },
    {
      "epoch": 0.8485511208310552,
      "grad_norm": 0.06005719676613808,
      "learning_rate": 0.00023956310679611648,
      "loss": 0.8149,
      "step": 388
    },
    {
      "epoch": 0.8507381082558775,
      "grad_norm": 0.0970982164144516,
      "learning_rate": 0.00023932038834951455,
      "loss": 0.7297,
      "step": 389
    },
    {
      "epoch": 0.8529250956806999,
      "grad_norm": 0.14135178923606873,
      "learning_rate": 0.0002390776699029126,
      "loss": 0.5565,
      "step": 390
    },
    {
      "epoch": 0.8551120831055221,
      "grad_norm": 0.07173170149326324,
      "learning_rate": 0.00023883495145631065,
      "loss": 0.7412,
      "step": 391
    },
    {
      "epoch": 0.8572990705303445,
      "grad_norm": 0.07636596262454987,
      "learning_rate": 0.00023859223300970873,
      "loss": 0.6125,
      "step": 392
    },
    {
      "epoch": 0.8594860579551667,
      "grad_norm": 0.10493489354848862,
      "learning_rate": 0.00023834951456310678,
      "loss": 0.66,
      "step": 393
    },
    {
      "epoch": 0.861673045379989,
      "grad_norm": 0.08531126379966736,
      "learning_rate": 0.00023810679611650483,
      "loss": 0.8011,
      "step": 394
    },
    {
      "epoch": 0.8638600328048114,
      "grad_norm": 0.0813411995768547,
      "learning_rate": 0.0002378640776699029,
      "loss": 0.7221,
      "step": 395
    },
    {
      "epoch": 0.8660470202296336,
      "grad_norm": 0.07029952853918076,
      "learning_rate": 0.00023762135922330096,
      "loss": 0.6196,
      "step": 396
    },
    {
      "epoch": 0.868234007654456,
      "grad_norm": 0.05257638543844223,
      "learning_rate": 0.000237378640776699,
      "loss": 0.5878,
      "step": 397
    },
    {
      "epoch": 0.8704209950792783,
      "grad_norm": 0.0511767715215683,
      "learning_rate": 0.00023713592233009706,
      "loss": 0.5656,
      "step": 398
    },
    {
      "epoch": 0.8726079825041007,
      "grad_norm": 0.04756753519177437,
      "learning_rate": 0.0002368932038834951,
      "loss": 0.6717,
      "step": 399
    },
    {
      "epoch": 0.8747949699289229,
      "grad_norm": 0.05377547815442085,
      "learning_rate": 0.0002366504854368932,
      "loss": 0.6772,
      "step": 400
    },
    {
      "epoch": 0.8769819573537452,
      "grad_norm": 0.048965323716402054,
      "learning_rate": 0.00023640776699029124,
      "loss": 0.7175,
      "step": 401
    },
    {
      "epoch": 0.8791689447785676,
      "grad_norm": 0.0994691550731659,
      "learning_rate": 0.0002361650485436893,
      "loss": 0.6364,
      "step": 402
    },
    {
      "epoch": 0.8813559322033898,
      "grad_norm": 0.11497332155704498,
      "learning_rate": 0.00023592233009708734,
      "loss": 0.8192,
      "step": 403
    },
    {
      "epoch": 0.8835429196282122,
      "grad_norm": 0.05721667408943176,
      "learning_rate": 0.00023567961165048542,
      "loss": 0.6384,
      "step": 404
    },
    {
      "epoch": 0.8857299070530344,
      "grad_norm": 0.09783297032117844,
      "learning_rate": 0.00023543689320388347,
      "loss": 0.73,
      "step": 405
    },
    {
      "epoch": 0.8879168944778567,
      "grad_norm": 0.06903696805238724,
      "learning_rate": 0.00023519417475728152,
      "loss": 0.6751,
      "step": 406
    },
    {
      "epoch": 0.8901038819026791,
      "grad_norm": 0.031625282019376755,
      "learning_rate": 0.0002349514563106796,
      "loss": 0.4768,
      "step": 407
    },
    {
      "epoch": 0.8922908693275013,
      "grad_norm": 0.09231697767972946,
      "learning_rate": 0.00023470873786407762,
      "loss": 0.5265,
      "step": 408
    },
    {
      "epoch": 0.8944778567523237,
      "grad_norm": 0.06142432987689972,
      "learning_rate": 0.0002344660194174757,
      "loss": 0.7142,
      "step": 409
    },
    {
      "epoch": 0.896664844177146,
      "grad_norm": 0.07211441546678543,
      "learning_rate": 0.00023422330097087378,
      "loss": 0.7355,
      "step": 410
    },
    {
      "epoch": 0.8988518316019682,
      "grad_norm": 0.05186145380139351,
      "learning_rate": 0.0002339805825242718,
      "loss": 0.7309,
      "step": 411
    },
    {
      "epoch": 0.9010388190267906,
      "grad_norm": 0.07550069689750671,
      "learning_rate": 0.00023373786407766988,
      "loss": 0.5684,
      "step": 412
    },
    {
      "epoch": 0.9032258064516129,
      "grad_norm": 0.11441147327423096,
      "learning_rate": 0.00023349514563106795,
      "loss": 0.7721,
      "step": 413
    },
    {
      "epoch": 0.9054127938764353,
      "grad_norm": 0.03658903390169144,
      "learning_rate": 0.00023325242718446598,
      "loss": 0.6225,
      "step": 414
    },
    {
      "epoch": 0.9075997813012575,
      "grad_norm": 0.0618315227329731,
      "learning_rate": 0.00023300970873786406,
      "loss": 0.6044,
      "step": 415
    },
    {
      "epoch": 0.9097867687260798,
      "grad_norm": 0.10633370280265808,
      "learning_rate": 0.00023276699029126213,
      "loss": 0.8217,
      "step": 416
    },
    {
      "epoch": 0.9119737561509021,
      "grad_norm": 0.09450804442167282,
      "learning_rate": 0.00023252427184466016,
      "loss": 0.6532,
      "step": 417
    },
    {
      "epoch": 0.9141607435757244,
      "grad_norm": 0.04320994019508362,
      "learning_rate": 0.00023228155339805823,
      "loss": 0.6539,
      "step": 418
    },
    {
      "epoch": 0.9163477310005468,
      "grad_norm": 0.1247217059135437,
      "learning_rate": 0.00023203883495145629,
      "loss": 0.6525,
      "step": 419
    },
    {
      "epoch": 0.918534718425369,
      "grad_norm": 0.06656447798013687,
      "learning_rate": 0.00023179611650485434,
      "loss": 0.6559,
      "step": 420
    },
    {
      "epoch": 0.9207217058501913,
      "grad_norm": 0.06622477620840073,
      "learning_rate": 0.0002315533980582524,
      "loss": 0.8571,
      "step": 421
    },
    {
      "epoch": 0.9229086932750137,
      "grad_norm": 0.07110453397035599,
      "learning_rate": 0.00023131067961165046,
      "loss": 0.6829,
      "step": 422
    },
    {
      "epoch": 0.9250956806998359,
      "grad_norm": 0.09340423345565796,
      "learning_rate": 0.00023106796116504851,
      "loss": 0.5177,
      "step": 423
    },
    {
      "epoch": 0.9272826681246583,
      "grad_norm": 0.08478957414627075,
      "learning_rate": 0.0002308252427184466,
      "loss": 0.5144,
      "step": 424
    },
    {
      "epoch": 0.9294696555494806,
      "grad_norm": 0.08036354184150696,
      "learning_rate": 0.00023058252427184464,
      "loss": 0.5751,
      "step": 425
    },
    {
      "epoch": 0.931656642974303,
      "grad_norm": 0.16781304776668549,
      "learning_rate": 0.0002303398058252427,
      "loss": 0.7269,
      "step": 426
    },
    {
      "epoch": 0.9338436303991252,
      "grad_norm": 0.04996452108025551,
      "learning_rate": 0.00023009708737864074,
      "loss": 0.9213,
      "step": 427
    },
    {
      "epoch": 0.9360306178239475,
      "grad_norm": 0.06372474879026413,
      "learning_rate": 0.00022985436893203882,
      "loss": 0.7192,
      "step": 428
    },
    {
      "epoch": 0.9382176052487698,
      "grad_norm": 0.08853182196617126,
      "learning_rate": 0.00022961165048543687,
      "loss": 0.6885,
      "step": 429
    },
    {
      "epoch": 0.9404045926735921,
      "grad_norm": 0.09669560939073563,
      "learning_rate": 0.00022936893203883492,
      "loss": 0.7755,
      "step": 430
    },
    {
      "epoch": 0.9425915800984145,
      "grad_norm": 0.08254298567771912,
      "learning_rate": 0.000229126213592233,
      "loss": 0.6438,
      "step": 431
    },
    {
      "epoch": 0.9447785675232367,
      "grad_norm": 0.05825486406683922,
      "learning_rate": 0.00022888349514563102,
      "loss": 0.6271,
      "step": 432
    },
    {
      "epoch": 0.946965554948059,
      "grad_norm": 0.07293659448623657,
      "learning_rate": 0.0002286407766990291,
      "loss": 0.7427,
      "step": 433
    },
    {
      "epoch": 0.9491525423728814,
      "grad_norm": 0.06341248005628586,
      "learning_rate": 0.00022839805825242718,
      "loss": 0.5967,
      "step": 434
    },
    {
      "epoch": 0.9513395297977036,
      "grad_norm": 0.05031775310635567,
      "learning_rate": 0.0002281553398058252,
      "loss": 0.8006,
      "step": 435
    },
    {
      "epoch": 0.953526517222526,
      "grad_norm": 0.07868911325931549,
      "learning_rate": 0.00022791262135922328,
      "loss": 0.7485,
      "step": 436
    },
    {
      "epoch": 0.9557135046473483,
      "grad_norm": 0.048034172505140305,
      "learning_rate": 0.00022766990291262136,
      "loss": 0.5678,
      "step": 437
    },
    {
      "epoch": 0.9579004920721705,
      "grad_norm": 0.24086016416549683,
      "learning_rate": 0.00022742718446601938,
      "loss": 0.5608,
      "step": 438
    },
    {
      "epoch": 0.9600874794969929,
      "grad_norm": 0.11558078229427338,
      "learning_rate": 0.00022718446601941746,
      "loss": 0.6857,
      "step": 439
    },
    {
      "epoch": 0.9622744669218152,
      "grad_norm": 0.0938619077205658,
      "learning_rate": 0.00022694174757281554,
      "loss": 0.7646,
      "step": 440
    },
    {
      "epoch": 0.9644614543466375,
      "grad_norm": 0.06463375687599182,
      "learning_rate": 0.00022669902912621356,
      "loss": 0.5471,
      "step": 441
    },
    {
      "epoch": 0.9666484417714598,
      "grad_norm": 0.09269668161869049,
      "learning_rate": 0.00022645631067961164,
      "loss": 0.599,
      "step": 442
    },
    {
      "epoch": 0.9688354291962821,
      "grad_norm": 0.05435742810368538,
      "learning_rate": 0.0002262135922330097,
      "loss": 0.6006,
      "step": 443
    },
    {
      "epoch": 0.9710224166211044,
      "grad_norm": 0.08216127008199692,
      "learning_rate": 0.00022597087378640774,
      "loss": 0.9168,
      "step": 444
    },
    {
      "epoch": 0.9732094040459267,
      "grad_norm": 0.09296419471502304,
      "learning_rate": 0.00022572815533980582,
      "loss": 0.7166,
      "step": 445
    },
    {
      "epoch": 0.9753963914707491,
      "grad_norm": 0.04655352607369423,
      "learning_rate": 0.00022548543689320387,
      "loss": 0.6178,
      "step": 446
    },
    {
      "epoch": 0.9775833788955713,
      "grad_norm": 0.06913512200117111,
      "learning_rate": 0.00022524271844660192,
      "loss": 0.6967,
      "step": 447
    },
    {
      "epoch": 0.9797703663203936,
      "grad_norm": 0.06842842698097229,
      "learning_rate": 0.000225,
      "loss": 0.6422,
      "step": 448
    },
    {
      "epoch": 0.981957353745216,
      "grad_norm": 0.05165997892618179,
      "learning_rate": 0.00022475728155339804,
      "loss": 0.8705,
      "step": 449
    },
    {
      "epoch": 0.9841443411700382,
      "grad_norm": 0.051989976316690445,
      "learning_rate": 0.0002245145631067961,
      "loss": 0.6084,
      "step": 450
    },
    {
      "epoch": 0.9863313285948606,
      "grad_norm": 0.04412755370140076,
      "learning_rate": 0.00022427184466019415,
      "loss": 0.894,
      "step": 451
    },
    {
      "epoch": 0.9885183160196829,
      "grad_norm": 0.0869666188955307,
      "learning_rate": 0.00022402912621359222,
      "loss": 0.8823,
      "step": 452
    },
    {
      "epoch": 0.9907053034445052,
      "grad_norm": 0.09454749524593353,
      "learning_rate": 0.00022378640776699027,
      "loss": 0.777,
      "step": 453
    },
    {
      "epoch": 0.9928922908693275,
      "grad_norm": 0.06162366643548012,
      "learning_rate": 0.00022354368932038832,
      "loss": 0.6904,
      "step": 454
    },
    {
      "epoch": 0.9950792782941498,
      "grad_norm": 0.06763439625501633,
      "learning_rate": 0.0002233009708737864,
      "loss": 0.624,
      "step": 455
    },
    {
      "epoch": 0.9972662657189721,
      "grad_norm": 0.09152118861675262,
      "learning_rate": 0.00022305825242718443,
      "loss": 0.8294,
      "step": 456
    },
    {
      "epoch": 0.9994532531437944,
      "grad_norm": 0.09418030828237534,
      "learning_rate": 0.0002228155339805825,
      "loss": 0.6718,
      "step": 457
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.304161936044693,
      "learning_rate": 0.00022257281553398058,
      "loss": 0.7043,
      "step": 458
    },
    {
      "epoch": 1.0021869874248224,
      "grad_norm": 0.0715775266289711,
      "learning_rate": 0.0002223300970873786,
      "loss": 0.8464,
      "step": 459
    },
    {
      "epoch": 1.0043739748496445,
      "grad_norm": 0.07122276723384857,
      "learning_rate": 0.00022208737864077668,
      "loss": 0.9641,
      "step": 460
    },
    {
      "epoch": 1.006560962274467,
      "grad_norm": 0.07358932495117188,
      "learning_rate": 0.00022184466019417476,
      "loss": 0.6549,
      "step": 461
    },
    {
      "epoch": 1.0087479496992893,
      "grad_norm": 0.07040970772504807,
      "learning_rate": 0.00022160194174757278,
      "loss": 0.4505,
      "step": 462
    },
    {
      "epoch": 1.0109349371241116,
      "grad_norm": 0.06049482896924019,
      "learning_rate": 0.00022135922330097086,
      "loss": 0.5927,
      "step": 463
    },
    {
      "epoch": 1.0131219245489338,
      "grad_norm": 0.0695071890950203,
      "learning_rate": 0.00022111650485436894,
      "loss": 0.5383,
      "step": 464
    },
    {
      "epoch": 1.0153089119737562,
      "grad_norm": 0.08132364600896835,
      "learning_rate": 0.00022087378640776696,
      "loss": 0.847,
      "step": 465
    },
    {
      "epoch": 1.0174958993985785,
      "grad_norm": 0.05544696003198624,
      "learning_rate": 0.00022063106796116504,
      "loss": 0.4345,
      "step": 466
    },
    {
      "epoch": 1.0196828868234007,
      "grad_norm": 0.05969703197479248,
      "learning_rate": 0.0002203883495145631,
      "loss": 0.4453,
      "step": 467
    },
    {
      "epoch": 1.021869874248223,
      "grad_norm": 0.1177251935005188,
      "learning_rate": 0.00022014563106796114,
      "loss": 0.7104,
      "step": 468
    },
    {
      "epoch": 1.0240568616730454,
      "grad_norm": 0.06920936703681946,
      "learning_rate": 0.00021990291262135922,
      "loss": 0.7134,
      "step": 469
    },
    {
      "epoch": 1.0262438490978676,
      "grad_norm": 0.07872872799634933,
      "learning_rate": 0.00021966019417475727,
      "loss": 0.7264,
      "step": 470
    },
    {
      "epoch": 1.02843083652269,
      "grad_norm": 0.07836636900901794,
      "learning_rate": 0.00021941747572815532,
      "loss": 0.67,
      "step": 471
    },
    {
      "epoch": 1.0306178239475123,
      "grad_norm": 0.11644013226032257,
      "learning_rate": 0.00021917475728155337,
      "loss": 0.6805,
      "step": 472
    },
    {
      "epoch": 1.0328048113723347,
      "grad_norm": 0.0698661059141159,
      "learning_rate": 0.00021893203883495145,
      "loss": 0.8073,
      "step": 473
    },
    {
      "epoch": 1.0349917987971569,
      "grad_norm": 0.11226359009742737,
      "learning_rate": 0.0002186893203883495,
      "loss": 0.7748,
      "step": 474
    },
    {
      "epoch": 1.0371787862219792,
      "grad_norm": 0.05350861698389053,
      "learning_rate": 0.00021844660194174755,
      "loss": 0.3718,
      "step": 475
    },
    {
      "epoch": 1.0393657736468016,
      "grad_norm": 0.08481690287590027,
      "learning_rate": 0.00021820388349514562,
      "loss": 0.7288,
      "step": 476
    },
    {
      "epoch": 1.0415527610716238,
      "grad_norm": 0.03595307841897011,
      "learning_rate": 0.00021796116504854368,
      "loss": 0.3986,
      "step": 477
    },
    {
      "epoch": 1.0437397484964461,
      "grad_norm": 0.03737505525350571,
      "learning_rate": 0.00021771844660194173,
      "loss": 0.7487,
      "step": 478
    },
    {
      "epoch": 1.0459267359212685,
      "grad_norm": 0.09312517195940018,
      "learning_rate": 0.0002174757281553398,
      "loss": 0.6514,
      "step": 479
    },
    {
      "epoch": 1.0481137233460907,
      "grad_norm": 0.04990383982658386,
      "learning_rate": 0.00021723300970873783,
      "loss": 0.5012,
      "step": 480
    },
    {
      "epoch": 1.050300710770913,
      "grad_norm": 0.051016416400671005,
      "learning_rate": 0.0002169902912621359,
      "loss": 0.6223,
      "step": 481
    },
    {
      "epoch": 1.0524876981957354,
      "grad_norm": 0.10985545814037323,
      "learning_rate": 0.00021674757281553398,
      "loss": 0.7192,
      "step": 482
    },
    {
      "epoch": 1.0546746856205578,
      "grad_norm": 0.13058781623840332,
      "learning_rate": 0.000216504854368932,
      "loss": 0.6935,
      "step": 483
    },
    {
      "epoch": 1.05686167304538,
      "grad_norm": 0.08312558382749557,
      "learning_rate": 0.00021626213592233008,
      "loss": 0.7496,
      "step": 484
    },
    {
      "epoch": 1.0590486604702023,
      "grad_norm": 0.1301640123128891,
      "learning_rate": 0.00021601941747572816,
      "loss": 0.6052,
      "step": 485
    },
    {
      "epoch": 1.0612356478950247,
      "grad_norm": 0.09434215724468231,
      "learning_rate": 0.00021577669902912618,
      "loss": 0.7594,
      "step": 486
    },
    {
      "epoch": 1.0634226353198468,
      "grad_norm": 0.10083501785993576,
      "learning_rate": 0.00021553398058252426,
      "loss": 0.7177,
      "step": 487
    },
    {
      "epoch": 1.0656096227446692,
      "grad_norm": 0.055385932326316833,
      "learning_rate": 0.00021529126213592234,
      "loss": 0.5035,
      "step": 488
    },
    {
      "epoch": 1.0677966101694916,
      "grad_norm": 0.04303668439388275,
      "learning_rate": 0.00021504854368932036,
      "loss": 0.6053,
      "step": 489
    },
    {
      "epoch": 1.0699835975943137,
      "grad_norm": 0.10392434149980545,
      "learning_rate": 0.00021480582524271844,
      "loss": 0.7089,
      "step": 490
    },
    {
      "epoch": 1.072170585019136,
      "grad_norm": 0.08201673626899719,
      "learning_rate": 0.0002145631067961165,
      "loss": 0.7632,
      "step": 491
    },
    {
      "epoch": 1.0743575724439585,
      "grad_norm": 0.045085739344358444,
      "learning_rate": 0.00021432038834951454,
      "loss": 0.6817,
      "step": 492
    },
    {
      "epoch": 1.0765445598687808,
      "grad_norm": 0.07135611772537231,
      "learning_rate": 0.00021407766990291262,
      "loss": 0.6935,
      "step": 493
    },
    {
      "epoch": 1.078731547293603,
      "grad_norm": 0.05138889327645302,
      "learning_rate": 0.00021383495145631067,
      "loss": 0.4929,
      "step": 494
    },
    {
      "epoch": 1.0809185347184254,
      "grad_norm": 0.08583348989486694,
      "learning_rate": 0.00021359223300970872,
      "loss": 0.5905,
      "step": 495
    },
    {
      "epoch": 1.0831055221432477,
      "grad_norm": 0.04269842803478241,
      "learning_rate": 0.00021334951456310677,
      "loss": 0.5433,
      "step": 496
    },
    {
      "epoch": 1.0852925095680699,
      "grad_norm": 0.08027777820825577,
      "learning_rate": 0.00021310679611650485,
      "loss": 0.6686,
      "step": 497
    },
    {
      "epoch": 1.0874794969928923,
      "grad_norm": 0.03481656312942505,
      "learning_rate": 0.0002128640776699029,
      "loss": 0.6998,
      "step": 498
    },
    {
      "epoch": 1.0896664844177146,
      "grad_norm": 0.07240082323551178,
      "learning_rate": 0.00021262135922330095,
      "loss": 0.607,
      "step": 499
    },
    {
      "epoch": 1.091853471842537,
      "grad_norm": 0.05366026610136032,
      "learning_rate": 0.000212378640776699,
      "loss": 1.2503,
      "step": 500
    },
    {
      "epoch": 1.0940404592673592,
      "grad_norm": 0.07942108064889908,
      "learning_rate": 0.00021213592233009705,
      "loss": 0.5248,
      "step": 501
    },
    {
      "epoch": 1.0962274466921815,
      "grad_norm": 0.10274556279182434,
      "learning_rate": 0.00021189320388349513,
      "loss": 0.7508,
      "step": 502
    },
    {
      "epoch": 1.098414434117004,
      "grad_norm": 0.0703059732913971,
      "learning_rate": 0.00021165048543689318,
      "loss": 0.7144,
      "step": 503
    },
    {
      "epoch": 1.100601421541826,
      "grad_norm": 0.05571672320365906,
      "learning_rate": 0.00021140776699029123,
      "loss": 0.6688,
      "step": 504
    },
    {
      "epoch": 1.1027884089666484,
      "grad_norm": 0.06484287977218628,
      "learning_rate": 0.0002111650485436893,
      "loss": 0.8054,
      "step": 505
    },
    {
      "epoch": 1.1049753963914708,
      "grad_norm": 0.06399201601743698,
      "learning_rate": 0.00021092233009708736,
      "loss": 0.7235,
      "step": 506
    },
    {
      "epoch": 1.1071623838162932,
      "grad_norm": 0.10094781219959259,
      "learning_rate": 0.0002106796116504854,
      "loss": 0.7598,
      "step": 507
    },
    {
      "epoch": 1.1093493712411153,
      "grad_norm": 0.058417245745658875,
      "learning_rate": 0.00021043689320388349,
      "loss": 0.7093,
      "step": 508
    },
    {
      "epoch": 1.1115363586659377,
      "grad_norm": 0.07317756116390228,
      "learning_rate": 0.0002101941747572815,
      "loss": 0.4897,
      "step": 509
    },
    {
      "epoch": 1.11372334609076,
      "grad_norm": 0.06851299107074738,
      "learning_rate": 0.00020995145631067959,
      "loss": 0.6625,
      "step": 510
    },
    {
      "epoch": 1.1159103335155822,
      "grad_norm": 0.02538035437464714,
      "learning_rate": 0.00020970873786407766,
      "loss": 0.5364,
      "step": 511
    },
    {
      "epoch": 1.1180973209404046,
      "grad_norm": 0.030080309137701988,
      "learning_rate": 0.0002094660194174757,
      "loss": 0.5308,
      "step": 512
    },
    {
      "epoch": 1.120284308365227,
      "grad_norm": 0.053607307374477386,
      "learning_rate": 0.00020922330097087376,
      "loss": 0.6043,
      "step": 513
    },
    {
      "epoch": 1.1224712957900491,
      "grad_norm": 0.06043059006333351,
      "learning_rate": 0.00020898058252427184,
      "loss": 0.7792,
      "step": 514
    },
    {
      "epoch": 1.1246582832148715,
      "grad_norm": 0.0692690908908844,
      "learning_rate": 0.00020873786407766987,
      "loss": 0.6497,
      "step": 515
    },
    {
      "epoch": 1.1268452706396939,
      "grad_norm": 0.041172128170728683,
      "learning_rate": 0.00020849514563106794,
      "loss": 0.5175,
      "step": 516
    },
    {
      "epoch": 1.129032258064516,
      "grad_norm": 0.030904235318303108,
      "learning_rate": 0.00020825242718446602,
      "loss": 0.7253,
      "step": 517
    },
    {
      "epoch": 1.1312192454893384,
      "grad_norm": 0.10942153632640839,
      "learning_rate": 0.00020800970873786404,
      "loss": 0.486,
      "step": 518
    },
    {
      "epoch": 1.1334062329141608,
      "grad_norm": 0.05069408193230629,
      "learning_rate": 0.00020776699029126212,
      "loss": 0.6905,
      "step": 519
    },
    {
      "epoch": 1.1355932203389831,
      "grad_norm": 0.08336110413074493,
      "learning_rate": 0.00020752427184466017,
      "loss": 0.3925,
      "step": 520
    },
    {
      "epoch": 1.1377802077638053,
      "grad_norm": 0.07145116478204727,
      "learning_rate": 0.00020728155339805822,
      "loss": 0.5243,
      "step": 521
    },
    {
      "epoch": 1.1399671951886277,
      "grad_norm": 0.06341274827718735,
      "learning_rate": 0.0002070388349514563,
      "loss": 0.541,
      "step": 522
    },
    {
      "epoch": 1.14215418261345,
      "grad_norm": 0.07657375186681747,
      "learning_rate": 0.00020679611650485435,
      "loss": 0.6905,
      "step": 523
    },
    {
      "epoch": 1.1443411700382722,
      "grad_norm": 0.055964358150959015,
      "learning_rate": 0.0002065533980582524,
      "loss": 0.7425,
      "step": 524
    },
    {
      "epoch": 1.1465281574630946,
      "grad_norm": 0.046841271221637726,
      "learning_rate": 0.00020631067961165045,
      "loss": 0.6434,
      "step": 525
    },
    {
      "epoch": 1.148715144887917,
      "grad_norm": 0.05508289113640785,
      "learning_rate": 0.00020606796116504853,
      "loss": 0.4618,
      "step": 526
    },
    {
      "epoch": 1.1509021323127393,
      "grad_norm": 0.08871844410896301,
      "learning_rate": 0.00020582524271844658,
      "loss": 0.5587,
      "step": 527
    },
    {
      "epoch": 1.1530891197375615,
      "grad_norm": 0.07026297599077225,
      "learning_rate": 0.00020558252427184463,
      "loss": 0.4971,
      "step": 528
    },
    {
      "epoch": 1.1552761071623838,
      "grad_norm": 0.07413173466920853,
      "learning_rate": 0.0002053398058252427,
      "loss": 0.7721,
      "step": 529
    },
    {
      "epoch": 1.1574630945872062,
      "grad_norm": 0.11381928622722626,
      "learning_rate": 0.00020509708737864073,
      "loss": 0.578,
      "step": 530
    },
    {
      "epoch": 1.1596500820120283,
      "grad_norm": 0.11267655342817307,
      "learning_rate": 0.0002048543689320388,
      "loss": 0.7266,
      "step": 531
    },
    {
      "epoch": 1.1618370694368507,
      "grad_norm": 0.05957537144422531,
      "learning_rate": 0.0002046116504854369,
      "loss": 0.6555,
      "step": 532
    },
    {
      "epoch": 1.164024056861673,
      "grad_norm": 0.07506536692380905,
      "learning_rate": 0.0002043689320388349,
      "loss": 0.5954,
      "step": 533
    },
    {
      "epoch": 1.1662110442864955,
      "grad_norm": 0.04716431349515915,
      "learning_rate": 0.000204126213592233,
      "loss": 0.7473,
      "step": 534
    },
    {
      "epoch": 1.1683980317113176,
      "grad_norm": 0.076003298163414,
      "learning_rate": 0.00020388349514563107,
      "loss": 0.7488,
      "step": 535
    },
    {
      "epoch": 1.17058501913614,
      "grad_norm": 0.07676059007644653,
      "learning_rate": 0.0002036407766990291,
      "loss": 0.5341,
      "step": 536
    },
    {
      "epoch": 1.1727720065609624,
      "grad_norm": 0.037800587713718414,
      "learning_rate": 0.00020339805825242717,
      "loss": 0.6851,
      "step": 537
    },
    {
      "epoch": 1.1749589939857845,
      "grad_norm": 0.05885176360607147,
      "learning_rate": 0.00020315533980582524,
      "loss": 0.5632,
      "step": 538
    },
    {
      "epoch": 1.177145981410607,
      "grad_norm": 0.048130929470062256,
      "learning_rate": 0.00020291262135922327,
      "loss": 0.6632,
      "step": 539
    },
    {
      "epoch": 1.1793329688354293,
      "grad_norm": 0.19910819828510284,
      "learning_rate": 0.00020266990291262135,
      "loss": 0.6354,
      "step": 540
    },
    {
      "epoch": 1.1815199562602514,
      "grad_norm": 0.07047896087169647,
      "learning_rate": 0.00020242718446601942,
      "loss": 0.8063,
      "step": 541
    },
    {
      "epoch": 1.1837069436850738,
      "grad_norm": 0.11079243570566177,
      "learning_rate": 0.00020218446601941745,
      "loss": 0.6696,
      "step": 542
    },
    {
      "epoch": 1.1858939311098962,
      "grad_norm": 0.0585208535194397,
      "learning_rate": 0.00020194174757281552,
      "loss": 0.5988,
      "step": 543
    },
    {
      "epoch": 1.1880809185347183,
      "grad_norm": 0.13656085729599,
      "learning_rate": 0.00020169902912621357,
      "loss": 0.5899,
      "step": 544
    },
    {
      "epoch": 1.1902679059595407,
      "grad_norm": 0.0442144051194191,
      "learning_rate": 0.00020145631067961163,
      "loss": 0.6737,
      "step": 545
    },
    {
      "epoch": 1.192454893384363,
      "grad_norm": 0.050628334283828735,
      "learning_rate": 0.0002012135922330097,
      "loss": 0.5829,
      "step": 546
    },
    {
      "epoch": 1.1946418808091854,
      "grad_norm": 0.0832197517156601,
      "learning_rate": 0.00020097087378640775,
      "loss": 0.6445,
      "step": 547
    },
    {
      "epoch": 1.1968288682340076,
      "grad_norm": 0.05916838347911835,
      "learning_rate": 0.0002007281553398058,
      "loss": 0.646,
      "step": 548
    },
    {
      "epoch": 1.19901585565883,
      "grad_norm": 0.04812471941113472,
      "learning_rate": 0.00020048543689320385,
      "loss": 0.6882,
      "step": 549
    },
    {
      "epoch": 1.2012028430836523,
      "grad_norm": 0.09752511233091354,
      "learning_rate": 0.00020024271844660193,
      "loss": 0.6365,
      "step": 550
    },
    {
      "epoch": 1.2033898305084745,
      "grad_norm": 0.07826195657253265,
      "learning_rate": 0.00019999999999999998,
      "loss": 0.6926,
      "step": 551
    },
    {
      "epoch": 1.2055768179332969,
      "grad_norm": 0.07991279661655426,
      "learning_rate": 0.00019975728155339803,
      "loss": 0.8615,
      "step": 552
    },
    {
      "epoch": 1.2077638053581192,
      "grad_norm": 0.06553139537572861,
      "learning_rate": 0.0001995145631067961,
      "loss": 0.648,
      "step": 553
    },
    {
      "epoch": 1.2099507927829416,
      "grad_norm": 0.03314918279647827,
      "learning_rate": 0.00019927184466019413,
      "loss": 0.6059,
      "step": 554
    },
    {
      "epoch": 1.2121377802077637,
      "grad_norm": 0.11380508542060852,
      "learning_rate": 0.0001990291262135922,
      "loss": 0.723,
      "step": 555
    },
    {
      "epoch": 1.2143247676325861,
      "grad_norm": 0.07652546465396881,
      "learning_rate": 0.0001987864077669903,
      "loss": 0.7122,
      "step": 556
    },
    {
      "epoch": 1.2165117550574085,
      "grad_norm": 0.03759431838989258,
      "learning_rate": 0.0001985436893203883,
      "loss": 0.367,
      "step": 557
    },
    {
      "epoch": 1.2186987424822306,
      "grad_norm": 0.05255873501300812,
      "learning_rate": 0.0001983009708737864,
      "loss": 0.6797,
      "step": 558
    },
    {
      "epoch": 1.220885729907053,
      "grad_norm": 0.051489852368831635,
      "learning_rate": 0.00019805825242718447,
      "loss": 0.5864,
      "step": 559
    },
    {
      "epoch": 1.2230727173318754,
      "grad_norm": 0.11085233092308044,
      "learning_rate": 0.0001978155339805825,
      "loss": 0.5889,
      "step": 560
    },
    {
      "epoch": 1.2252597047566978,
      "grad_norm": 0.06055936589837074,
      "learning_rate": 0.00019757281553398057,
      "loss": 0.6423,
      "step": 561
    },
    {
      "epoch": 1.22744669218152,
      "grad_norm": 0.04101664200425148,
      "learning_rate": 0.00019733009708737865,
      "loss": 0.5727,
      "step": 562
    },
    {
      "epoch": 1.2296336796063423,
      "grad_norm": 0.03171277046203613,
      "learning_rate": 0.00019708737864077667,
      "loss": 0.7185,
      "step": 563
    },
    {
      "epoch": 1.2318206670311647,
      "grad_norm": 0.06788396835327148,
      "learning_rate": 0.00019684466019417475,
      "loss": 0.6122,
      "step": 564
    },
    {
      "epoch": 1.2340076544559868,
      "grad_norm": 0.08057110011577606,
      "learning_rate": 0.0001966019417475728,
      "loss": 0.7807,
      "step": 565
    },
    {
      "epoch": 1.2361946418808092,
      "grad_norm": 0.05691580846905708,
      "learning_rate": 0.00019635922330097085,
      "loss": 0.6171,
      "step": 566
    },
    {
      "epoch": 1.2383816293056316,
      "grad_norm": 0.14753122627735138,
      "learning_rate": 0.00019611650485436893,
      "loss": 0.6721,
      "step": 567
    },
    {
      "epoch": 1.2405686167304537,
      "grad_norm": 0.10009617358446121,
      "learning_rate": 0.00019587378640776698,
      "loss": 0.7494,
      "step": 568
    },
    {
      "epoch": 1.242755604155276,
      "grad_norm": 0.07308393716812134,
      "learning_rate": 0.00019563106796116503,
      "loss": 0.6882,
      "step": 569
    },
    {
      "epoch": 1.2449425915800985,
      "grad_norm": 0.062100425362586975,
      "learning_rate": 0.0001953883495145631,
      "loss": 0.8307,
      "step": 570
    },
    {
      "epoch": 1.2471295790049206,
      "grad_norm": 0.06036095321178436,
      "learning_rate": 0.00019514563106796116,
      "loss": 0.6629,
      "step": 571
    },
    {
      "epoch": 1.249316566429743,
      "grad_norm": 0.0912746712565422,
      "learning_rate": 0.0001949029126213592,
      "loss": 0.5981,
      "step": 572
    },
    {
      "epoch": 1.2515035538545654,
      "grad_norm": 0.07584258913993835,
      "learning_rate": 0.00019466019417475726,
      "loss": 0.6035,
      "step": 573
    },
    {
      "epoch": 1.2536905412793877,
      "grad_norm": 0.06942015886306763,
      "learning_rate": 0.00019441747572815533,
      "loss": 0.9066,
      "step": 574
    },
    {
      "epoch": 1.2558775287042099,
      "grad_norm": 0.188585102558136,
      "learning_rate": 0.00019417475728155338,
      "loss": 0.5576,
      "step": 575
    },
    {
      "epoch": 1.2580645161290323,
      "grad_norm": 0.06378776580095291,
      "learning_rate": 0.00019393203883495143,
      "loss": 0.7096,
      "step": 576
    },
    {
      "epoch": 1.2602515035538546,
      "grad_norm": 0.093607597053051,
      "learning_rate": 0.0001936893203883495,
      "loss": 0.6168,
      "step": 577
    },
    {
      "epoch": 1.2624384909786768,
      "grad_norm": 0.06908252835273743,
      "learning_rate": 0.00019344660194174754,
      "loss": 0.5409,
      "step": 578
    },
    {
      "epoch": 1.2646254784034991,
      "grad_norm": 0.07703085243701935,
      "learning_rate": 0.00019320388349514561,
      "loss": 0.592,
      "step": 579
    },
    {
      "epoch": 1.2668124658283215,
      "grad_norm": 0.068319171667099,
      "learning_rate": 0.0001929611650485437,
      "loss": 0.6522,
      "step": 580
    },
    {
      "epoch": 1.268999453253144,
      "grad_norm": 0.04304174333810806,
      "learning_rate": 0.00019271844660194171,
      "loss": 0.8218,
      "step": 581
    },
    {
      "epoch": 1.271186440677966,
      "grad_norm": 0.0900990441441536,
      "learning_rate": 0.0001924757281553398,
      "loss": 1.0345,
      "step": 582
    },
    {
      "epoch": 1.2733734281027884,
      "grad_norm": 0.04922797903418541,
      "learning_rate": 0.00019223300970873787,
      "loss": 0.6132,
      "step": 583
    },
    {
      "epoch": 1.2755604155276108,
      "grad_norm": 0.047148652374744415,
      "learning_rate": 0.0001919902912621359,
      "loss": 0.3725,
      "step": 584
    },
    {
      "epoch": 1.277747402952433,
      "grad_norm": 0.2023891806602478,
      "learning_rate": 0.00019174757281553397,
      "loss": 0.5327,
      "step": 585
    },
    {
      "epoch": 1.2799343903772553,
      "grad_norm": 0.11520843207836151,
      "learning_rate": 0.00019150485436893205,
      "loss": 0.8382,
      "step": 586
    },
    {
      "epoch": 1.2821213778020777,
      "grad_norm": 0.07389979809522629,
      "learning_rate": 0.00019126213592233007,
      "loss": 0.568,
      "step": 587
    },
    {
      "epoch": 1.2843083652269,
      "grad_norm": 0.055056143552064896,
      "learning_rate": 0.00019101941747572815,
      "loss": 0.6447,
      "step": 588
    },
    {
      "epoch": 1.2864953526517222,
      "grad_norm": 0.06612139940261841,
      "learning_rate": 0.0001907766990291262,
      "loss": 0.5929,
      "step": 589
    },
    {
      "epoch": 1.2886823400765446,
      "grad_norm": 0.0247971098870039,
      "learning_rate": 0.00019053398058252425,
      "loss": 0.5537,
      "step": 590
    },
    {
      "epoch": 1.2908693275013667,
      "grad_norm": 0.15111826360225677,
      "learning_rate": 0.00019029126213592233,
      "loss": 0.7372,
      "step": 591
    },
    {
      "epoch": 1.293056314926189,
      "grad_norm": 0.08081436157226562,
      "learning_rate": 0.00019004854368932038,
      "loss": 0.6935,
      "step": 592
    },
    {
      "epoch": 1.2952433023510115,
      "grad_norm": 0.10907871276140213,
      "learning_rate": 0.00018980582524271843,
      "loss": 0.6666,
      "step": 593
    },
    {
      "epoch": 1.2974302897758339,
      "grad_norm": 0.135869100689888,
      "learning_rate": 0.00018956310679611648,
      "loss": 0.6372,
      "step": 594
    },
    {
      "epoch": 1.2996172772006562,
      "grad_norm": 0.034416068345308304,
      "learning_rate": 0.00018932038834951456,
      "loss": 0.4763,
      "step": 595
    },
    {
      "epoch": 1.3018042646254784,
      "grad_norm": 0.12647610902786255,
      "learning_rate": 0.0001890776699029126,
      "loss": 0.8464,
      "step": 596
    },
    {
      "epoch": 1.3039912520503008,
      "grad_norm": 0.057269513607025146,
      "learning_rate": 0.00018883495145631066,
      "loss": 0.5798,
      "step": 597
    },
    {
      "epoch": 1.306178239475123,
      "grad_norm": 0.0600140281021595,
      "learning_rate": 0.00018859223300970874,
      "loss": 0.5804,
      "step": 598
    },
    {
      "epoch": 1.3083652268999453,
      "grad_norm": 0.19099771976470947,
      "learning_rate": 0.0001883495145631068,
      "loss": 0.685,
      "step": 599
    },
    {
      "epoch": 1.3105522143247677,
      "grad_norm": 0.050372760742902756,
      "learning_rate": 0.00018810679611650484,
      "loss": 0.5087,
      "step": 600
    },
    {
      "epoch": 1.31273920174959,
      "grad_norm": 0.11331050843000412,
      "learning_rate": 0.00018786407766990291,
      "loss": 0.5442,
      "step": 601
    },
    {
      "epoch": 1.3149261891744122,
      "grad_norm": 0.06795575469732285,
      "learning_rate": 0.00018762135922330094,
      "loss": 0.637,
      "step": 602
    },
    {
      "epoch": 1.3171131765992345,
      "grad_norm": 0.16096124053001404,
      "learning_rate": 0.00018737864077669902,
      "loss": 0.696,
      "step": 603
    },
    {
      "epoch": 1.319300164024057,
      "grad_norm": 0.06096538528800011,
      "learning_rate": 0.00018713592233009707,
      "loss": 0.6776,
      "step": 604
    },
    {
      "epoch": 1.321487151448879,
      "grad_norm": 0.18116861581802368,
      "learning_rate": 0.00018689320388349512,
      "loss": 0.6609,
      "step": 605
    },
    {
      "epoch": 1.3236741388737014,
      "grad_norm": 0.07999300211668015,
      "learning_rate": 0.0001866504854368932,
      "loss": 0.839,
      "step": 606
    },
    {
      "epoch": 1.3258611262985238,
      "grad_norm": 0.06887108087539673,
      "learning_rate": 0.00018640776699029122,
      "loss": 0.4534,
      "step": 607
    },
    {
      "epoch": 1.3280481137233462,
      "grad_norm": 0.06105407699942589,
      "learning_rate": 0.0001861650485436893,
      "loss": 0.5984,
      "step": 608
    },
    {
      "epoch": 1.3302351011481683,
      "grad_norm": 0.07335031777620316,
      "learning_rate": 0.00018592233009708737,
      "loss": 0.8253,
      "step": 609
    },
    {
      "epoch": 1.3324220885729907,
      "grad_norm": 0.0456896536052227,
      "learning_rate": 0.0001856796116504854,
      "loss": 0.7217,
      "step": 610
    },
    {
      "epoch": 1.334609075997813,
      "grad_norm": 0.3200141191482544,
      "learning_rate": 0.00018543689320388347,
      "loss": 0.7451,
      "step": 611
    },
    {
      "epoch": 1.3367960634226352,
      "grad_norm": 0.06651178002357483,
      "learning_rate": 0.00018519417475728155,
      "loss": 0.7326,
      "step": 612
    },
    {
      "epoch": 1.3389830508474576,
      "grad_norm": 0.042020298540592194,
      "learning_rate": 0.00018495145631067957,
      "loss": 0.4071,
      "step": 613
    },
    {
      "epoch": 1.34117003827228,
      "grad_norm": 0.07132259756326675,
      "learning_rate": 0.00018470873786407765,
      "loss": 0.7612,
      "step": 614
    },
    {
      "epoch": 1.3433570256971024,
      "grad_norm": 0.11206962168216705,
      "learning_rate": 0.00018446601941747573,
      "loss": 0.6456,
      "step": 615
    },
    {
      "epoch": 1.3455440131219245,
      "grad_norm": 0.05653239041566849,
      "learning_rate": 0.00018422330097087375,
      "loss": 0.8114,
      "step": 616
    },
    {
      "epoch": 1.3477310005467469,
      "grad_norm": 0.14098510146141052,
      "learning_rate": 0.00018398058252427183,
      "loss": 0.7555,
      "step": 617
    },
    {
      "epoch": 1.349917987971569,
      "grad_norm": 0.07498617470264435,
      "learning_rate": 0.00018373786407766988,
      "loss": 0.6413,
      "step": 618
    },
    {
      "epoch": 1.3521049753963914,
      "grad_norm": 0.06883811205625534,
      "learning_rate": 0.00018349514563106793,
      "loss": 0.5517,
      "step": 619
    },
    {
      "epoch": 1.3542919628212138,
      "grad_norm": 0.15621449053287506,
      "learning_rate": 0.000183252427184466,
      "loss": 0.7654,
      "step": 620
    },
    {
      "epoch": 1.3564789502460362,
      "grad_norm": 0.10155853629112244,
      "learning_rate": 0.00018300970873786406,
      "loss": 0.7674,
      "step": 621
    },
    {
      "epoch": 1.3586659376708585,
      "grad_norm": 0.11119658499956131,
      "learning_rate": 0.0001827669902912621,
      "loss": 0.8727,
      "step": 622
    },
    {
      "epoch": 1.3608529250956807,
      "grad_norm": 0.08053936064243317,
      "learning_rate": 0.00018252427184466016,
      "loss": 0.7164,
      "step": 623
    },
    {
      "epoch": 1.363039912520503,
      "grad_norm": 0.10677488148212433,
      "learning_rate": 0.00018228155339805824,
      "loss": 0.7041,
      "step": 624
    },
    {
      "epoch": 1.3652268999453252,
      "grad_norm": 0.0674259215593338,
      "learning_rate": 0.0001820388349514563,
      "loss": 0.4037,
      "step": 625
    },
    {
      "epoch": 1.3674138873701476,
      "grad_norm": 0.07473250478506088,
      "learning_rate": 0.00018179611650485434,
      "loss": 0.6386,
      "step": 626
    },
    {
      "epoch": 1.36960087479497,
      "grad_norm": 0.1293839067220688,
      "learning_rate": 0.00018155339805825242,
      "loss": 0.6953,
      "step": 627
    },
    {
      "epoch": 1.3717878622197923,
      "grad_norm": 0.11126802861690521,
      "learning_rate": 0.00018131067961165047,
      "loss": 0.4841,
      "step": 628
    },
    {
      "epoch": 1.3739748496446145,
      "grad_norm": 0.053122710436582565,
      "learning_rate": 0.00018106796116504852,
      "loss": 0.6054,
      "step": 629
    },
    {
      "epoch": 1.3761618370694368,
      "grad_norm": 0.10750580579042435,
      "learning_rate": 0.0001808252427184466,
      "loss": 0.6955,
      "step": 630
    },
    {
      "epoch": 1.3783488244942592,
      "grad_norm": 0.10743537545204163,
      "learning_rate": 0.00018058252427184462,
      "loss": 0.6787,
      "step": 631
    },
    {
      "epoch": 1.3805358119190814,
      "grad_norm": 0.04577861353754997,
      "learning_rate": 0.0001803398058252427,
      "loss": 0.6591,
      "step": 632
    },
    {
      "epoch": 1.3827227993439037,
      "grad_norm": 0.09178069978952408,
      "learning_rate": 0.00018009708737864077,
      "loss": 0.6936,
      "step": 633
    },
    {
      "epoch": 1.3849097867687261,
      "grad_norm": 0.1732717752456665,
      "learning_rate": 0.0001798543689320388,
      "loss": 0.7492,
      "step": 634
    },
    {
      "epoch": 1.3870967741935485,
      "grad_norm": 0.054739419370889664,
      "learning_rate": 0.00017961165048543688,
      "loss": 0.5586,
      "step": 635
    },
    {
      "epoch": 1.3892837616183706,
      "grad_norm": 0.11255872249603271,
      "learning_rate": 0.00017936893203883495,
      "loss": 0.636,
      "step": 636
    },
    {
      "epoch": 1.391470749043193,
      "grad_norm": 0.14757251739501953,
      "learning_rate": 0.00017912621359223298,
      "loss": 0.5243,
      "step": 637
    },
    {
      "epoch": 1.3936577364680154,
      "grad_norm": 0.044511694461107254,
      "learning_rate": 0.00017888349514563105,
      "loss": 0.4847,
      "step": 638
    },
    {
      "epoch": 1.3958447238928375,
      "grad_norm": 0.09428581595420837,
      "learning_rate": 0.00017864077669902913,
      "loss": 1.1749,
      "step": 639
    },
    {
      "epoch": 1.39803171131766,
      "grad_norm": 0.06652595102787018,
      "learning_rate": 0.00017839805825242716,
      "loss": 0.8209,
      "step": 640
    },
    {
      "epoch": 1.4002186987424823,
      "grad_norm": 0.087835893034935,
      "learning_rate": 0.00017815533980582523,
      "loss": 0.5784,
      "step": 641
    },
    {
      "epoch": 1.4024056861673047,
      "grad_norm": 0.09346538782119751,
      "learning_rate": 0.00017791262135922328,
      "loss": 0.5932,
      "step": 642
    },
    {
      "epoch": 1.4045926735921268,
      "grad_norm": 0.10662012547254562,
      "learning_rate": 0.00017766990291262133,
      "loss": 0.6372,
      "step": 643
    },
    {
      "epoch": 1.4067796610169492,
      "grad_norm": 0.11840318143367767,
      "learning_rate": 0.0001774271844660194,
      "loss": 0.6369,
      "step": 644
    },
    {
      "epoch": 1.4089666484417713,
      "grad_norm": 0.06892179697751999,
      "learning_rate": 0.00017718446601941746,
      "loss": 0.5671,
      "step": 645
    },
    {
      "epoch": 1.4111536358665937,
      "grad_norm": 0.10891327261924744,
      "learning_rate": 0.0001769417475728155,
      "loss": 0.9284,
      "step": 646
    },
    {
      "epoch": 1.413340623291416,
      "grad_norm": 0.09791342914104462,
      "learning_rate": 0.00017669902912621356,
      "loss": 0.7292,
      "step": 647
    },
    {
      "epoch": 1.4155276107162384,
      "grad_norm": 0.09937909245491028,
      "learning_rate": 0.00017645631067961164,
      "loss": 0.8155,
      "step": 648
    },
    {
      "epoch": 1.4177145981410608,
      "grad_norm": 0.09038559347391129,
      "learning_rate": 0.0001762135922330097,
      "loss": 0.4662,
      "step": 649
    },
    {
      "epoch": 1.419901585565883,
      "grad_norm": 0.058239519596099854,
      "learning_rate": 0.00017597087378640774,
      "loss": 0.6582,
      "step": 650
    },
    {
      "epoch": 1.4220885729907053,
      "grad_norm": 0.0631416067481041,
      "learning_rate": 0.00017572815533980582,
      "loss": 0.695,
      "step": 651
    },
    {
      "epoch": 1.4242755604155275,
      "grad_norm": 0.1288028061389923,
      "learning_rate": 0.00017548543689320387,
      "loss": 0.648,
      "step": 652
    },
    {
      "epoch": 1.4264625478403499,
      "grad_norm": 0.03047804720699787,
      "learning_rate": 0.00017524271844660192,
      "loss": 0.5446,
      "step": 653
    },
    {
      "epoch": 1.4286495352651722,
      "grad_norm": 0.060944609344005585,
      "learning_rate": 0.000175,
      "loss": 0.658,
      "step": 654
    },
    {
      "epoch": 1.4308365226899946,
      "grad_norm": 0.056785643100738525,
      "learning_rate": 0.00017475728155339802,
      "loss": 0.5258,
      "step": 655
    },
    {
      "epoch": 1.4330235101148168,
      "grad_norm": 0.058587562292814255,
      "learning_rate": 0.0001745145631067961,
      "loss": 0.6653,
      "step": 656
    },
    {
      "epoch": 1.4352104975396391,
      "grad_norm": 0.22061674296855927,
      "learning_rate": 0.00017427184466019418,
      "loss": 0.7656,
      "step": 657
    },
    {
      "epoch": 1.4373974849644615,
      "grad_norm": 0.06048043444752693,
      "learning_rate": 0.0001740291262135922,
      "loss": 0.7472,
      "step": 658
    },
    {
      "epoch": 1.4395844723892837,
      "grad_norm": 0.10482992976903915,
      "learning_rate": 0.00017378640776699028,
      "loss": 0.6674,
      "step": 659
    },
    {
      "epoch": 1.441771459814106,
      "grad_norm": 0.08470641821622849,
      "learning_rate": 0.00017354368932038836,
      "loss": 0.665,
      "step": 660
    },
    {
      "epoch": 1.4439584472389284,
      "grad_norm": 0.11567983031272888,
      "learning_rate": 0.00017330097087378638,
      "loss": 0.7653,
      "step": 661
    },
    {
      "epoch": 1.4461454346637508,
      "grad_norm": 0.09336879104375839,
      "learning_rate": 0.00017305825242718446,
      "loss": 0.897,
      "step": 662
    },
    {
      "epoch": 1.448332422088573,
      "grad_norm": 0.1675858199596405,
      "learning_rate": 0.00017281553398058253,
      "loss": 0.829,
      "step": 663
    },
    {
      "epoch": 1.4505194095133953,
      "grad_norm": 0.06564716249704361,
      "learning_rate": 0.00017257281553398056,
      "loss": 0.7922,
      "step": 664
    },
    {
      "epoch": 1.4527063969382177,
      "grad_norm": 0.10239650309085846,
      "learning_rate": 0.00017233009708737864,
      "loss": 0.6929,
      "step": 665
    },
    {
      "epoch": 1.4548933843630398,
      "grad_norm": 0.053650129586458206,
      "learning_rate": 0.00017208737864077669,
      "loss": 0.6141,
      "step": 666
    },
    {
      "epoch": 1.4570803717878622,
      "grad_norm": 0.07837307453155518,
      "learning_rate": 0.00017184466019417474,
      "loss": 0.8482,
      "step": 667
    },
    {
      "epoch": 1.4592673592126846,
      "grad_norm": 0.03846685588359833,
      "learning_rate": 0.00017160194174757281,
      "loss": 0.7327,
      "step": 668
    },
    {
      "epoch": 1.461454346637507,
      "grad_norm": 0.059876035898923874,
      "learning_rate": 0.00017135922330097086,
      "loss": 0.8028,
      "step": 669
    },
    {
      "epoch": 1.463641334062329,
      "grad_norm": 0.061472248286008835,
      "learning_rate": 0.00017111650485436891,
      "loss": 0.6939,
      "step": 670
    },
    {
      "epoch": 1.4658283214871515,
      "grad_norm": 0.12179437279701233,
      "learning_rate": 0.00017087378640776697,
      "loss": 0.6649,
      "step": 671
    },
    {
      "epoch": 1.4680153089119736,
      "grad_norm": 0.09186365455389023,
      "learning_rate": 0.00017063106796116504,
      "loss": 0.6011,
      "step": 672
    },
    {
      "epoch": 1.470202296336796,
      "grad_norm": 0.05658876150846481,
      "learning_rate": 0.0001703883495145631,
      "loss": 0.6425,
      "step": 673
    },
    {
      "epoch": 1.4723892837616184,
      "grad_norm": 0.08521269261837006,
      "learning_rate": 0.00017014563106796114,
      "loss": 0.6321,
      "step": 674
    },
    {
      "epoch": 1.4745762711864407,
      "grad_norm": 0.05281525105237961,
      "learning_rate": 0.00016990291262135922,
      "loss": 0.6789,
      "step": 675
    },
    {
      "epoch": 1.4767632586112631,
      "grad_norm": 0.08585121482610703,
      "learning_rate": 0.00016966019417475724,
      "loss": 0.6413,
      "step": 676
    },
    {
      "epoch": 1.4789502460360853,
      "grad_norm": 0.07132274657487869,
      "learning_rate": 0.00016941747572815532,
      "loss": 0.9009,
      "step": 677
    },
    {
      "epoch": 1.4811372334609076,
      "grad_norm": 0.06283555924892426,
      "learning_rate": 0.0001691747572815534,
      "loss": 0.5174,
      "step": 678
    },
    {
      "epoch": 1.4833242208857298,
      "grad_norm": 0.12625133991241455,
      "learning_rate": 0.00016893203883495142,
      "loss": 0.7434,
      "step": 679
    },
    {
      "epoch": 1.4855112083105522,
      "grad_norm": 0.08590378612279892,
      "learning_rate": 0.0001686893203883495,
      "loss": 0.6517,
      "step": 680
    },
    {
      "epoch": 1.4876981957353745,
      "grad_norm": 0.07998991757631302,
      "learning_rate": 0.00016844660194174758,
      "loss": 0.8598,
      "step": 681
    },
    {
      "epoch": 1.489885183160197,
      "grad_norm": 0.08209861814975739,
      "learning_rate": 0.0001682038834951456,
      "loss": 0.5312,
      "step": 682
    },
    {
      "epoch": 1.492072170585019,
      "grad_norm": 0.08130457997322083,
      "learning_rate": 0.00016796116504854368,
      "loss": 0.6749,
      "step": 683
    },
    {
      "epoch": 1.4942591580098414,
      "grad_norm": 0.06565511971712112,
      "learning_rate": 0.00016771844660194176,
      "loss": 0.5273,
      "step": 684
    },
    {
      "epoch": 1.4964461454346638,
      "grad_norm": 0.03930141031742096,
      "learning_rate": 0.00016747572815533978,
      "loss": 0.7349,
      "step": 685
    },
    {
      "epoch": 1.498633132859486,
      "grad_norm": 0.08142934739589691,
      "learning_rate": 0.00016723300970873786,
      "loss": 0.7705,
      "step": 686
    },
    {
      "epoch": 1.5008201202843083,
      "grad_norm": 0.11566407233476639,
      "learning_rate": 0.0001669902912621359,
      "loss": 0.6753,
      "step": 687
    },
    {
      "epoch": 1.5030071077091307,
      "grad_norm": 0.06479275226593018,
      "learning_rate": 0.00016674757281553396,
      "loss": 0.6477,
      "step": 688
    },
    {
      "epoch": 1.505194095133953,
      "grad_norm": 0.07941816747188568,
      "learning_rate": 0.00016650485436893204,
      "loss": 0.7045,
      "step": 689
    },
    {
      "epoch": 1.5073810825587752,
      "grad_norm": 0.12151027470827103,
      "learning_rate": 0.0001662621359223301,
      "loss": 0.6896,
      "step": 690
    },
    {
      "epoch": 1.5095680699835976,
      "grad_norm": 0.08807893842458725,
      "learning_rate": 0.00016601941747572814,
      "loss": 0.6048,
      "step": 691
    },
    {
      "epoch": 1.5117550574084198,
      "grad_norm": 0.03695622459053993,
      "learning_rate": 0.00016577669902912622,
      "loss": 0.6582,
      "step": 692
    },
    {
      "epoch": 1.5139420448332421,
      "grad_norm": 0.05755792185664177,
      "learning_rate": 0.00016553398058252427,
      "loss": 0.6326,
      "step": 693
    },
    {
      "epoch": 1.5161290322580645,
      "grad_norm": 0.1037437841296196,
      "learning_rate": 0.00016529126213592232,
      "loss": 0.6608,
      "step": 694
    },
    {
      "epoch": 1.5183160196828869,
      "grad_norm": 0.0674048662185669,
      "learning_rate": 0.00016504854368932037,
      "loss": 0.5077,
      "step": 695
    },
    {
      "epoch": 1.5205030071077092,
      "grad_norm": 0.1347750574350357,
      "learning_rate": 0.00016480582524271844,
      "loss": 0.6933,
      "step": 696
    },
    {
      "epoch": 1.5226899945325314,
      "grad_norm": 0.05024397000670433,
      "learning_rate": 0.0001645631067961165,
      "loss": 0.7558,
      "step": 697
    },
    {
      "epoch": 1.5248769819573538,
      "grad_norm": 0.0631837323307991,
      "learning_rate": 0.00016432038834951455,
      "loss": 0.5842,
      "step": 698
    },
    {
      "epoch": 1.527063969382176,
      "grad_norm": 0.09257268905639648,
      "learning_rate": 0.00016407766990291262,
      "loss": 0.6996,
      "step": 699
    },
    {
      "epoch": 1.5292509568069983,
      "grad_norm": 0.11887294054031372,
      "learning_rate": 0.00016383495145631065,
      "loss": 0.5101,
      "step": 700
    },
    {
      "epoch": 1.5314379442318207,
      "grad_norm": 0.07113930583000183,
      "learning_rate": 0.00016359223300970872,
      "loss": 0.7546,
      "step": 701
    },
    {
      "epoch": 1.533624931656643,
      "grad_norm": 0.10183428227901459,
      "learning_rate": 0.0001633495145631068,
      "loss": 0.6312,
      "step": 702
    },
    {
      "epoch": 1.5358119190814654,
      "grad_norm": 0.06275193393230438,
      "learning_rate": 0.00016310679611650483,
      "loss": 0.5942,
      "step": 703
    },
    {
      "epoch": 1.5379989065062876,
      "grad_norm": 0.06637417525053024,
      "learning_rate": 0.0001628640776699029,
      "loss": 0.6245,
      "step": 704
    },
    {
      "epoch": 1.54018589393111,
      "grad_norm": 0.11270612478256226,
      "learning_rate": 0.00016262135922330098,
      "loss": 0.9069,
      "step": 705
    },
    {
      "epoch": 1.542372881355932,
      "grad_norm": 0.07012913376092911,
      "learning_rate": 0.000162378640776699,
      "loss": 0.7436,
      "step": 706
    },
    {
      "epoch": 1.5445598687807545,
      "grad_norm": 0.07748250663280487,
      "learning_rate": 0.00016213592233009708,
      "loss": 0.6251,
      "step": 707
    },
    {
      "epoch": 1.5467468562055768,
      "grad_norm": 0.0590163953602314,
      "learning_rate": 0.0001618932038834951,
      "loss": 0.7188,
      "step": 708
    },
    {
      "epoch": 1.5489338436303992,
      "grad_norm": 0.042455244809389114,
      "learning_rate": 0.00016165048543689318,
      "loss": 0.6997,
      "step": 709
    },
    {
      "epoch": 1.5511208310552216,
      "grad_norm": 0.13864417374134064,
      "learning_rate": 0.00016140776699029126,
      "loss": 0.7744,
      "step": 710
    },
    {
      "epoch": 1.5533078184800437,
      "grad_norm": 0.06101079657673836,
      "learning_rate": 0.00016116504854368928,
      "loss": 0.658,
      "step": 711
    },
    {
      "epoch": 1.5554948059048659,
      "grad_norm": 0.07164673507213593,
      "learning_rate": 0.00016092233009708736,
      "loss": 0.5964,
      "step": 712
    },
    {
      "epoch": 1.5576817933296883,
      "grad_norm": 0.1606537252664566,
      "learning_rate": 0.00016067961165048544,
      "loss": 0.7858,
      "step": 713
    },
    {
      "epoch": 1.5598687807545106,
      "grad_norm": 0.06193002313375473,
      "learning_rate": 0.00016043689320388346,
      "loss": 0.9423,
      "step": 714
    },
    {
      "epoch": 1.562055768179333,
      "grad_norm": 0.11733701825141907,
      "learning_rate": 0.00016019417475728154,
      "loss": 0.6538,
      "step": 715
    },
    {
      "epoch": 1.5642427556041554,
      "grad_norm": 0.0617966391146183,
      "learning_rate": 0.00015995145631067962,
      "loss": 0.8811,
      "step": 716
    },
    {
      "epoch": 1.5664297430289775,
      "grad_norm": 0.045651547610759735,
      "learning_rate": 0.00015970873786407764,
      "loss": 0.4535,
      "step": 717
    },
    {
      "epoch": 1.5686167304538,
      "grad_norm": 0.05577806383371353,
      "learning_rate": 0.00015946601941747572,
      "loss": 0.8585,
      "step": 718
    },
    {
      "epoch": 1.570803717878622,
      "grad_norm": 0.04707050323486328,
      "learning_rate": 0.00015922330097087377,
      "loss": 0.6141,
      "step": 719
    },
    {
      "epoch": 1.5729907053034444,
      "grad_norm": 0.10123418271541595,
      "learning_rate": 0.00015898058252427182,
      "loss": 0.6322,
      "step": 720
    },
    {
      "epoch": 1.5751776927282668,
      "grad_norm": 0.058243077248334885,
      "learning_rate": 0.0001587378640776699,
      "loss": 0.4855,
      "step": 721
    },
    {
      "epoch": 1.5773646801530892,
      "grad_norm": 0.07926628738641739,
      "learning_rate": 0.00015849514563106795,
      "loss": 0.7209,
      "step": 722
    },
    {
      "epoch": 1.5795516675779115,
      "grad_norm": 0.053030792623758316,
      "learning_rate": 0.000158252427184466,
      "loss": 0.5231,
      "step": 723
    },
    {
      "epoch": 1.5817386550027337,
      "grad_norm": 0.10164280235767365,
      "learning_rate": 0.00015800970873786405,
      "loss": 0.6191,
      "step": 724
    },
    {
      "epoch": 1.583925642427556,
      "grad_norm": 0.080946184694767,
      "learning_rate": 0.00015776699029126213,
      "loss": 0.4119,
      "step": 725
    },
    {
      "epoch": 1.5861126298523782,
      "grad_norm": 0.10418912023305893,
      "learning_rate": 0.00015752427184466018,
      "loss": 0.6786,
      "step": 726
    },
    {
      "epoch": 1.5882996172772006,
      "grad_norm": 0.08983733505010605,
      "learning_rate": 0.00015728155339805823,
      "loss": 0.7682,
      "step": 727
    },
    {
      "epoch": 1.590486604702023,
      "grad_norm": 0.07781610637903214,
      "learning_rate": 0.0001570388349514563,
      "loss": 0.6476,
      "step": 728
    },
    {
      "epoch": 1.5926735921268453,
      "grad_norm": 0.048014696687459946,
      "learning_rate": 0.00015679611650485433,
      "loss": 0.5202,
      "step": 729
    },
    {
      "epoch": 1.5948605795516677,
      "grad_norm": 0.04506755992770195,
      "learning_rate": 0.0001565533980582524,
      "loss": 0.6491,
      "step": 730
    },
    {
      "epoch": 1.5970475669764899,
      "grad_norm": 0.0790894404053688,
      "learning_rate": 0.00015631067961165048,
      "loss": 0.8726,
      "step": 731
    },
    {
      "epoch": 1.5992345544013122,
      "grad_norm": 0.0461219921708107,
      "learning_rate": 0.0001560679611650485,
      "loss": 0.6627,
      "step": 732
    },
    {
      "epoch": 1.6014215418261344,
      "grad_norm": 0.055332399904727936,
      "learning_rate": 0.00015582524271844658,
      "loss": 0.6403,
      "step": 733
    },
    {
      "epoch": 1.6036085292509568,
      "grad_norm": 0.06580759584903717,
      "learning_rate": 0.00015558252427184466,
      "loss": 0.6241,
      "step": 734
    },
    {
      "epoch": 1.6057955166757791,
      "grad_norm": 0.05455014482140541,
      "learning_rate": 0.00015533980582524269,
      "loss": 0.5521,
      "step": 735
    },
    {
      "epoch": 1.6079825041006015,
      "grad_norm": 0.0879889652132988,
      "learning_rate": 0.00015509708737864076,
      "loss": 0.8286,
      "step": 736
    },
    {
      "epoch": 1.6101694915254239,
      "grad_norm": 0.07845278829336166,
      "learning_rate": 0.00015485436893203884,
      "loss": 0.6807,
      "step": 737
    },
    {
      "epoch": 1.612356478950246,
      "grad_norm": 0.1058611050248146,
      "learning_rate": 0.00015461165048543686,
      "loss": 0.7281,
      "step": 738
    },
    {
      "epoch": 1.6145434663750682,
      "grad_norm": 0.09188025444746017,
      "learning_rate": 0.00015436893203883494,
      "loss": 0.9997,
      "step": 739
    },
    {
      "epoch": 1.6167304537998906,
      "grad_norm": 0.07776019722223282,
      "learning_rate": 0.000154126213592233,
      "loss": 0.6304,
      "step": 740
    },
    {
      "epoch": 1.618917441224713,
      "grad_norm": 0.11982621252536774,
      "learning_rate": 0.00015388349514563104,
      "loss": 0.7336,
      "step": 741
    },
    {
      "epoch": 1.6211044286495353,
      "grad_norm": 0.13907882571220398,
      "learning_rate": 0.00015364077669902912,
      "loss": 0.7224,
      "step": 742
    },
    {
      "epoch": 1.6232914160743577,
      "grad_norm": 0.08288092911243439,
      "learning_rate": 0.00015339805825242717,
      "loss": 0.6346,
      "step": 743
    },
    {
      "epoch": 1.6254784034991798,
      "grad_norm": 0.07805322110652924,
      "learning_rate": 0.00015315533980582522,
      "loss": 0.7299,
      "step": 744
    },
    {
      "epoch": 1.6276653909240022,
      "grad_norm": 0.09408170729875565,
      "learning_rate": 0.0001529126213592233,
      "loss": 0.7052,
      "step": 745
    },
    {
      "epoch": 1.6298523783488243,
      "grad_norm": 0.05800126492977142,
      "learning_rate": 0.00015266990291262135,
      "loss": 0.7289,
      "step": 746
    },
    {
      "epoch": 1.6320393657736467,
      "grad_norm": 0.07356402277946472,
      "learning_rate": 0.0001524271844660194,
      "loss": 0.7533,
      "step": 747
    },
    {
      "epoch": 1.634226353198469,
      "grad_norm": 0.045158419758081436,
      "learning_rate": 0.00015218446601941745,
      "loss": 0.4564,
      "step": 748
    },
    {
      "epoch": 1.6364133406232915,
      "grad_norm": 0.05181990936398506,
      "learning_rate": 0.00015194174757281553,
      "loss": 0.5431,
      "step": 749
    },
    {
      "epoch": 1.6386003280481138,
      "grad_norm": 0.12281020730733871,
      "learning_rate": 0.00015169902912621358,
      "loss": 0.9977,
      "step": 750
    },
    {
      "epoch": 1.640787315472936,
      "grad_norm": 0.10029571503400803,
      "learning_rate": 0.00015145631067961163,
      "loss": 0.4878,
      "step": 751
    },
    {
      "epoch": 1.6429743028977584,
      "grad_norm": 0.27176743745803833,
      "learning_rate": 0.0001512135922330097,
      "loss": 0.4482,
      "step": 752
    },
    {
      "epoch": 1.6451612903225805,
      "grad_norm": 0.07689711451530457,
      "learning_rate": 0.00015097087378640773,
      "loss": 0.6518,
      "step": 753
    },
    {
      "epoch": 1.647348277747403,
      "grad_norm": 0.09448367357254028,
      "learning_rate": 0.0001507281553398058,
      "loss": 0.7245,
      "step": 754
    },
    {
      "epoch": 1.6495352651722253,
      "grad_norm": 0.09228754788637161,
      "learning_rate": 0.00015048543689320389,
      "loss": 0.6763,
      "step": 755
    },
    {
      "epoch": 1.6517222525970476,
      "grad_norm": 0.062209248542785645,
      "learning_rate": 0.0001502427184466019,
      "loss": 0.5551,
      "step": 756
    },
    {
      "epoch": 1.65390924002187,
      "grad_norm": 0.13062815368175507,
      "learning_rate": 0.00015,
      "loss": 0.8073,
      "step": 757
    },
    {
      "epoch": 1.6560962274466922,
      "grad_norm": 0.07445967942476273,
      "learning_rate": 0.00014975728155339804,
      "loss": 0.7943,
      "step": 758
    },
    {
      "epoch": 1.6582832148715145,
      "grad_norm": 0.04933764040470123,
      "learning_rate": 0.00014951456310679611,
      "loss": 0.5679,
      "step": 759
    },
    {
      "epoch": 1.6604702022963367,
      "grad_norm": 0.08024493604898453,
      "learning_rate": 0.00014927184466019417,
      "loss": 0.8256,
      "step": 760
    },
    {
      "epoch": 1.662657189721159,
      "grad_norm": 0.07997776567935944,
      "learning_rate": 0.00014902912621359222,
      "loss": 0.651,
      "step": 761
    },
    {
      "epoch": 1.6648441771459814,
      "grad_norm": 0.0992245227098465,
      "learning_rate": 0.0001487864077669903,
      "loss": 0.5129,
      "step": 762
    },
    {
      "epoch": 1.6670311645708038,
      "grad_norm": 0.09031302481889725,
      "learning_rate": 0.00014854368932038834,
      "loss": 0.5117,
      "step": 763
    },
    {
      "epoch": 1.6692181519956262,
      "grad_norm": 0.08211041986942291,
      "learning_rate": 0.0001483009708737864,
      "loss": 0.6673,
      "step": 764
    },
    {
      "epoch": 1.6714051394204483,
      "grad_norm": 0.07226278632879257,
      "learning_rate": 0.00014805825242718447,
      "loss": 1.2885,
      "step": 765
    },
    {
      "epoch": 1.6735921268452705,
      "grad_norm": 0.04525578394532204,
      "learning_rate": 0.00014781553398058252,
      "loss": 0.6017,
      "step": 766
    },
    {
      "epoch": 1.6757791142700929,
      "grad_norm": 0.08551250398159027,
      "learning_rate": 0.00014757281553398057,
      "loss": 0.5044,
      "step": 767
    },
    {
      "epoch": 1.6779661016949152,
      "grad_norm": 0.09782097488641739,
      "learning_rate": 0.00014733009708737862,
      "loss": 0.7231,
      "step": 768
    },
    {
      "epoch": 1.6801530891197376,
      "grad_norm": 0.08225250244140625,
      "learning_rate": 0.00014708737864077667,
      "loss": 0.706,
      "step": 769
    },
    {
      "epoch": 1.68234007654456,
      "grad_norm": 0.10632199048995972,
      "learning_rate": 0.00014684466019417475,
      "loss": 0.7661,
      "step": 770
    },
    {
      "epoch": 1.6845270639693821,
      "grad_norm": 0.0609552338719368,
      "learning_rate": 0.0001466019417475728,
      "loss": 0.5343,
      "step": 771
    },
    {
      "epoch": 1.6867140513942045,
      "grad_norm": 0.07229924201965332,
      "learning_rate": 0.00014635922330097085,
      "loss": 0.5003,
      "step": 772
    },
    {
      "epoch": 1.6889010388190266,
      "grad_norm": 0.10183531045913696,
      "learning_rate": 0.0001461165048543689,
      "loss": 0.7977,
      "step": 773
    },
    {
      "epoch": 1.691088026243849,
      "grad_norm": 0.05217605456709862,
      "learning_rate": 0.00014587378640776698,
      "loss": 0.6733,
      "step": 774
    },
    {
      "epoch": 1.6932750136686714,
      "grad_norm": 0.14263714849948883,
      "learning_rate": 0.00014563106796116503,
      "loss": 0.7619,
      "step": 775
    },
    {
      "epoch": 1.6954620010934938,
      "grad_norm": 0.07814032584428787,
      "learning_rate": 0.00014538834951456308,
      "loss": 0.7111,
      "step": 776
    },
    {
      "epoch": 1.6976489885183161,
      "grad_norm": 0.10231511294841766,
      "learning_rate": 0.00014514563106796116,
      "loss": 0.7288,
      "step": 777
    },
    {
      "epoch": 1.6998359759431383,
      "grad_norm": 0.08485434204339981,
      "learning_rate": 0.0001449029126213592,
      "loss": 0.6357,
      "step": 778
    },
    {
      "epoch": 1.7020229633679607,
      "grad_norm": 0.07320071011781693,
      "learning_rate": 0.00014466019417475726,
      "loss": 0.6858,
      "step": 779
    },
    {
      "epoch": 1.7042099507927828,
      "grad_norm": 0.059365514665842056,
      "learning_rate": 0.00014441747572815534,
      "loss": 0.5363,
      "step": 780
    },
    {
      "epoch": 1.7063969382176052,
      "grad_norm": 0.18387210369110107,
      "learning_rate": 0.0001441747572815534,
      "loss": 0.377,
      "step": 781
    },
    {
      "epoch": 1.7085839256424276,
      "grad_norm": 0.12494487315416336,
      "learning_rate": 0.00014393203883495144,
      "loss": 0.7818,
      "step": 782
    },
    {
      "epoch": 1.71077091306725,
      "grad_norm": 0.06817726045846939,
      "learning_rate": 0.0001436893203883495,
      "loss": 0.6386,
      "step": 783
    },
    {
      "epoch": 1.7129579004920723,
      "grad_norm": 0.24346807599067688,
      "learning_rate": 0.00014344660194174757,
      "loss": 0.6238,
      "step": 784
    },
    {
      "epoch": 1.7151448879168945,
      "grad_norm": 0.11059845983982086,
      "learning_rate": 0.00014320388349514562,
      "loss": 0.6998,
      "step": 785
    },
    {
      "epoch": 1.7173318753417168,
      "grad_norm": 0.08616719394922256,
      "learning_rate": 0.00014296116504854367,
      "loss": 0.6289,
      "step": 786
    },
    {
      "epoch": 1.719518862766539,
      "grad_norm": 0.08878134191036224,
      "learning_rate": 0.00014271844660194175,
      "loss": 0.4928,
      "step": 787
    },
    {
      "epoch": 1.7217058501913614,
      "grad_norm": 0.046061642467975616,
      "learning_rate": 0.0001424757281553398,
      "loss": 0.4056,
      "step": 788
    },
    {
      "epoch": 1.7238928376161837,
      "grad_norm": 0.06339546293020248,
      "learning_rate": 0.00014223300970873785,
      "loss": 0.9644,
      "step": 789
    },
    {
      "epoch": 1.726079825041006,
      "grad_norm": 0.08085539191961288,
      "learning_rate": 0.00014199029126213592,
      "loss": 0.589,
      "step": 790
    },
    {
      "epoch": 1.7282668124658285,
      "grad_norm": 0.09716881811618805,
      "learning_rate": 0.00014174757281553398,
      "loss": 0.7105,
      "step": 791
    },
    {
      "epoch": 1.7304537998906506,
      "grad_norm": 0.08055704832077026,
      "learning_rate": 0.00014150485436893203,
      "loss": 0.7049,
      "step": 792
    },
    {
      "epoch": 1.7326407873154728,
      "grad_norm": 0.033506009727716446,
      "learning_rate": 0.00014126213592233008,
      "loss": 0.55,
      "step": 793
    },
    {
      "epoch": 1.7348277747402951,
      "grad_norm": 0.08695266395807266,
      "learning_rate": 0.00014101941747572815,
      "loss": 0.8698,
      "step": 794
    },
    {
      "epoch": 1.7370147621651175,
      "grad_norm": 0.10492417961359024,
      "learning_rate": 0.0001407766990291262,
      "loss": 0.5841,
      "step": 795
    },
    {
      "epoch": 1.73920174958994,
      "grad_norm": 0.0260971337556839,
      "learning_rate": 0.00014053398058252425,
      "loss": 0.4715,
      "step": 796
    },
    {
      "epoch": 1.7413887370147623,
      "grad_norm": 0.14263994991779327,
      "learning_rate": 0.0001402912621359223,
      "loss": 0.8341,
      "step": 797
    },
    {
      "epoch": 1.7435757244395844,
      "grad_norm": 0.08548615872859955,
      "learning_rate": 0.00014004854368932036,
      "loss": 0.608,
      "step": 798
    },
    {
      "epoch": 1.7457627118644068,
      "grad_norm": 0.11541998386383057,
      "learning_rate": 0.00013980582524271843,
      "loss": 0.8728,
      "step": 799
    },
    {
      "epoch": 1.747949699289229,
      "grad_norm": 0.10113636404275894,
      "learning_rate": 0.00013956310679611648,
      "loss": 0.5703,
      "step": 800
    },
    {
      "epoch": 1.7501366867140513,
      "grad_norm": 0.09393016993999481,
      "learning_rate": 0.00013932038834951453,
      "loss": 0.6279,
      "step": 801
    },
    {
      "epoch": 1.7523236741388737,
      "grad_norm": 0.0811694860458374,
      "learning_rate": 0.0001390776699029126,
      "loss": 0.5184,
      "step": 802
    },
    {
      "epoch": 1.754510661563696,
      "grad_norm": 0.10583005845546722,
      "learning_rate": 0.00013883495145631066,
      "loss": 0.6799,
      "step": 803
    },
    {
      "epoch": 1.7566976489885184,
      "grad_norm": 0.09318964928388596,
      "learning_rate": 0.0001385922330097087,
      "loss": 0.8732,
      "step": 804
    },
    {
      "epoch": 1.7588846364133406,
      "grad_norm": 0.07826653122901917,
      "learning_rate": 0.0001383495145631068,
      "loss": 0.8013,
      "step": 805
    },
    {
      "epoch": 1.761071623838163,
      "grad_norm": 0.06191803142428398,
      "learning_rate": 0.00013810679611650484,
      "loss": 0.6102,
      "step": 806
    },
    {
      "epoch": 1.763258611262985,
      "grad_norm": 0.08127430081367493,
      "learning_rate": 0.0001378640776699029,
      "loss": 1.0328,
      "step": 807
    },
    {
      "epoch": 1.7654455986878075,
      "grad_norm": 0.046586234122514725,
      "learning_rate": 0.00013762135922330097,
      "loss": 0.8868,
      "step": 808
    },
    {
      "epoch": 1.7676325861126299,
      "grad_norm": 0.08679065853357315,
      "learning_rate": 0.00013737864077669902,
      "loss": 0.4166,
      "step": 809
    },
    {
      "epoch": 1.7698195735374522,
      "grad_norm": 0.11666709929704666,
      "learning_rate": 0.00013713592233009707,
      "loss": 0.7552,
      "step": 810
    },
    {
      "epoch": 1.7720065609622746,
      "grad_norm": 0.08766624331474304,
      "learning_rate": 0.00013689320388349515,
      "loss": 0.7544,
      "step": 811
    },
    {
      "epoch": 1.7741935483870968,
      "grad_norm": 0.04604002833366394,
      "learning_rate": 0.0001366504854368932,
      "loss": 0.6628,
      "step": 812
    },
    {
      "epoch": 1.7763805358119191,
      "grad_norm": 0.08578313142061234,
      "learning_rate": 0.00013640776699029125,
      "loss": 0.5542,
      "step": 813
    },
    {
      "epoch": 1.7785675232367413,
      "grad_norm": 0.07900494337081909,
      "learning_rate": 0.00013616504854368933,
      "loss": 0.8792,
      "step": 814
    },
    {
      "epoch": 1.7807545106615636,
      "grad_norm": 0.08219628036022186,
      "learning_rate": 0.00013592233009708738,
      "loss": 0.6798,
      "step": 815
    },
    {
      "epoch": 1.782941498086386,
      "grad_norm": 0.08442230522632599,
      "learning_rate": 0.00013567961165048543,
      "loss": 0.7634,
      "step": 816
    },
    {
      "epoch": 1.7851284855112084,
      "grad_norm": 0.12211817502975464,
      "learning_rate": 0.00013543689320388348,
      "loss": 0.5944,
      "step": 817
    },
    {
      "epoch": 1.7873154729360308,
      "grad_norm": 0.052023954689502716,
      "learning_rate": 0.00013519417475728156,
      "loss": 0.7413,
      "step": 818
    },
    {
      "epoch": 1.789502460360853,
      "grad_norm": 0.05588607117533684,
      "learning_rate": 0.0001349514563106796,
      "loss": 0.6026,
      "step": 819
    },
    {
      "epoch": 1.791689447785675,
      "grad_norm": 0.09938627481460571,
      "learning_rate": 0.00013470873786407766,
      "loss": 0.4907,
      "step": 820
    },
    {
      "epoch": 1.7938764352104974,
      "grad_norm": 0.07085568457841873,
      "learning_rate": 0.0001344660194174757,
      "loss": 0.7812,
      "step": 821
    },
    {
      "epoch": 1.7960634226353198,
      "grad_norm": 0.06302408128976822,
      "learning_rate": 0.00013422330097087376,
      "loss": 1.1212,
      "step": 822
    },
    {
      "epoch": 1.7982504100601422,
      "grad_norm": 0.07288993149995804,
      "learning_rate": 0.00013398058252427184,
      "loss": 0.3592,
      "step": 823
    },
    {
      "epoch": 1.8004373974849646,
      "grad_norm": 0.10824166238307953,
      "learning_rate": 0.00013373786407766989,
      "loss": 0.7594,
      "step": 824
    },
    {
      "epoch": 1.8026243849097867,
      "grad_norm": 0.06682860851287842,
      "learning_rate": 0.00013349514563106794,
      "loss": 0.7621,
      "step": 825
    },
    {
      "epoch": 1.804811372334609,
      "grad_norm": 0.058759741485118866,
      "learning_rate": 0.00013325242718446601,
      "loss": 0.5391,
      "step": 826
    },
    {
      "epoch": 1.8069983597594312,
      "grad_norm": 0.06587909162044525,
      "learning_rate": 0.00013300970873786406,
      "loss": 0.8572,
      "step": 827
    },
    {
      "epoch": 1.8091853471842536,
      "grad_norm": 0.08297627419233322,
      "learning_rate": 0.00013276699029126212,
      "loss": 0.5941,
      "step": 828
    },
    {
      "epoch": 1.811372334609076,
      "grad_norm": 0.06561224162578583,
      "learning_rate": 0.0001325242718446602,
      "loss": 0.6561,
      "step": 829
    },
    {
      "epoch": 1.8135593220338984,
      "grad_norm": 0.0836096704006195,
      "learning_rate": 0.00013228155339805824,
      "loss": 0.7184,
      "step": 830
    },
    {
      "epoch": 1.8157463094587207,
      "grad_norm": 0.059412457048892975,
      "learning_rate": 0.0001320388349514563,
      "loss": 0.7218,
      "step": 831
    },
    {
      "epoch": 1.8179332968835429,
      "grad_norm": 0.1909535527229309,
      "learning_rate": 0.00013179611650485437,
      "loss": 0.8877,
      "step": 832
    },
    {
      "epoch": 1.8201202843083653,
      "grad_norm": 0.12529511749744415,
      "learning_rate": 0.00013155339805825242,
      "loss": 1.0633,
      "step": 833
    },
    {
      "epoch": 1.8223072717331874,
      "grad_norm": 0.08508165180683136,
      "learning_rate": 0.00013131067961165047,
      "loss": 0.788,
      "step": 834
    },
    {
      "epoch": 1.8244942591580098,
      "grad_norm": 0.13272306323051453,
      "learning_rate": 0.00013106796116504852,
      "loss": 0.7453,
      "step": 835
    },
    {
      "epoch": 1.8266812465828322,
      "grad_norm": 0.08672458678483963,
      "learning_rate": 0.0001308252427184466,
      "loss": 0.5658,
      "step": 836
    },
    {
      "epoch": 1.8288682340076545,
      "grad_norm": 0.05675327405333519,
      "learning_rate": 0.00013058252427184465,
      "loss": 0.7884,
      "step": 837
    },
    {
      "epoch": 1.831055221432477,
      "grad_norm": 0.05675899609923363,
      "learning_rate": 0.0001303398058252427,
      "loss": 0.65,
      "step": 838
    },
    {
      "epoch": 1.833242208857299,
      "grad_norm": 0.07699860632419586,
      "learning_rate": 0.00013009708737864078,
      "loss": 0.7413,
      "step": 839
    },
    {
      "epoch": 1.8354291962821214,
      "grad_norm": 0.06006371229887009,
      "learning_rate": 0.00012985436893203883,
      "loss": 0.4697,
      "step": 840
    },
    {
      "epoch": 1.8376161837069436,
      "grad_norm": 0.13671620190143585,
      "learning_rate": 0.00012961165048543688,
      "loss": 0.5431,
      "step": 841
    },
    {
      "epoch": 1.839803171131766,
      "grad_norm": 0.07552420347929001,
      "learning_rate": 0.00012936893203883493,
      "loss": 0.6848,
      "step": 842
    },
    {
      "epoch": 1.8419901585565883,
      "grad_norm": 0.07906556129455566,
      "learning_rate": 0.000129126213592233,
      "loss": 0.7319,
      "step": 843
    },
    {
      "epoch": 1.8441771459814107,
      "grad_norm": 0.16908115148544312,
      "learning_rate": 0.00012888349514563106,
      "loss": 0.7233,
      "step": 844
    },
    {
      "epoch": 1.846364133406233,
      "grad_norm": 0.05891953036189079,
      "learning_rate": 0.0001286407766990291,
      "loss": 0.469,
      "step": 845
    },
    {
      "epoch": 1.8485511208310552,
      "grad_norm": 0.059181615710258484,
      "learning_rate": 0.00012839805825242716,
      "loss": 0.6132,
      "step": 846
    },
    {
      "epoch": 1.8507381082558774,
      "grad_norm": 0.04733738303184509,
      "learning_rate": 0.00012815533980582524,
      "loss": 0.5983,
      "step": 847
    },
    {
      "epoch": 1.8529250956806997,
      "grad_norm": 0.10546132922172546,
      "learning_rate": 0.0001279126213592233,
      "loss": 0.6189,
      "step": 848
    },
    {
      "epoch": 1.8551120831055221,
      "grad_norm": 0.04662831500172615,
      "learning_rate": 0.00012766990291262134,
      "loss": 0.5541,
      "step": 849
    },
    {
      "epoch": 1.8572990705303445,
      "grad_norm": 0.03423594310879707,
      "learning_rate": 0.0001274271844660194,
      "loss": 0.5169,
      "step": 850
    },
    {
      "epoch": 1.8594860579551669,
      "grad_norm": 0.07898862659931183,
      "learning_rate": 0.00012718446601941747,
      "loss": 0.6811,
      "step": 851
    },
    {
      "epoch": 1.861673045379989,
      "grad_norm": 0.06376733630895615,
      "learning_rate": 0.00012694174757281552,
      "loss": 0.6015,
      "step": 852
    },
    {
      "epoch": 1.8638600328048114,
      "grad_norm": 0.11817888915538788,
      "learning_rate": 0.00012669902912621357,
      "loss": 0.7551,
      "step": 853
    },
    {
      "epoch": 1.8660470202296335,
      "grad_norm": 0.04637736827135086,
      "learning_rate": 0.00012645631067961165,
      "loss": 0.6324,
      "step": 854
    },
    {
      "epoch": 1.868234007654456,
      "grad_norm": 0.11976910382509232,
      "learning_rate": 0.0001262135922330097,
      "loss": 0.8507,
      "step": 855
    },
    {
      "epoch": 1.8704209950792783,
      "grad_norm": 0.08364604413509369,
      "learning_rate": 0.00012597087378640775,
      "loss": 0.7253,
      "step": 856
    },
    {
      "epoch": 1.8726079825041007,
      "grad_norm": 0.08944633603096008,
      "learning_rate": 0.00012572815533980582,
      "loss": 0.6782,
      "step": 857
    },
    {
      "epoch": 1.874794969928923,
      "grad_norm": 0.14449813961982727,
      "learning_rate": 0.00012548543689320387,
      "loss": 0.8016,
      "step": 858
    },
    {
      "epoch": 1.8769819573537452,
      "grad_norm": 0.07508597522974014,
      "learning_rate": 0.00012524271844660192,
      "loss": 0.7011,
      "step": 859
    },
    {
      "epoch": 1.8791689447785676,
      "grad_norm": 0.07832931727170944,
      "learning_rate": 0.000125,
      "loss": 0.6775,
      "step": 860
    },
    {
      "epoch": 1.8813559322033897,
      "grad_norm": 0.05429183319211006,
      "learning_rate": 0.00012475728155339805,
      "loss": 0.8538,
      "step": 861
    },
    {
      "epoch": 1.883542919628212,
      "grad_norm": 0.05742572247982025,
      "learning_rate": 0.0001245145631067961,
      "loss": 0.5853,
      "step": 862
    },
    {
      "epoch": 1.8857299070530344,
      "grad_norm": 0.07888288795948029,
      "learning_rate": 0.00012427184466019418,
      "loss": 0.6769,
      "step": 863
    },
    {
      "epoch": 1.8879168944778568,
      "grad_norm": 0.1497015655040741,
      "learning_rate": 0.00012402912621359223,
      "loss": 0.769,
      "step": 864
    },
    {
      "epoch": 1.8901038819026792,
      "grad_norm": 0.05974101275205612,
      "learning_rate": 0.00012378640776699028,
      "loss": 0.6417,
      "step": 865
    },
    {
      "epoch": 1.8922908693275013,
      "grad_norm": 0.06933821737766266,
      "learning_rate": 0.00012354368932038833,
      "loss": 0.8024,
      "step": 866
    },
    {
      "epoch": 1.8944778567523237,
      "grad_norm": 0.08834031224250793,
      "learning_rate": 0.0001233009708737864,
      "loss": 0.5945,
      "step": 867
    },
    {
      "epoch": 1.8966648441771459,
      "grad_norm": 0.06061748042702675,
      "learning_rate": 0.00012305825242718446,
      "loss": 0.7633,
      "step": 868
    },
    {
      "epoch": 1.8988518316019682,
      "grad_norm": 0.04227694123983383,
      "learning_rate": 0.0001228155339805825,
      "loss": 0.5573,
      "step": 869
    },
    {
      "epoch": 1.9010388190267906,
      "grad_norm": 0.05616821348667145,
      "learning_rate": 0.00012257281553398056,
      "loss": 0.557,
      "step": 870
    },
    {
      "epoch": 1.903225806451613,
      "grad_norm": 0.06088731437921524,
      "learning_rate": 0.0001223300970873786,
      "loss": 0.6185,
      "step": 871
    },
    {
      "epoch": 1.9054127938764354,
      "grad_norm": 0.053946834057569504,
      "learning_rate": 0.0001220873786407767,
      "loss": 0.597,
      "step": 872
    },
    {
      "epoch": 1.9075997813012575,
      "grad_norm": 0.10907828062772751,
      "learning_rate": 0.00012184466019417475,
      "loss": 0.6573,
      "step": 873
    },
    {
      "epoch": 1.9097867687260797,
      "grad_norm": 0.18307583034038544,
      "learning_rate": 0.0001216019417475728,
      "loss": 0.6346,
      "step": 874
    },
    {
      "epoch": 1.911973756150902,
      "grad_norm": 0.09590666741132736,
      "learning_rate": 0.00012135922330097087,
      "loss": 0.7205,
      "step": 875
    },
    {
      "epoch": 1.9141607435757244,
      "grad_norm": 0.061872418969869614,
      "learning_rate": 0.00012111650485436892,
      "loss": 0.4657,
      "step": 876
    },
    {
      "epoch": 1.9163477310005468,
      "grad_norm": 0.056165486574172974,
      "learning_rate": 0.00012087378640776697,
      "loss": 0.8182,
      "step": 877
    },
    {
      "epoch": 1.9185347184253692,
      "grad_norm": 0.0575888454914093,
      "learning_rate": 0.00012063106796116505,
      "loss": 0.5488,
      "step": 878
    },
    {
      "epoch": 1.9207217058501913,
      "grad_norm": 0.04158012568950653,
      "learning_rate": 0.0001203883495145631,
      "loss": 0.5699,
      "step": 879
    },
    {
      "epoch": 1.9229086932750137,
      "grad_norm": 0.05806119740009308,
      "learning_rate": 0.00012014563106796115,
      "loss": 0.7537,
      "step": 880
    },
    {
      "epoch": 1.9250956806998358,
      "grad_norm": 0.059674229472875595,
      "learning_rate": 0.00011990291262135923,
      "loss": 0.7854,
      "step": 881
    },
    {
      "epoch": 1.9272826681246582,
      "grad_norm": 0.16574355959892273,
      "learning_rate": 0.00011966019417475728,
      "loss": 0.7689,
      "step": 882
    },
    {
      "epoch": 1.9294696555494806,
      "grad_norm": 0.12689882516860962,
      "learning_rate": 0.00011941747572815533,
      "loss": 0.6747,
      "step": 883
    },
    {
      "epoch": 1.931656642974303,
      "grad_norm": 0.0645318478345871,
      "learning_rate": 0.00011917475728155339,
      "loss": 0.6583,
      "step": 884
    },
    {
      "epoch": 1.9338436303991253,
      "grad_norm": 0.0930570513010025,
      "learning_rate": 0.00011893203883495146,
      "loss": 0.5295,
      "step": 885
    },
    {
      "epoch": 1.9360306178239475,
      "grad_norm": 0.16208307445049286,
      "learning_rate": 0.0001186893203883495,
      "loss": 0.8123,
      "step": 886
    },
    {
      "epoch": 1.9382176052487698,
      "grad_norm": 0.11295409500598907,
      "learning_rate": 0.00011844660194174756,
      "loss": 0.8346,
      "step": 887
    },
    {
      "epoch": 1.940404592673592,
      "grad_norm": 0.17686349153518677,
      "learning_rate": 0.00011820388349514562,
      "loss": 0.5105,
      "step": 888
    },
    {
      "epoch": 1.9425915800984144,
      "grad_norm": 0.045305948704481125,
      "learning_rate": 0.00011796116504854367,
      "loss": 0.6025,
      "step": 889
    },
    {
      "epoch": 1.9447785675232367,
      "grad_norm": 0.26349976658821106,
      "learning_rate": 0.00011771844660194173,
      "loss": 0.4674,
      "step": 890
    },
    {
      "epoch": 1.9469655549480591,
      "grad_norm": 0.08130199462175369,
      "learning_rate": 0.0001174757281553398,
      "loss": 0.6241,
      "step": 891
    },
    {
      "epoch": 1.9491525423728815,
      "grad_norm": 0.12196088582277298,
      "learning_rate": 0.00011723300970873785,
      "loss": 0.6839,
      "step": 892
    },
    {
      "epoch": 1.9513395297977036,
      "grad_norm": 0.08335163444280624,
      "learning_rate": 0.0001169902912621359,
      "loss": 0.6898,
      "step": 893
    },
    {
      "epoch": 1.953526517222526,
      "grad_norm": 0.06309807300567627,
      "learning_rate": 0.00011674757281553398,
      "loss": 0.5075,
      "step": 894
    },
    {
      "epoch": 1.9557135046473482,
      "grad_norm": 0.09318676590919495,
      "learning_rate": 0.00011650485436893203,
      "loss": 0.6401,
      "step": 895
    },
    {
      "epoch": 1.9579004920721705,
      "grad_norm": 0.0993434339761734,
      "learning_rate": 0.00011626213592233008,
      "loss": 0.5846,
      "step": 896
    },
    {
      "epoch": 1.960087479496993,
      "grad_norm": 0.06654249131679535,
      "learning_rate": 0.00011601941747572814,
      "loss": 0.4916,
      "step": 897
    },
    {
      "epoch": 1.9622744669218153,
      "grad_norm": 0.06536954641342163,
      "learning_rate": 0.0001157766990291262,
      "loss": 0.7887,
      "step": 898
    },
    {
      "epoch": 1.9644614543466377,
      "grad_norm": 0.08677750080823898,
      "learning_rate": 0.00011553398058252426,
      "loss": 0.4943,
      "step": 899
    },
    {
      "epoch": 1.9666484417714598,
      "grad_norm": 0.09694842249155045,
      "learning_rate": 0.00011529126213592232,
      "loss": 0.6401,
      "step": 900
    },
    {
      "epoch": 1.968835429196282,
      "grad_norm": 0.045057326555252075,
      "learning_rate": 0.00011504854368932037,
      "loss": 0.7471,
      "step": 901
    },
    {
      "epoch": 1.9710224166211043,
      "grad_norm": 0.13406231999397278,
      "learning_rate": 0.00011480582524271844,
      "loss": 0.6352,
      "step": 902
    },
    {
      "epoch": 1.9732094040459267,
      "grad_norm": 0.07525835931301117,
      "learning_rate": 0.0001145631067961165,
      "loss": 0.6649,
      "step": 903
    },
    {
      "epoch": 1.975396391470749,
      "grad_norm": 0.08038943260908127,
      "learning_rate": 0.00011432038834951455,
      "loss": 0.5342,
      "step": 904
    },
    {
      "epoch": 1.9775833788955715,
      "grad_norm": 0.11779956519603729,
      "learning_rate": 0.0001140776699029126,
      "loss": 0.7336,
      "step": 905
    },
    {
      "epoch": 1.9797703663203936,
      "grad_norm": 0.1321367770433426,
      "learning_rate": 0.00011383495145631068,
      "loss": 0.7451,
      "step": 906
    },
    {
      "epoch": 1.981957353745216,
      "grad_norm": 0.12764184176921844,
      "learning_rate": 0.00011359223300970873,
      "loss": 0.7046,
      "step": 907
    },
    {
      "epoch": 1.9841443411700381,
      "grad_norm": 0.04532116651535034,
      "learning_rate": 0.00011334951456310678,
      "loss": 0.7603,
      "step": 908
    },
    {
      "epoch": 1.9863313285948605,
      "grad_norm": 0.043816618621349335,
      "learning_rate": 0.00011310679611650484,
      "loss": 0.8493,
      "step": 909
    },
    {
      "epoch": 1.9885183160196829,
      "grad_norm": 0.07102162390947342,
      "learning_rate": 0.00011286407766990291,
      "loss": 0.6751,
      "step": 910
    },
    {
      "epoch": 1.9907053034445052,
      "grad_norm": 0.07698041200637817,
      "learning_rate": 0.00011262135922330096,
      "loss": 0.8755,
      "step": 911
    },
    {
      "epoch": 1.9928922908693276,
      "grad_norm": 0.07542385905981064,
      "learning_rate": 0.00011237864077669902,
      "loss": 0.6766,
      "step": 912
    },
    {
      "epoch": 1.9950792782941498,
      "grad_norm": 0.08312849700450897,
      "learning_rate": 0.00011213592233009707,
      "loss": 0.665,
      "step": 913
    },
    {
      "epoch": 1.9972662657189721,
      "grad_norm": 0.08149036020040512,
      "learning_rate": 0.00011189320388349514,
      "loss": 0.5601,
      "step": 914
    },
    {
      "epoch": 1.9994532531437943,
      "grad_norm": 0.08313512802124023,
      "learning_rate": 0.0001116504854368932,
      "loss": 0.7797,
      "step": 915
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.2948603332042694,
      "learning_rate": 0.00011140776699029125,
      "loss": 0.7282,
      "step": 916
    },
    {
      "epoch": 2.0021869874248224,
      "grad_norm": 0.05372385308146477,
      "learning_rate": 0.0001111650485436893,
      "loss": 0.5473,
      "step": 917
    },
    {
      "epoch": 2.0043739748496447,
      "grad_norm": 0.09125916659832001,
      "learning_rate": 0.00011092233009708738,
      "loss": 0.5244,
      "step": 918
    },
    {
      "epoch": 2.006560962274467,
      "grad_norm": 0.10436462610960007,
      "learning_rate": 0.00011067961165048543,
      "loss": 0.6836,
      "step": 919
    },
    {
      "epoch": 2.008747949699289,
      "grad_norm": 0.0616326779127121,
      "learning_rate": 0.00011043689320388348,
      "loss": 0.5314,
      "step": 920
    },
    {
      "epoch": 2.0109349371241114,
      "grad_norm": 0.0691252276301384,
      "learning_rate": 0.00011019417475728154,
      "loss": 0.4933,
      "step": 921
    },
    {
      "epoch": 2.013121924548934,
      "grad_norm": 0.06891285628080368,
      "learning_rate": 0.00010995145631067961,
      "loss": 0.577,
      "step": 922
    },
    {
      "epoch": 2.015308911973756,
      "grad_norm": 0.07588429003953934,
      "learning_rate": 0.00010970873786407766,
      "loss": 0.78,
      "step": 923
    },
    {
      "epoch": 2.0174958993985785,
      "grad_norm": 0.05341513082385063,
      "learning_rate": 0.00010946601941747572,
      "loss": 0.6743,
      "step": 924
    },
    {
      "epoch": 2.019682886823401,
      "grad_norm": 0.060330722481012344,
      "learning_rate": 0.00010922330097087377,
      "loss": 0.6878,
      "step": 925
    },
    {
      "epoch": 2.0218698742482233,
      "grad_norm": 0.04241131246089935,
      "learning_rate": 0.00010898058252427184,
      "loss": 0.5132,
      "step": 926
    },
    {
      "epoch": 2.024056861673045,
      "grad_norm": 0.06371448934078217,
      "learning_rate": 0.0001087378640776699,
      "loss": 0.3898,
      "step": 927
    },
    {
      "epoch": 2.0262438490978676,
      "grad_norm": 0.08609224110841751,
      "learning_rate": 0.00010849514563106795,
      "loss": 0.6915,
      "step": 928
    },
    {
      "epoch": 2.02843083652269,
      "grad_norm": 0.08693238347768784,
      "learning_rate": 0.000108252427184466,
      "loss": 0.7032,
      "step": 929
    },
    {
      "epoch": 2.0306178239475123,
      "grad_norm": 0.10312940180301666,
      "learning_rate": 0.00010800970873786408,
      "loss": 0.7286,
      "step": 930
    },
    {
      "epoch": 2.0328048113723347,
      "grad_norm": 0.03645063191652298,
      "learning_rate": 0.00010776699029126213,
      "loss": 0.5828,
      "step": 931
    },
    {
      "epoch": 2.034991798797157,
      "grad_norm": 0.12635767459869385,
      "learning_rate": 0.00010752427184466018,
      "loss": 0.6497,
      "step": 932
    },
    {
      "epoch": 2.037178786221979,
      "grad_norm": 0.04273942857980728,
      "learning_rate": 0.00010728155339805825,
      "loss": 0.9746,
      "step": 933
    },
    {
      "epoch": 2.0393657736468014,
      "grad_norm": 0.07975783199071884,
      "learning_rate": 0.00010703883495145631,
      "loss": 0.7937,
      "step": 934
    },
    {
      "epoch": 2.0415527610716238,
      "grad_norm": 0.05579470098018646,
      "learning_rate": 0.00010679611650485436,
      "loss": 0.3576,
      "step": 935
    },
    {
      "epoch": 2.043739748496446,
      "grad_norm": 0.08263109624385834,
      "learning_rate": 0.00010655339805825242,
      "loss": 1.3163,
      "step": 936
    },
    {
      "epoch": 2.0459267359212685,
      "grad_norm": 0.05702097713947296,
      "learning_rate": 0.00010631067961165047,
      "loss": 0.6933,
      "step": 937
    },
    {
      "epoch": 2.048113723346091,
      "grad_norm": 0.08686594665050507,
      "learning_rate": 0.00010606796116504853,
      "loss": 0.7076,
      "step": 938
    },
    {
      "epoch": 2.0503007107709132,
      "grad_norm": 0.06772071868181229,
      "learning_rate": 0.00010582524271844659,
      "loss": 0.8234,
      "step": 939
    },
    {
      "epoch": 2.052487698195735,
      "grad_norm": 0.09771222621202469,
      "learning_rate": 0.00010558252427184465,
      "loss": 0.6862,
      "step": 940
    },
    {
      "epoch": 2.0546746856205576,
      "grad_norm": 0.07454249262809753,
      "learning_rate": 0.0001053398058252427,
      "loss": 0.7456,
      "step": 941
    },
    {
      "epoch": 2.05686167304538,
      "grad_norm": 0.15167352557182312,
      "learning_rate": 0.00010509708737864075,
      "loss": 0.6754,
      "step": 942
    },
    {
      "epoch": 2.0590486604702023,
      "grad_norm": 0.06704775243997574,
      "learning_rate": 0.00010485436893203883,
      "loss": 0.6393,
      "step": 943
    },
    {
      "epoch": 2.0612356478950247,
      "grad_norm": 0.08005604147911072,
      "learning_rate": 0.00010461165048543688,
      "loss": 0.6547,
      "step": 944
    },
    {
      "epoch": 2.063422635319847,
      "grad_norm": 0.09053482115268707,
      "learning_rate": 0.00010436893203883493,
      "loss": 0.6647,
      "step": 945
    },
    {
      "epoch": 2.0656096227446694,
      "grad_norm": 0.040814295411109924,
      "learning_rate": 0.00010412621359223301,
      "loss": 0.5523,
      "step": 946
    },
    {
      "epoch": 2.0677966101694913,
      "grad_norm": 0.16581493616104126,
      "learning_rate": 0.00010388349514563106,
      "loss": 0.6994,
      "step": 947
    },
    {
      "epoch": 2.0699835975943137,
      "grad_norm": 0.0843871682882309,
      "learning_rate": 0.00010364077669902911,
      "loss": 0.6573,
      "step": 948
    },
    {
      "epoch": 2.072170585019136,
      "grad_norm": 0.32330793142318726,
      "learning_rate": 0.00010339805825242718,
      "loss": 0.5952,
      "step": 949
    },
    {
      "epoch": 2.0743575724439585,
      "grad_norm": 0.06891205906867981,
      "learning_rate": 0.00010315533980582523,
      "loss": 0.5727,
      "step": 950
    },
    {
      "epoch": 2.076544559868781,
      "grad_norm": 0.14751192927360535,
      "learning_rate": 0.00010291262135922329,
      "loss": 0.655,
      "step": 951
    },
    {
      "epoch": 2.078731547293603,
      "grad_norm": 0.12010716646909714,
      "learning_rate": 0.00010266990291262135,
      "loss": 0.7208,
      "step": 952
    },
    {
      "epoch": 2.0809185347184256,
      "grad_norm": 0.056473225355148315,
      "learning_rate": 0.0001024271844660194,
      "loss": 0.4184,
      "step": 953
    },
    {
      "epoch": 2.0831055221432475,
      "grad_norm": 0.05701868236064911,
      "learning_rate": 0.00010218446601941746,
      "loss": 0.5916,
      "step": 954
    },
    {
      "epoch": 2.08529250956807,
      "grad_norm": 0.1446901112794876,
      "learning_rate": 0.00010194174757281553,
      "loss": 0.8088,
      "step": 955
    },
    {
      "epoch": 2.0874794969928923,
      "grad_norm": 0.12073302268981934,
      "learning_rate": 0.00010169902912621358,
      "loss": 0.6987,
      "step": 956
    },
    {
      "epoch": 2.0896664844177146,
      "grad_norm": 0.10358598083257675,
      "learning_rate": 0.00010145631067961163,
      "loss": 0.5885,
      "step": 957
    },
    {
      "epoch": 2.091853471842537,
      "grad_norm": 0.04258884862065315,
      "learning_rate": 0.00010121359223300971,
      "loss": 0.4822,
      "step": 958
    },
    {
      "epoch": 2.0940404592673594,
      "grad_norm": 0.08908698707818985,
      "learning_rate": 0.00010097087378640776,
      "loss": 0.6025,
      "step": 959
    },
    {
      "epoch": 2.0962274466921813,
      "grad_norm": 0.06096798926591873,
      "learning_rate": 0.00010072815533980581,
      "loss": 0.6388,
      "step": 960
    },
    {
      "epoch": 2.0984144341170037,
      "grad_norm": 0.05718372389674187,
      "learning_rate": 0.00010048543689320388,
      "loss": 0.3815,
      "step": 961
    },
    {
      "epoch": 2.100601421541826,
      "grad_norm": 0.08511919528245926,
      "learning_rate": 0.00010024271844660193,
      "loss": 0.6558,
      "step": 962
    },
    {
      "epoch": 2.1027884089666484,
      "grad_norm": 0.08019436150789261,
      "learning_rate": 9.999999999999999e-05,
      "loss": 0.6916,
      "step": 963
    },
    {
      "epoch": 2.104975396391471,
      "grad_norm": 0.06560107320547104,
      "learning_rate": 9.975728155339806e-05,
      "loss": 0.5873,
      "step": 964
    },
    {
      "epoch": 2.107162383816293,
      "grad_norm": 0.14055019617080688,
      "learning_rate": 9.95145631067961e-05,
      "loss": 0.5676,
      "step": 965
    },
    {
      "epoch": 2.1093493712411155,
      "grad_norm": 0.1176968365907669,
      "learning_rate": 9.927184466019416e-05,
      "loss": 0.5483,
      "step": 966
    },
    {
      "epoch": 2.1115363586659375,
      "grad_norm": 0.07665383070707321,
      "learning_rate": 9.902912621359223e-05,
      "loss": 0.5784,
      "step": 967
    },
    {
      "epoch": 2.11372334609076,
      "grad_norm": 0.1384451985359192,
      "learning_rate": 9.878640776699028e-05,
      "loss": 0.6778,
      "step": 968
    },
    {
      "epoch": 2.115910333515582,
      "grad_norm": 0.09405054897069931,
      "learning_rate": 9.854368932038833e-05,
      "loss": 0.7837,
      "step": 969
    },
    {
      "epoch": 2.1180973209404046,
      "grad_norm": 0.06697643548250198,
      "learning_rate": 9.83009708737864e-05,
      "loss": 0.5959,
      "step": 970
    },
    {
      "epoch": 2.120284308365227,
      "grad_norm": 0.09054021537303925,
      "learning_rate": 9.805825242718446e-05,
      "loss": 0.8132,
      "step": 971
    },
    {
      "epoch": 2.1224712957900493,
      "grad_norm": 0.1512996107339859,
      "learning_rate": 9.781553398058251e-05,
      "loss": 0.8451,
      "step": 972
    },
    {
      "epoch": 2.1246582832148717,
      "grad_norm": 0.10180766880512238,
      "learning_rate": 9.757281553398058e-05,
      "loss": 0.949,
      "step": 973
    },
    {
      "epoch": 2.1268452706396936,
      "grad_norm": 0.0628434494137764,
      "learning_rate": 9.733009708737863e-05,
      "loss": 0.6542,
      "step": 974
    },
    {
      "epoch": 2.129032258064516,
      "grad_norm": 0.07519734650850296,
      "learning_rate": 9.708737864077669e-05,
      "loss": 0.7679,
      "step": 975
    },
    {
      "epoch": 2.1312192454893384,
      "grad_norm": 0.07029229402542114,
      "learning_rate": 9.684466019417476e-05,
      "loss": 0.5905,
      "step": 976
    },
    {
      "epoch": 2.1334062329141608,
      "grad_norm": 0.06242423877120018,
      "learning_rate": 9.660194174757281e-05,
      "loss": 0.4672,
      "step": 977
    },
    {
      "epoch": 2.135593220338983,
      "grad_norm": 0.044176701456308365,
      "learning_rate": 9.635922330097086e-05,
      "loss": 0.6701,
      "step": 978
    },
    {
      "epoch": 2.1377802077638055,
      "grad_norm": 0.09683780372142792,
      "learning_rate": 9.611650485436893e-05,
      "loss": 0.8281,
      "step": 979
    },
    {
      "epoch": 2.1399671951886274,
      "grad_norm": 0.08462493866682053,
      "learning_rate": 9.587378640776699e-05,
      "loss": 0.569,
      "step": 980
    },
    {
      "epoch": 2.14215418261345,
      "grad_norm": 0.10928475111722946,
      "learning_rate": 9.563106796116504e-05,
      "loss": 0.7517,
      "step": 981
    },
    {
      "epoch": 2.144341170038272,
      "grad_norm": 0.045466918498277664,
      "learning_rate": 9.53883495145631e-05,
      "loss": 0.6727,
      "step": 982
    },
    {
      "epoch": 2.1465281574630946,
      "grad_norm": 0.10792270302772522,
      "learning_rate": 9.514563106796116e-05,
      "loss": 0.6238,
      "step": 983
    },
    {
      "epoch": 2.148715144887917,
      "grad_norm": 0.09913436323404312,
      "learning_rate": 9.490291262135921e-05,
      "loss": 0.6343,
      "step": 984
    },
    {
      "epoch": 2.1509021323127393,
      "grad_norm": 0.06535235792398453,
      "learning_rate": 9.466019417475728e-05,
      "loss": 0.449,
      "step": 985
    },
    {
      "epoch": 2.1530891197375617,
      "grad_norm": 0.09585700929164886,
      "learning_rate": 9.441747572815533e-05,
      "loss": 0.6492,
      "step": 986
    },
    {
      "epoch": 2.1552761071623836,
      "grad_norm": 0.05615914240479469,
      "learning_rate": 9.41747572815534e-05,
      "loss": 0.7197,
      "step": 987
    },
    {
      "epoch": 2.157463094587206,
      "grad_norm": 0.04286925494670868,
      "learning_rate": 9.393203883495146e-05,
      "loss": 0.7092,
      "step": 988
    },
    {
      "epoch": 2.1596500820120283,
      "grad_norm": 0.19396215677261353,
      "learning_rate": 9.368932038834951e-05,
      "loss": 0.6476,
      "step": 989
    },
    {
      "epoch": 2.1618370694368507,
      "grad_norm": 0.05644046515226364,
      "learning_rate": 9.344660194174756e-05,
      "loss": 0.5788,
      "step": 990
    },
    {
      "epoch": 2.164024056861673,
      "grad_norm": 0.10071127116680145,
      "learning_rate": 9.320388349514561e-05,
      "loss": 0.7672,
      "step": 991
    },
    {
      "epoch": 2.1662110442864955,
      "grad_norm": 0.2811974883079529,
      "learning_rate": 9.296116504854369e-05,
      "loss": 0.6406,
      "step": 992
    },
    {
      "epoch": 2.168398031711318,
      "grad_norm": 0.050698939710855484,
      "learning_rate": 9.271844660194174e-05,
      "loss": 0.5947,
      "step": 993
    },
    {
      "epoch": 2.1705850191361398,
      "grad_norm": 0.06642589718103409,
      "learning_rate": 9.247572815533979e-05,
      "loss": 0.6463,
      "step": 994
    },
    {
      "epoch": 2.172772006560962,
      "grad_norm": 0.10926207154989243,
      "learning_rate": 9.223300970873787e-05,
      "loss": 0.6987,
      "step": 995
    },
    {
      "epoch": 2.1749589939857845,
      "grad_norm": 0.0996134802699089,
      "learning_rate": 9.199029126213592e-05,
      "loss": 0.5775,
      "step": 996
    },
    {
      "epoch": 2.177145981410607,
      "grad_norm": 0.06016650050878525,
      "learning_rate": 9.174757281553397e-05,
      "loss": 0.7562,
      "step": 997
    },
    {
      "epoch": 2.1793329688354293,
      "grad_norm": 0.07599005103111267,
      "learning_rate": 9.150485436893203e-05,
      "loss": 0.5302,
      "step": 998
    },
    {
      "epoch": 2.1815199562602516,
      "grad_norm": 0.13693709671497345,
      "learning_rate": 9.126213592233008e-05,
      "loss": 0.5294,
      "step": 999
    },
    {
      "epoch": 2.183706943685074,
      "grad_norm": 0.15747599303722382,
      "learning_rate": 9.101941747572814e-05,
      "loss": 0.5905,
      "step": 1000
    },
    {
      "epoch": 2.185893931109896,
      "grad_norm": 0.0751093253493309,
      "learning_rate": 9.077669902912621e-05,
      "loss": 0.5914,
      "step": 1001
    },
    {
      "epoch": 2.1880809185347183,
      "grad_norm": 0.06828130036592484,
      "learning_rate": 9.053398058252426e-05,
      "loss": 0.5183,
      "step": 1002
    },
    {
      "epoch": 2.1902679059595407,
      "grad_norm": 0.0638083666563034,
      "learning_rate": 9.029126213592231e-05,
      "loss": 0.6065,
      "step": 1003
    },
    {
      "epoch": 2.192454893384363,
      "grad_norm": 0.10457223653793335,
      "learning_rate": 9.004854368932039e-05,
      "loss": 0.5831,
      "step": 1004
    },
    {
      "epoch": 2.1946418808091854,
      "grad_norm": 0.08131687343120575,
      "learning_rate": 8.980582524271844e-05,
      "loss": 0.6268,
      "step": 1005
    },
    {
      "epoch": 2.196828868234008,
      "grad_norm": 0.06330417096614838,
      "learning_rate": 8.956310679611649e-05,
      "loss": 0.5261,
      "step": 1006
    },
    {
      "epoch": 2.19901585565883,
      "grad_norm": 0.056995972990989685,
      "learning_rate": 8.932038834951457e-05,
      "loss": 0.5737,
      "step": 1007
    },
    {
      "epoch": 2.201202843083652,
      "grad_norm": 0.10710989683866501,
      "learning_rate": 8.907766990291262e-05,
      "loss": 0.5823,
      "step": 1008
    },
    {
      "epoch": 2.2033898305084745,
      "grad_norm": 0.08252164721488953,
      "learning_rate": 8.883495145631067e-05,
      "loss": 0.5026,
      "step": 1009
    },
    {
      "epoch": 2.205576817933297,
      "grad_norm": 0.09622621536254883,
      "learning_rate": 8.859223300970873e-05,
      "loss": 0.6359,
      "step": 1010
    },
    {
      "epoch": 2.2077638053581192,
      "grad_norm": 0.15953707695007324,
      "learning_rate": 8.834951456310678e-05,
      "loss": 0.8133,
      "step": 1011
    },
    {
      "epoch": 2.2099507927829416,
      "grad_norm": 0.02987244725227356,
      "learning_rate": 8.810679611650485e-05,
      "loss": 0.6058,
      "step": 1012
    },
    {
      "epoch": 2.212137780207764,
      "grad_norm": 0.0736922174692154,
      "learning_rate": 8.786407766990291e-05,
      "loss": 0.7735,
      "step": 1013
    },
    {
      "epoch": 2.2143247676325863,
      "grad_norm": 0.06956300139427185,
      "learning_rate": 8.762135922330096e-05,
      "loss": 0.8033,
      "step": 1014
    },
    {
      "epoch": 2.2165117550574083,
      "grad_norm": 0.06588080525398254,
      "learning_rate": 8.737864077669901e-05,
      "loss": 0.59,
      "step": 1015
    },
    {
      "epoch": 2.2186987424822306,
      "grad_norm": 0.0655379667878151,
      "learning_rate": 8.713592233009709e-05,
      "loss": 0.433,
      "step": 1016
    },
    {
      "epoch": 2.220885729907053,
      "grad_norm": 0.0930158868432045,
      "learning_rate": 8.689320388349514e-05,
      "loss": 0.7287,
      "step": 1017
    },
    {
      "epoch": 2.2230727173318754,
      "grad_norm": 0.10869035869836807,
      "learning_rate": 8.665048543689319e-05,
      "loss": 0.8585,
      "step": 1018
    },
    {
      "epoch": 2.2252597047566978,
      "grad_norm": 0.10342340171337128,
      "learning_rate": 8.640776699029127e-05,
      "loss": 0.8002,
      "step": 1019
    },
    {
      "epoch": 2.22744669218152,
      "grad_norm": 0.2504841685295105,
      "learning_rate": 8.616504854368932e-05,
      "loss": 0.5052,
      "step": 1020
    },
    {
      "epoch": 2.229633679606342,
      "grad_norm": 0.07130393385887146,
      "learning_rate": 8.592233009708737e-05,
      "loss": 0.6461,
      "step": 1021
    },
    {
      "epoch": 2.2318206670311644,
      "grad_norm": 0.06542620062828064,
      "learning_rate": 8.567961165048543e-05,
      "loss": 0.5922,
      "step": 1022
    },
    {
      "epoch": 2.234007654455987,
      "grad_norm": 0.05315176770091057,
      "learning_rate": 8.543689320388348e-05,
      "loss": 0.5041,
      "step": 1023
    },
    {
      "epoch": 2.236194641880809,
      "grad_norm": 0.06824400275945663,
      "learning_rate": 8.519417475728155e-05,
      "loss": 0.6448,
      "step": 1024
    },
    {
      "epoch": 2.2383816293056316,
      "grad_norm": 0.11105392873287201,
      "learning_rate": 8.495145631067961e-05,
      "loss": 0.6203,
      "step": 1025
    },
    {
      "epoch": 2.240568616730454,
      "grad_norm": 0.08662376552820206,
      "learning_rate": 8.470873786407766e-05,
      "loss": 0.7169,
      "step": 1026
    },
    {
      "epoch": 2.2427556041552763,
      "grad_norm": 0.07787991315126419,
      "learning_rate": 8.446601941747571e-05,
      "loss": 0.7499,
      "step": 1027
    },
    {
      "epoch": 2.2449425915800982,
      "grad_norm": 0.0659622848033905,
      "learning_rate": 8.422330097087379e-05,
      "loss": 1.0476,
      "step": 1028
    },
    {
      "epoch": 2.2471295790049206,
      "grad_norm": 0.08333081007003784,
      "learning_rate": 8.398058252427184e-05,
      "loss": 0.5645,
      "step": 1029
    },
    {
      "epoch": 2.249316566429743,
      "grad_norm": 0.0681498721241951,
      "learning_rate": 8.373786407766989e-05,
      "loss": 0.5698,
      "step": 1030
    },
    {
      "epoch": 2.2515035538545654,
      "grad_norm": 0.042218856513500214,
      "learning_rate": 8.349514563106795e-05,
      "loss": 0.7866,
      "step": 1031
    },
    {
      "epoch": 2.2536905412793877,
      "grad_norm": 0.1502753645181656,
      "learning_rate": 8.325242718446602e-05,
      "loss": 0.7288,
      "step": 1032
    },
    {
      "epoch": 2.25587752870421,
      "grad_norm": 0.04378807172179222,
      "learning_rate": 8.300970873786407e-05,
      "loss": 0.663,
      "step": 1033
    },
    {
      "epoch": 2.258064516129032,
      "grad_norm": 0.06042996048927307,
      "learning_rate": 8.276699029126213e-05,
      "loss": 0.7021,
      "step": 1034
    },
    {
      "epoch": 2.2602515035538544,
      "grad_norm": 0.13028943538665771,
      "learning_rate": 8.252427184466018e-05,
      "loss": 0.662,
      "step": 1035
    },
    {
      "epoch": 2.2624384909786768,
      "grad_norm": 0.10280822962522507,
      "learning_rate": 8.228155339805825e-05,
      "loss": 0.7972,
      "step": 1036
    },
    {
      "epoch": 2.264625478403499,
      "grad_norm": 0.08380752801895142,
      "learning_rate": 8.203883495145631e-05,
      "loss": 0.8213,
      "step": 1037
    },
    {
      "epoch": 2.2668124658283215,
      "grad_norm": 0.0534130297601223,
      "learning_rate": 8.179611650485436e-05,
      "loss": 0.6322,
      "step": 1038
    },
    {
      "epoch": 2.268999453253144,
      "grad_norm": 0.029188288375735283,
      "learning_rate": 8.155339805825241e-05,
      "loss": 1.1116,
      "step": 1039
    },
    {
      "epoch": 2.2711864406779663,
      "grad_norm": 0.07449432462453842,
      "learning_rate": 8.131067961165049e-05,
      "loss": 0.7614,
      "step": 1040
    },
    {
      "epoch": 2.273373428102788,
      "grad_norm": 0.13352523744106293,
      "learning_rate": 8.106796116504854e-05,
      "loss": 0.7191,
      "step": 1041
    },
    {
      "epoch": 2.2755604155276106,
      "grad_norm": 0.18024753034114838,
      "learning_rate": 8.082524271844659e-05,
      "loss": 0.6954,
      "step": 1042
    },
    {
      "epoch": 2.277747402952433,
      "grad_norm": 0.07135174423456192,
      "learning_rate": 8.058252427184464e-05,
      "loss": 0.7497,
      "step": 1043
    },
    {
      "epoch": 2.2799343903772553,
      "grad_norm": 0.04793510586023331,
      "learning_rate": 8.033980582524272e-05,
      "loss": 0.6165,
      "step": 1044
    },
    {
      "epoch": 2.2821213778020777,
      "grad_norm": 0.12190023809671402,
      "learning_rate": 8.009708737864077e-05,
      "loss": 0.8184,
      "step": 1045
    },
    {
      "epoch": 2.2843083652269,
      "grad_norm": 0.06986870616674423,
      "learning_rate": 7.985436893203882e-05,
      "loss": 0.7382,
      "step": 1046
    },
    {
      "epoch": 2.2864953526517224,
      "grad_norm": 0.044186268001794815,
      "learning_rate": 7.961165048543688e-05,
      "loss": 0.6763,
      "step": 1047
    },
    {
      "epoch": 2.2886823400765444,
      "grad_norm": 0.08646490424871445,
      "learning_rate": 7.936893203883495e-05,
      "loss": 0.8968,
      "step": 1048
    },
    {
      "epoch": 2.2908693275013667,
      "grad_norm": 0.14186227321624756,
      "learning_rate": 7.9126213592233e-05,
      "loss": 0.7973,
      "step": 1049
    },
    {
      "epoch": 2.293056314926189,
      "grad_norm": 0.2580704987049103,
      "learning_rate": 7.888349514563106e-05,
      "loss": 0.5353,
      "step": 1050
    },
    {
      "epoch": 2.2952433023510115,
      "grad_norm": 0.08344961702823639,
      "learning_rate": 7.864077669902911e-05,
      "loss": 0.4765,
      "step": 1051
    },
    {
      "epoch": 2.297430289775834,
      "grad_norm": 0.04605596140027046,
      "learning_rate": 7.839805825242716e-05,
      "loss": 0.5672,
      "step": 1052
    },
    {
      "epoch": 2.2996172772006562,
      "grad_norm": 0.08017672598361969,
      "learning_rate": 7.815533980582524e-05,
      "loss": 0.6236,
      "step": 1053
    },
    {
      "epoch": 2.3018042646254786,
      "grad_norm": 0.0792083665728569,
      "learning_rate": 7.791262135922329e-05,
      "loss": 0.6233,
      "step": 1054
    },
    {
      "epoch": 2.3039912520503005,
      "grad_norm": 3.3572826385498047,
      "learning_rate": 7.766990291262134e-05,
      "loss": 0.6117,
      "step": 1055
    },
    {
      "epoch": 2.306178239475123,
      "grad_norm": 0.0807461366057396,
      "learning_rate": 7.742718446601942e-05,
      "loss": 0.7919,
      "step": 1056
    },
    {
      "epoch": 2.3083652268999453,
      "grad_norm": 0.0449124239385128,
      "learning_rate": 7.718446601941747e-05,
      "loss": 0.4221,
      "step": 1057
    },
    {
      "epoch": 2.3105522143247677,
      "grad_norm": 0.03787796199321747,
      "learning_rate": 7.694174757281552e-05,
      "loss": 0.582,
      "step": 1058
    },
    {
      "epoch": 2.31273920174959,
      "grad_norm": 0.04931551218032837,
      "learning_rate": 7.669902912621359e-05,
      "loss": 0.569,
      "step": 1059
    },
    {
      "epoch": 2.3149261891744124,
      "grad_norm": 0.08203588426113129,
      "learning_rate": 7.645631067961165e-05,
      "loss": 0.5703,
      "step": 1060
    },
    {
      "epoch": 2.3171131765992348,
      "grad_norm": 0.07741877436637878,
      "learning_rate": 7.62135922330097e-05,
      "loss": 0.5266,
      "step": 1061
    },
    {
      "epoch": 2.3193001640240567,
      "grad_norm": 0.07360921800136566,
      "learning_rate": 7.597087378640776e-05,
      "loss": 0.5674,
      "step": 1062
    },
    {
      "epoch": 2.321487151448879,
      "grad_norm": 0.0719866156578064,
      "learning_rate": 7.572815533980581e-05,
      "loss": 0.8207,
      "step": 1063
    },
    {
      "epoch": 2.3236741388737014,
      "grad_norm": 0.0980333462357521,
      "learning_rate": 7.548543689320387e-05,
      "loss": 0.6523,
      "step": 1064
    },
    {
      "epoch": 2.325861126298524,
      "grad_norm": 0.06781519204378128,
      "learning_rate": 7.524271844660194e-05,
      "loss": 0.5863,
      "step": 1065
    },
    {
      "epoch": 2.328048113723346,
      "grad_norm": 0.14128249883651733,
      "learning_rate": 7.5e-05,
      "loss": 0.7345,
      "step": 1066
    },
    {
      "epoch": 2.3302351011481686,
      "grad_norm": 0.0773346945643425,
      "learning_rate": 7.475728155339806e-05,
      "loss": 0.5622,
      "step": 1067
    },
    {
      "epoch": 2.332422088572991,
      "grad_norm": 0.06902678310871124,
      "learning_rate": 7.451456310679611e-05,
      "loss": 0.432,
      "step": 1068
    },
    {
      "epoch": 2.334609075997813,
      "grad_norm": 0.05239726975560188,
      "learning_rate": 7.427184466019417e-05,
      "loss": 0.5932,
      "step": 1069
    },
    {
      "epoch": 2.3367960634226352,
      "grad_norm": 0.0757569819688797,
      "learning_rate": 7.402912621359224e-05,
      "loss": 0.7471,
      "step": 1070
    },
    {
      "epoch": 2.3389830508474576,
      "grad_norm": 0.07388169318437576,
      "learning_rate": 7.378640776699029e-05,
      "loss": 0.8203,
      "step": 1071
    },
    {
      "epoch": 2.34117003827228,
      "grad_norm": 0.06949525326490402,
      "learning_rate": 7.354368932038834e-05,
      "loss": 0.7012,
      "step": 1072
    },
    {
      "epoch": 2.3433570256971024,
      "grad_norm": 0.19210799038410187,
      "learning_rate": 7.33009708737864e-05,
      "loss": 0.8093,
      "step": 1073
    },
    {
      "epoch": 2.3455440131219247,
      "grad_norm": 0.06348125636577606,
      "learning_rate": 7.305825242718445e-05,
      "loss": 0.6288,
      "step": 1074
    },
    {
      "epoch": 2.347731000546747,
      "grad_norm": 0.10887488722801208,
      "learning_rate": 7.281553398058252e-05,
      "loss": 0.787,
      "step": 1075
    },
    {
      "epoch": 2.349917987971569,
      "grad_norm": 0.02196069434285164,
      "learning_rate": 7.257281553398058e-05,
      "loss": 0.4349,
      "step": 1076
    },
    {
      "epoch": 2.3521049753963914,
      "grad_norm": 0.09422359615564346,
      "learning_rate": 7.233009708737863e-05,
      "loss": 0.6258,
      "step": 1077
    },
    {
      "epoch": 2.354291962821214,
      "grad_norm": 0.07091114670038223,
      "learning_rate": 7.20873786407767e-05,
      "loss": 0.629,
      "step": 1078
    },
    {
      "epoch": 2.356478950246036,
      "grad_norm": 0.07532557845115662,
      "learning_rate": 7.184466019417474e-05,
      "loss": 0.5756,
      "step": 1079
    },
    {
      "epoch": 2.3586659376708585,
      "grad_norm": 0.08222904056310654,
      "learning_rate": 7.160194174757281e-05,
      "loss": 1.0221,
      "step": 1080
    },
    {
      "epoch": 2.3608529250956805,
      "grad_norm": 0.10531949996948242,
      "learning_rate": 7.135922330097087e-05,
      "loss": 0.7223,
      "step": 1081
    },
    {
      "epoch": 2.363039912520503,
      "grad_norm": 0.16746295988559723,
      "learning_rate": 7.111650485436892e-05,
      "loss": 0.628,
      "step": 1082
    },
    {
      "epoch": 2.365226899945325,
      "grad_norm": 0.030869996175169945,
      "learning_rate": 7.087378640776699e-05,
      "loss": 0.5624,
      "step": 1083
    },
    {
      "epoch": 2.3674138873701476,
      "grad_norm": 0.0833895355463028,
      "learning_rate": 7.063106796116504e-05,
      "loss": 0.7341,
      "step": 1084
    },
    {
      "epoch": 2.36960087479497,
      "grad_norm": 0.05863624066114426,
      "learning_rate": 7.03883495145631e-05,
      "loss": 0.5487,
      "step": 1085
    },
    {
      "epoch": 2.3717878622197923,
      "grad_norm": 0.0941510945558548,
      "learning_rate": 7.014563106796115e-05,
      "loss": 0.7639,
      "step": 1086
    },
    {
      "epoch": 2.3739748496446147,
      "grad_norm": 0.08358798176050186,
      "learning_rate": 6.990291262135922e-05,
      "loss": 0.7019,
      "step": 1087
    },
    {
      "epoch": 2.3761618370694366,
      "grad_norm": 0.1289578527212143,
      "learning_rate": 6.966019417475727e-05,
      "loss": 0.6672,
      "step": 1088
    },
    {
      "epoch": 2.378348824494259,
      "grad_norm": 0.03130876645445824,
      "learning_rate": 6.941747572815533e-05,
      "loss": 0.9502,
      "step": 1089
    },
    {
      "epoch": 2.3805358119190814,
      "grad_norm": 0.05892246961593628,
      "learning_rate": 6.91747572815534e-05,
      "loss": 0.4054,
      "step": 1090
    },
    {
      "epoch": 2.3827227993439037,
      "grad_norm": 0.08211269229650497,
      "learning_rate": 6.893203883495145e-05,
      "loss": 0.6134,
      "step": 1091
    },
    {
      "epoch": 2.384909786768726,
      "grad_norm": 0.14080072939395905,
      "learning_rate": 6.868932038834951e-05,
      "loss": 0.8127,
      "step": 1092
    },
    {
      "epoch": 2.3870967741935485,
      "grad_norm": 0.08313527703285217,
      "learning_rate": 6.844660194174757e-05,
      "loss": 0.6494,
      "step": 1093
    },
    {
      "epoch": 2.389283761618371,
      "grad_norm": 0.12474166601896286,
      "learning_rate": 6.820388349514562e-05,
      "loss": 0.5539,
      "step": 1094
    },
    {
      "epoch": 2.391470749043193,
      "grad_norm": 0.08614606410264969,
      "learning_rate": 6.796116504854369e-05,
      "loss": 0.6714,
      "step": 1095
    },
    {
      "epoch": 2.393657736468015,
      "grad_norm": 0.07203420996665955,
      "learning_rate": 6.771844660194174e-05,
      "loss": 0.4596,
      "step": 1096
    },
    {
      "epoch": 2.3958447238928375,
      "grad_norm": 0.08504094183444977,
      "learning_rate": 6.74757281553398e-05,
      "loss": 0.5993,
      "step": 1097
    },
    {
      "epoch": 2.39803171131766,
      "grad_norm": 0.1589978188276291,
      "learning_rate": 6.723300970873785e-05,
      "loss": 0.6799,
      "step": 1098
    },
    {
      "epoch": 2.4002186987424823,
      "grad_norm": 0.0418832041323185,
      "learning_rate": 6.699029126213592e-05,
      "loss": 0.6429,
      "step": 1099
    },
    {
      "epoch": 2.4024056861673047,
      "grad_norm": 0.06284458935260773,
      "learning_rate": 6.674757281553397e-05,
      "loss": 0.6468,
      "step": 1100
    },
    {
      "epoch": 2.404592673592127,
      "grad_norm": 0.10806804895401001,
      "learning_rate": 6.650485436893203e-05,
      "loss": 0.9486,
      "step": 1101
    },
    {
      "epoch": 2.406779661016949,
      "grad_norm": 0.08519349247217178,
      "learning_rate": 6.62621359223301e-05,
      "loss": 0.4344,
      "step": 1102
    },
    {
      "epoch": 2.4089666484417713,
      "grad_norm": 0.05331065133213997,
      "learning_rate": 6.601941747572815e-05,
      "loss": 0.7634,
      "step": 1103
    },
    {
      "epoch": 2.4111536358665937,
      "grad_norm": 0.049161165952682495,
      "learning_rate": 6.577669902912621e-05,
      "loss": 0.5412,
      "step": 1104
    },
    {
      "epoch": 2.413340623291416,
      "grad_norm": 0.04863851144909859,
      "learning_rate": 6.553398058252426e-05,
      "loss": 0.5051,
      "step": 1105
    },
    {
      "epoch": 2.4155276107162384,
      "grad_norm": 0.05954289436340332,
      "learning_rate": 6.529126213592233e-05,
      "loss": 0.8638,
      "step": 1106
    },
    {
      "epoch": 2.417714598141061,
      "grad_norm": 0.12451896071434021,
      "learning_rate": 6.504854368932039e-05,
      "loss": 0.6294,
      "step": 1107
    },
    {
      "epoch": 2.419901585565883,
      "grad_norm": 0.07101567834615707,
      "learning_rate": 6.480582524271844e-05,
      "loss": 0.6948,
      "step": 1108
    },
    {
      "epoch": 2.422088572990705,
      "grad_norm": 0.0766870379447937,
      "learning_rate": 6.45631067961165e-05,
      "loss": 0.7316,
      "step": 1109
    },
    {
      "epoch": 2.4242755604155275,
      "grad_norm": 0.04993129521608353,
      "learning_rate": 6.432038834951455e-05,
      "loss": 0.5916,
      "step": 1110
    },
    {
      "epoch": 2.42646254784035,
      "grad_norm": 0.10340382903814316,
      "learning_rate": 6.407766990291262e-05,
      "loss": 0.4282,
      "step": 1111
    },
    {
      "epoch": 2.4286495352651722,
      "grad_norm": 0.10339093208312988,
      "learning_rate": 6.383495145631067e-05,
      "loss": 0.6349,
      "step": 1112
    },
    {
      "epoch": 2.4308365226899946,
      "grad_norm": 0.13234268128871918,
      "learning_rate": 6.359223300970873e-05,
      "loss": 0.8036,
      "step": 1113
    },
    {
      "epoch": 2.433023510114817,
      "grad_norm": 0.07981318235397339,
      "learning_rate": 6.334951456310678e-05,
      "loss": 0.6513,
      "step": 1114
    },
    {
      "epoch": 2.4352104975396394,
      "grad_norm": 0.13569766283035278,
      "learning_rate": 6.310679611650485e-05,
      "loss": 0.8387,
      "step": 1115
    },
    {
      "epoch": 2.4373974849644613,
      "grad_norm": 0.09119431674480438,
      "learning_rate": 6.286407766990291e-05,
      "loss": 0.585,
      "step": 1116
    },
    {
      "epoch": 2.4395844723892837,
      "grad_norm": 0.1081055998802185,
      "learning_rate": 6.262135922330096e-05,
      "loss": 0.6452,
      "step": 1117
    },
    {
      "epoch": 2.441771459814106,
      "grad_norm": 0.11525560915470123,
      "learning_rate": 6.237864077669903e-05,
      "loss": 0.7167,
      "step": 1118
    },
    {
      "epoch": 2.4439584472389284,
      "grad_norm": 0.06122585013508797,
      "learning_rate": 6.213592233009709e-05,
      "loss": 0.5866,
      "step": 1119
    },
    {
      "epoch": 2.446145434663751,
      "grad_norm": 0.09825724363327026,
      "learning_rate": 6.189320388349514e-05,
      "loss": 0.5137,
      "step": 1120
    },
    {
      "epoch": 2.448332422088573,
      "grad_norm": 0.02984674647450447,
      "learning_rate": 6.16504854368932e-05,
      "loss": 0.4119,
      "step": 1121
    },
    {
      "epoch": 2.4505194095133955,
      "grad_norm": 0.06197787821292877,
      "learning_rate": 6.140776699029126e-05,
      "loss": 0.6159,
      "step": 1122
    },
    {
      "epoch": 2.4527063969382175,
      "grad_norm": 0.04322591796517372,
      "learning_rate": 6.11650485436893e-05,
      "loss": 0.7058,
      "step": 1123
    },
    {
      "epoch": 2.45489338436304,
      "grad_norm": 0.03457969054579735,
      "learning_rate": 6.092233009708738e-05,
      "loss": 0.4582,
      "step": 1124
    },
    {
      "epoch": 2.457080371787862,
      "grad_norm": 0.07612497359514236,
      "learning_rate": 6.0679611650485434e-05,
      "loss": 0.5578,
      "step": 1125
    },
    {
      "epoch": 2.4592673592126846,
      "grad_norm": 0.05250145122408867,
      "learning_rate": 6.0436893203883485e-05,
      "loss": 0.591,
      "step": 1126
    },
    {
      "epoch": 2.461454346637507,
      "grad_norm": 0.11965347081422806,
      "learning_rate": 6.019417475728155e-05,
      "loss": 0.7173,
      "step": 1127
    },
    {
      "epoch": 2.4636413340623293,
      "grad_norm": 0.057462405413389206,
      "learning_rate": 5.995145631067961e-05,
      "loss": 0.5775,
      "step": 1128
    },
    {
      "epoch": 2.4658283214871517,
      "grad_norm": 0.0856429934501648,
      "learning_rate": 5.9708737864077663e-05,
      "loss": 0.5766,
      "step": 1129
    },
    {
      "epoch": 2.4680153089119736,
      "grad_norm": 0.09746617823839188,
      "learning_rate": 5.946601941747573e-05,
      "loss": 0.4959,
      "step": 1130
    },
    {
      "epoch": 2.470202296336796,
      "grad_norm": 0.07750790566205978,
      "learning_rate": 5.922330097087378e-05,
      "loss": 0.6716,
      "step": 1131
    },
    {
      "epoch": 2.4723892837616184,
      "grad_norm": 0.10556381940841675,
      "learning_rate": 5.8980582524271835e-05,
      "loss": 0.8194,
      "step": 1132
    },
    {
      "epoch": 2.4745762711864407,
      "grad_norm": 0.047729332000017166,
      "learning_rate": 5.87378640776699e-05,
      "loss": 0.6458,
      "step": 1133
    },
    {
      "epoch": 2.476763258611263,
      "grad_norm": 0.09303979575634003,
      "learning_rate": 5.849514563106795e-05,
      "loss": 0.6032,
      "step": 1134
    },
    {
      "epoch": 2.478950246036085,
      "grad_norm": 0.06340254843235016,
      "learning_rate": 5.8252427184466014e-05,
      "loss": 0.5289,
      "step": 1135
    },
    {
      "epoch": 2.4811372334609074,
      "grad_norm": 0.03329901024699211,
      "learning_rate": 5.800970873786407e-05,
      "loss": 0.896,
      "step": 1136
    },
    {
      "epoch": 2.48332422088573,
      "grad_norm": 0.1342897117137909,
      "learning_rate": 5.776699029126213e-05,
      "loss": 0.5008,
      "step": 1137
    },
    {
      "epoch": 2.485511208310552,
      "grad_norm": 0.03303806483745575,
      "learning_rate": 5.7524271844660186e-05,
      "loss": 0.4037,
      "step": 1138
    },
    {
      "epoch": 2.4876981957353745,
      "grad_norm": 0.0739361122250557,
      "learning_rate": 5.728155339805825e-05,
      "loss": 0.7167,
      "step": 1139
    },
    {
      "epoch": 2.489885183160197,
      "grad_norm": 0.06320933997631073,
      "learning_rate": 5.70388349514563e-05,
      "loss": 0.919,
      "step": 1140
    },
    {
      "epoch": 2.4920721705850193,
      "grad_norm": 0.0679958239197731,
      "learning_rate": 5.6796116504854364e-05,
      "loss": 0.731,
      "step": 1141
    },
    {
      "epoch": 2.494259158009841,
      "grad_norm": 0.12814053893089294,
      "learning_rate": 5.655339805825242e-05,
      "loss": 0.8651,
      "step": 1142
    },
    {
      "epoch": 2.4964461454346636,
      "grad_norm": 0.07267304509878159,
      "learning_rate": 5.631067961165048e-05,
      "loss": 0.4623,
      "step": 1143
    },
    {
      "epoch": 2.498633132859486,
      "grad_norm": 0.06321340799331665,
      "learning_rate": 5.6067961165048536e-05,
      "loss": 0.5501,
      "step": 1144
    },
    {
      "epoch": 2.5008201202843083,
      "grad_norm": 0.10648145526647568,
      "learning_rate": 5.58252427184466e-05,
      "loss": 0.5048,
      "step": 1145
    },
    {
      "epoch": 2.5030071077091307,
      "grad_norm": 0.10429555922746658,
      "learning_rate": 5.558252427184465e-05,
      "loss": 0.701,
      "step": 1146
    },
    {
      "epoch": 2.505194095133953,
      "grad_norm": 0.06531582027673721,
      "learning_rate": 5.5339805825242715e-05,
      "loss": 0.446,
      "step": 1147
    },
    {
      "epoch": 2.5073810825587755,
      "grad_norm": 0.04627282917499542,
      "learning_rate": 5.509708737864077e-05,
      "loss": 0.5879,
      "step": 1148
    },
    {
      "epoch": 2.5095680699835974,
      "grad_norm": 0.12454242259263992,
      "learning_rate": 5.485436893203883e-05,
      "loss": 0.5276,
      "step": 1149
    },
    {
      "epoch": 2.5117550574084198,
      "grad_norm": 0.09565190225839615,
      "learning_rate": 5.461165048543689e-05,
      "loss": 0.6706,
      "step": 1150
    },
    {
      "epoch": 2.513942044833242,
      "grad_norm": 0.09251926094293594,
      "learning_rate": 5.436893203883495e-05,
      "loss": 0.6097,
      "step": 1151
    },
    {
      "epoch": 2.5161290322580645,
      "grad_norm": 0.08162148296833038,
      "learning_rate": 5.4126213592233e-05,
      "loss": 0.5359,
      "step": 1152
    },
    {
      "epoch": 2.518316019682887,
      "grad_norm": 0.041675105690956116,
      "learning_rate": 5.3883495145631065e-05,
      "loss": 0.8033,
      "step": 1153
    },
    {
      "epoch": 2.5205030071077092,
      "grad_norm": 0.039802707731723785,
      "learning_rate": 5.364077669902912e-05,
      "loss": 0.6052,
      "step": 1154
    },
    {
      "epoch": 2.5226899945325316,
      "grad_norm": 0.08948308974504471,
      "learning_rate": 5.339805825242718e-05,
      "loss": 0.6798,
      "step": 1155
    },
    {
      "epoch": 2.5248769819573536,
      "grad_norm": 0.10361290723085403,
      "learning_rate": 5.315533980582524e-05,
      "loss": 0.5986,
      "step": 1156
    },
    {
      "epoch": 2.527063969382176,
      "grad_norm": 0.1291511505842209,
      "learning_rate": 5.2912621359223295e-05,
      "loss": 0.6655,
      "step": 1157
    },
    {
      "epoch": 2.5292509568069983,
      "grad_norm": 0.15041688084602356,
      "learning_rate": 5.266990291262135e-05,
      "loss": 0.7115,
      "step": 1158
    },
    {
      "epoch": 2.5314379442318207,
      "grad_norm": 0.08857720345258713,
      "learning_rate": 5.2427184466019416e-05,
      "loss": 0.7579,
      "step": 1159
    },
    {
      "epoch": 2.533624931656643,
      "grad_norm": 0.08004803210496902,
      "learning_rate": 5.2184466019417466e-05,
      "loss": 0.6388,
      "step": 1160
    },
    {
      "epoch": 2.5358119190814654,
      "grad_norm": 0.06408461928367615,
      "learning_rate": 5.194174757281553e-05,
      "loss": 0.6373,
      "step": 1161
    },
    {
      "epoch": 2.537998906506288,
      "grad_norm": 0.08298526704311371,
      "learning_rate": 5.169902912621359e-05,
      "loss": 0.7151,
      "step": 1162
    },
    {
      "epoch": 2.5401858939311097,
      "grad_norm": 0.09231656789779663,
      "learning_rate": 5.1456310679611645e-05,
      "loss": 0.6188,
      "step": 1163
    },
    {
      "epoch": 2.542372881355932,
      "grad_norm": 0.11329259723424911,
      "learning_rate": 5.12135922330097e-05,
      "loss": 0.6681,
      "step": 1164
    },
    {
      "epoch": 2.5445598687807545,
      "grad_norm": 0.17237365245819092,
      "learning_rate": 5.0970873786407766e-05,
      "loss": 0.8227,
      "step": 1165
    },
    {
      "epoch": 2.546746856205577,
      "grad_norm": 0.07260779291391373,
      "learning_rate": 5.072815533980582e-05,
      "loss": 0.6832,
      "step": 1166
    },
    {
      "epoch": 2.548933843630399,
      "grad_norm": 0.0579003281891346,
      "learning_rate": 5.048543689320388e-05,
      "loss": 0.5133,
      "step": 1167
    },
    {
      "epoch": 2.5511208310552216,
      "grad_norm": 0.05579472705721855,
      "learning_rate": 5.024271844660194e-05,
      "loss": 0.5569,
      "step": 1168
    },
    {
      "epoch": 2.553307818480044,
      "grad_norm": 0.08632995188236237,
      "learning_rate": 4.9999999999999996e-05,
      "loss": 0.8297,
      "step": 1169
    },
    {
      "epoch": 2.555494805904866,
      "grad_norm": 0.05742983520030975,
      "learning_rate": 4.975728155339805e-05,
      "loss": 0.5625,
      "step": 1170
    },
    {
      "epoch": 2.5576817933296883,
      "grad_norm": 0.09453807771205902,
      "learning_rate": 4.951456310679612e-05,
      "loss": 0.5199,
      "step": 1171
    },
    {
      "epoch": 2.5598687807545106,
      "grad_norm": 0.07085605710744858,
      "learning_rate": 4.927184466019417e-05,
      "loss": 0.6006,
      "step": 1172
    },
    {
      "epoch": 2.562055768179333,
      "grad_norm": 0.1340280920267105,
      "learning_rate": 4.902912621359223e-05,
      "loss": 0.6783,
      "step": 1173
    },
    {
      "epoch": 2.5642427556041554,
      "grad_norm": 0.08021483570337296,
      "learning_rate": 4.878640776699029e-05,
      "loss": 0.4783,
      "step": 1174
    },
    {
      "epoch": 2.5664297430289773,
      "grad_norm": 0.04842549189925194,
      "learning_rate": 4.8543689320388346e-05,
      "loss": 0.6662,
      "step": 1175
    },
    {
      "epoch": 2.5686167304538,
      "grad_norm": 0.1178601086139679,
      "learning_rate": 4.8300970873786403e-05,
      "loss": 0.7801,
      "step": 1176
    },
    {
      "epoch": 2.570803717878622,
      "grad_norm": 0.06883268803358078,
      "learning_rate": 4.805825242718447e-05,
      "loss": 0.5078,
      "step": 1177
    },
    {
      "epoch": 2.5729907053034444,
      "grad_norm": 0.05313848704099655,
      "learning_rate": 4.781553398058252e-05,
      "loss": 0.7318,
      "step": 1178
    },
    {
      "epoch": 2.575177692728267,
      "grad_norm": 0.05223860591650009,
      "learning_rate": 4.757281553398058e-05,
      "loss": 0.5506,
      "step": 1179
    },
    {
      "epoch": 2.577364680153089,
      "grad_norm": 0.06799974292516708,
      "learning_rate": 4.733009708737864e-05,
      "loss": 0.7171,
      "step": 1180
    },
    {
      "epoch": 2.5795516675779115,
      "grad_norm": 0.05223657935857773,
      "learning_rate": 4.70873786407767e-05,
      "loss": 0.6697,
      "step": 1181
    },
    {
      "epoch": 2.5817386550027335,
      "grad_norm": 0.12002090364694595,
      "learning_rate": 4.6844660194174754e-05,
      "loss": 0.5765,
      "step": 1182
    },
    {
      "epoch": 2.5839256424275563,
      "grad_norm": 0.112289659678936,
      "learning_rate": 4.6601941747572804e-05,
      "loss": 0.6459,
      "step": 1183
    },
    {
      "epoch": 2.586112629852378,
      "grad_norm": 0.06973439455032349,
      "learning_rate": 4.635922330097087e-05,
      "loss": 0.6429,
      "step": 1184
    },
    {
      "epoch": 2.5882996172772006,
      "grad_norm": 0.08749748021364212,
      "learning_rate": 4.611650485436893e-05,
      "loss": 0.6056,
      "step": 1185
    },
    {
      "epoch": 2.590486604702023,
      "grad_norm": 0.07339344173669815,
      "learning_rate": 4.587378640776698e-05,
      "loss": 0.6424,
      "step": 1186
    },
    {
      "epoch": 2.5926735921268453,
      "grad_norm": 0.06114998459815979,
      "learning_rate": 4.563106796116504e-05,
      "loss": 0.5619,
      "step": 1187
    },
    {
      "epoch": 2.5948605795516677,
      "grad_norm": 0.10229608416557312,
      "learning_rate": 4.5388349514563104e-05,
      "loss": 0.486,
      "step": 1188
    },
    {
      "epoch": 2.5970475669764896,
      "grad_norm": 0.10732418298721313,
      "learning_rate": 4.5145631067961155e-05,
      "loss": 0.646,
      "step": 1189
    },
    {
      "epoch": 2.5992345544013125,
      "grad_norm": 0.09944164752960205,
      "learning_rate": 4.490291262135922e-05,
      "loss": 0.6848,
      "step": 1190
    },
    {
      "epoch": 2.6014215418261344,
      "grad_norm": 0.05396636202931404,
      "learning_rate": 4.466019417475728e-05,
      "loss": 0.6142,
      "step": 1191
    },
    {
      "epoch": 2.6036085292509568,
      "grad_norm": 0.12629559636116028,
      "learning_rate": 4.4417475728155334e-05,
      "loss": 0.4058,
      "step": 1192
    },
    {
      "epoch": 2.605795516675779,
      "grad_norm": 0.10848679393529892,
      "learning_rate": 4.417475728155339e-05,
      "loss": 0.7677,
      "step": 1193
    },
    {
      "epoch": 2.6079825041006015,
      "grad_norm": 0.06883646547794342,
      "learning_rate": 4.3932038834951455e-05,
      "loss": 0.6009,
      "step": 1194
    },
    {
      "epoch": 2.610169491525424,
      "grad_norm": 0.0868680477142334,
      "learning_rate": 4.3689320388349505e-05,
      "loss": 0.684,
      "step": 1195
    },
    {
      "epoch": 2.612356478950246,
      "grad_norm": 0.028191180899739265,
      "learning_rate": 4.344660194174757e-05,
      "loss": 0.5871,
      "step": 1196
    },
    {
      "epoch": 2.614543466375068,
      "grad_norm": 0.07453961670398712,
      "learning_rate": 4.3203883495145634e-05,
      "loss": 0.5428,
      "step": 1197
    },
    {
      "epoch": 2.6167304537998906,
      "grad_norm": 0.08585245162248611,
      "learning_rate": 4.2961165048543684e-05,
      "loss": 0.6162,
      "step": 1198
    },
    {
      "epoch": 2.618917441224713,
      "grad_norm": 0.052546802908182144,
      "learning_rate": 4.271844660194174e-05,
      "loss": 0.5801,
      "step": 1199
    },
    {
      "epoch": 2.6211044286495353,
      "grad_norm": 0.131413534283638,
      "learning_rate": 4.2475728155339805e-05,
      "loss": 0.9167,
      "step": 1200
    },
    {
      "epoch": 2.6232914160743577,
      "grad_norm": 0.22036047279834747,
      "learning_rate": 4.2233009708737856e-05,
      "loss": 0.5283,
      "step": 1201
    },
    {
      "epoch": 2.62547840349918,
      "grad_norm": 0.3202528953552246,
      "learning_rate": 4.199029126213592e-05,
      "loss": 0.7045,
      "step": 1202
    },
    {
      "epoch": 2.627665390924002,
      "grad_norm": 0.04368839040398598,
      "learning_rate": 4.174757281553398e-05,
      "loss": 0.5529,
      "step": 1203
    },
    {
      "epoch": 2.6298523783488243,
      "grad_norm": 0.04791286960244179,
      "learning_rate": 4.1504854368932035e-05,
      "loss": 0.5689,
      "step": 1204
    },
    {
      "epoch": 2.6320393657736467,
      "grad_norm": 0.04749435931444168,
      "learning_rate": 4.126213592233009e-05,
      "loss": 0.4968,
      "step": 1205
    },
    {
      "epoch": 2.634226353198469,
      "grad_norm": 0.06943855434656143,
      "learning_rate": 4.1019417475728156e-05,
      "loss": 0.6496,
      "step": 1206
    },
    {
      "epoch": 2.6364133406232915,
      "grad_norm": 0.059361230581998825,
      "learning_rate": 4.0776699029126206e-05,
      "loss": 0.601,
      "step": 1207
    },
    {
      "epoch": 2.638600328048114,
      "grad_norm": 0.11798574030399323,
      "learning_rate": 4.053398058252427e-05,
      "loss": 0.6548,
      "step": 1208
    },
    {
      "epoch": 2.640787315472936,
      "grad_norm": 0.05566198378801346,
      "learning_rate": 4.029126213592232e-05,
      "loss": 0.4842,
      "step": 1209
    },
    {
      "epoch": 2.642974302897758,
      "grad_norm": 0.07326702028512955,
      "learning_rate": 4.0048543689320385e-05,
      "loss": 0.5298,
      "step": 1210
    },
    {
      "epoch": 2.6451612903225805,
      "grad_norm": 0.05375058576464653,
      "learning_rate": 3.980582524271844e-05,
      "loss": 0.6232,
      "step": 1211
    },
    {
      "epoch": 2.647348277747403,
      "grad_norm": 0.05424814671278,
      "learning_rate": 3.95631067961165e-05,
      "loss": 0.5531,
      "step": 1212
    },
    {
      "epoch": 2.6495352651722253,
      "grad_norm": 0.04809017851948738,
      "learning_rate": 3.932038834951456e-05,
      "loss": 0.615,
      "step": 1213
    },
    {
      "epoch": 2.6517222525970476,
      "grad_norm": 0.07291999459266663,
      "learning_rate": 3.907766990291262e-05,
      "loss": 0.5235,
      "step": 1214
    },
    {
      "epoch": 2.65390924002187,
      "grad_norm": 0.11024939268827438,
      "learning_rate": 3.883495145631067e-05,
      "loss": 0.7097,
      "step": 1215
    },
    {
      "epoch": 2.6560962274466924,
      "grad_norm": 0.11854353547096252,
      "learning_rate": 3.8592233009708736e-05,
      "loss": 0.7801,
      "step": 1216
    },
    {
      "epoch": 2.6582832148715143,
      "grad_norm": 0.07645071297883987,
      "learning_rate": 3.834951456310679e-05,
      "loss": 0.632,
      "step": 1217
    },
    {
      "epoch": 2.6604702022963367,
      "grad_norm": 0.055733490735292435,
      "learning_rate": 3.810679611650485e-05,
      "loss": 0.7028,
      "step": 1218
    },
    {
      "epoch": 2.662657189721159,
      "grad_norm": 0.08349243551492691,
      "learning_rate": 3.786407766990291e-05,
      "loss": 0.6578,
      "step": 1219
    },
    {
      "epoch": 2.6648441771459814,
      "grad_norm": 0.10217994451522827,
      "learning_rate": 3.762135922330097e-05,
      "loss": 0.6112,
      "step": 1220
    },
    {
      "epoch": 2.667031164570804,
      "grad_norm": 0.09897356480360031,
      "learning_rate": 3.737864077669903e-05,
      "loss": 0.8556,
      "step": 1221
    },
    {
      "epoch": 2.669218151995626,
      "grad_norm": 0.06858976930379868,
      "learning_rate": 3.7135922330097086e-05,
      "loss": 0.4973,
      "step": 1222
    },
    {
      "epoch": 2.6714051394204485,
      "grad_norm": 0.06404196470975876,
      "learning_rate": 3.689320388349514e-05,
      "loss": 0.6602,
      "step": 1223
    },
    {
      "epoch": 2.6735921268452705,
      "grad_norm": 0.07733612507581711,
      "learning_rate": 3.66504854368932e-05,
      "loss": 0.4758,
      "step": 1224
    },
    {
      "epoch": 2.675779114270093,
      "grad_norm": 0.09863855689764023,
      "learning_rate": 3.640776699029126e-05,
      "loss": 0.6417,
      "step": 1225
    },
    {
      "epoch": 2.6779661016949152,
      "grad_norm": 0.08200982958078384,
      "learning_rate": 3.6165048543689315e-05,
      "loss": 0.6043,
      "step": 1226
    },
    {
      "epoch": 2.6801530891197376,
      "grad_norm": 0.07447304576635361,
      "learning_rate": 3.592233009708737e-05,
      "loss": 0.5741,
      "step": 1227
    },
    {
      "epoch": 2.68234007654456,
      "grad_norm": 0.05752440541982651,
      "learning_rate": 3.5679611650485437e-05,
      "loss": 1.4945,
      "step": 1228
    },
    {
      "epoch": 2.684527063969382,
      "grad_norm": 0.13874761760234833,
      "learning_rate": 3.5436893203883494e-05,
      "loss": 0.6187,
      "step": 1229
    },
    {
      "epoch": 2.6867140513942047,
      "grad_norm": 0.04540708661079407,
      "learning_rate": 3.519417475728155e-05,
      "loss": 0.4935,
      "step": 1230
    },
    {
      "epoch": 2.6889010388190266,
      "grad_norm": 0.04991524666547775,
      "learning_rate": 3.495145631067961e-05,
      "loss": 0.6385,
      "step": 1231
    },
    {
      "epoch": 2.691088026243849,
      "grad_norm": 0.06925421953201294,
      "learning_rate": 3.4708737864077666e-05,
      "loss": 0.6022,
      "step": 1232
    },
    {
      "epoch": 2.6932750136686714,
      "grad_norm": 0.1184193342924118,
      "learning_rate": 3.446601941747572e-05,
      "loss": 0.603,
      "step": 1233
    },
    {
      "epoch": 2.6954620010934938,
      "grad_norm": 0.054968900978565216,
      "learning_rate": 3.422330097087379e-05,
      "loss": 0.7349,
      "step": 1234
    },
    {
      "epoch": 2.697648988518316,
      "grad_norm": 0.053675152361392975,
      "learning_rate": 3.3980582524271844e-05,
      "loss": 0.6316,
      "step": 1235
    },
    {
      "epoch": 2.699835975943138,
      "grad_norm": 0.08283350616693497,
      "learning_rate": 3.37378640776699e-05,
      "loss": 0.6284,
      "step": 1236
    },
    {
      "epoch": 2.702022963367961,
      "grad_norm": 0.06438212096691132,
      "learning_rate": 3.349514563106796e-05,
      "loss": 0.6579,
      "step": 1237
    },
    {
      "epoch": 2.704209950792783,
      "grad_norm": 0.08433155715465546,
      "learning_rate": 3.3252427184466016e-05,
      "loss": 0.5701,
      "step": 1238
    },
    {
      "epoch": 2.706396938217605,
      "grad_norm": 0.059264179319143295,
      "learning_rate": 3.3009708737864073e-05,
      "loss": 0.7575,
      "step": 1239
    },
    {
      "epoch": 2.7085839256424276,
      "grad_norm": 0.0719296857714653,
      "learning_rate": 3.276699029126213e-05,
      "loss": 0.8028,
      "step": 1240
    },
    {
      "epoch": 2.71077091306725,
      "grad_norm": 0.12577201426029205,
      "learning_rate": 3.2524271844660195e-05,
      "loss": 0.6245,
      "step": 1241
    },
    {
      "epoch": 2.7129579004920723,
      "grad_norm": 0.06721910089254379,
      "learning_rate": 3.228155339805825e-05,
      "loss": 0.6522,
      "step": 1242
    },
    {
      "epoch": 2.7151448879168942,
      "grad_norm": 0.06833085417747498,
      "learning_rate": 3.203883495145631e-05,
      "loss": 0.5381,
      "step": 1243
    },
    {
      "epoch": 2.717331875341717,
      "grad_norm": 0.08729522675275803,
      "learning_rate": 3.179611650485437e-05,
      "loss": 0.7297,
      "step": 1244
    },
    {
      "epoch": 2.719518862766539,
      "grad_norm": 0.08239904046058655,
      "learning_rate": 3.1553398058252424e-05,
      "loss": 0.5976,
      "step": 1245
    },
    {
      "epoch": 2.7217058501913614,
      "grad_norm": 0.04949947074055672,
      "learning_rate": 3.131067961165048e-05,
      "loss": 0.593,
      "step": 1246
    },
    {
      "epoch": 2.7238928376161837,
      "grad_norm": 0.06915730983018875,
      "learning_rate": 3.1067961165048545e-05,
      "loss": 0.8052,
      "step": 1247
    },
    {
      "epoch": 2.726079825041006,
      "grad_norm": 0.1225258857011795,
      "learning_rate": 3.08252427184466e-05,
      "loss": 0.7143,
      "step": 1248
    },
    {
      "epoch": 2.7282668124658285,
      "grad_norm": 0.08791057020425797,
      "learning_rate": 3.058252427184465e-05,
      "loss": 0.5169,
      "step": 1249
    },
    {
      "epoch": 2.7304537998906504,
      "grad_norm": 0.09579493850469589,
      "learning_rate": 3.0339805825242717e-05,
      "loss": 0.5776,
      "step": 1250
    },
    {
      "epoch": 2.7326407873154728,
      "grad_norm": 0.11249995231628418,
      "learning_rate": 3.0097087378640774e-05,
      "loss": 0.7811,
      "step": 1251
    },
    {
      "epoch": 2.734827774740295,
      "grad_norm": 0.09353052079677582,
      "learning_rate": 2.9854368932038832e-05,
      "loss": 0.6474,
      "step": 1252
    },
    {
      "epoch": 2.7370147621651175,
      "grad_norm": 0.0948183462023735,
      "learning_rate": 2.961165048543689e-05,
      "loss": 0.6142,
      "step": 1253
    },
    {
      "epoch": 2.73920174958994,
      "grad_norm": 0.03197981417179108,
      "learning_rate": 2.936893203883495e-05,
      "loss": 0.5058,
      "step": 1254
    },
    {
      "epoch": 2.7413887370147623,
      "grad_norm": 0.08017853647470474,
      "learning_rate": 2.9126213592233007e-05,
      "loss": 0.583,
      "step": 1255
    },
    {
      "epoch": 2.7435757244395846,
      "grad_norm": 0.04022474214434624,
      "learning_rate": 2.8883495145631064e-05,
      "loss": 0.8782,
      "step": 1256
    },
    {
      "epoch": 2.7457627118644066,
      "grad_norm": 0.061203911900520325,
      "learning_rate": 2.8640776699029125e-05,
      "loss": 0.7384,
      "step": 1257
    },
    {
      "epoch": 2.747949699289229,
      "grad_norm": 0.07927464693784714,
      "learning_rate": 2.8398058252427182e-05,
      "loss": 0.651,
      "step": 1258
    },
    {
      "epoch": 2.7501366867140513,
      "grad_norm": 0.041599441319704056,
      "learning_rate": 2.815533980582524e-05,
      "loss": 0.7055,
      "step": 1259
    },
    {
      "epoch": 2.7523236741388737,
      "grad_norm": 0.05352627485990524,
      "learning_rate": 2.79126213592233e-05,
      "loss": 0.5113,
      "step": 1260
    },
    {
      "epoch": 2.754510661563696,
      "grad_norm": 0.06345394998788834,
      "learning_rate": 2.7669902912621357e-05,
      "loss": 0.6726,
      "step": 1261
    },
    {
      "epoch": 2.7566976489885184,
      "grad_norm": 0.059153128415346146,
      "learning_rate": 2.7427184466019415e-05,
      "loss": 0.3546,
      "step": 1262
    },
    {
      "epoch": 2.758884636413341,
      "grad_norm": 0.08715004473924637,
      "learning_rate": 2.7184466019417475e-05,
      "loss": 0.6837,
      "step": 1263
    },
    {
      "epoch": 2.7610716238381627,
      "grad_norm": 0.05116257816553116,
      "learning_rate": 2.6941747572815533e-05,
      "loss": 0.7118,
      "step": 1264
    },
    {
      "epoch": 2.763258611262985,
      "grad_norm": 0.118152916431427,
      "learning_rate": 2.669902912621359e-05,
      "loss": 0.6275,
      "step": 1265
    },
    {
      "epoch": 2.7654455986878075,
      "grad_norm": 0.04079804942011833,
      "learning_rate": 2.6456310679611647e-05,
      "loss": 0.4964,
      "step": 1266
    },
    {
      "epoch": 2.76763258611263,
      "grad_norm": 0.1402898132801056,
      "learning_rate": 2.6213592233009708e-05,
      "loss": 0.6728,
      "step": 1267
    },
    {
      "epoch": 2.7698195735374522,
      "grad_norm": 0.10157422721385956,
      "learning_rate": 2.5970873786407765e-05,
      "loss": 0.6064,
      "step": 1268
    },
    {
      "epoch": 2.7720065609622746,
      "grad_norm": 0.06446008384227753,
      "learning_rate": 2.5728155339805823e-05,
      "loss": 0.7651,
      "step": 1269
    },
    {
      "epoch": 2.774193548387097,
      "grad_norm": 0.07324625551700592,
      "learning_rate": 2.5485436893203883e-05,
      "loss": 0.6383,
      "step": 1270
    },
    {
      "epoch": 2.776380535811919,
      "grad_norm": 0.07557366043329239,
      "learning_rate": 2.524271844660194e-05,
      "loss": 0.704,
      "step": 1271
    },
    {
      "epoch": 2.7785675232367413,
      "grad_norm": 0.08134450018405914,
      "learning_rate": 2.4999999999999998e-05,
      "loss": 0.6886,
      "step": 1272
    },
    {
      "epoch": 2.7807545106615636,
      "grad_norm": 0.04855785891413689,
      "learning_rate": 2.475728155339806e-05,
      "loss": 0.741,
      "step": 1273
    },
    {
      "epoch": 2.782941498086386,
      "grad_norm": 0.06953386217355728,
      "learning_rate": 2.4514563106796116e-05,
      "loss": 0.5516,
      "step": 1274
    },
    {
      "epoch": 2.7851284855112084,
      "grad_norm": 0.10012605786323547,
      "learning_rate": 2.4271844660194173e-05,
      "loss": 0.7545,
      "step": 1275
    },
    {
      "epoch": 2.7873154729360308,
      "grad_norm": 0.157820925116539,
      "learning_rate": 2.4029126213592234e-05,
      "loss": 0.7965,
      "step": 1276
    },
    {
      "epoch": 2.789502460360853,
      "grad_norm": 0.04620638117194176,
      "learning_rate": 2.378640776699029e-05,
      "loss": 0.7223,
      "step": 1277
    },
    {
      "epoch": 2.791689447785675,
      "grad_norm": 0.08428886532783508,
      "learning_rate": 2.354368932038835e-05,
      "loss": 0.6298,
      "step": 1278
    },
    {
      "epoch": 2.7938764352104974,
      "grad_norm": 0.0942365750670433,
      "learning_rate": 2.3300970873786402e-05,
      "loss": 0.7266,
      "step": 1279
    },
    {
      "epoch": 2.79606342263532,
      "grad_norm": 0.05135050788521767,
      "learning_rate": 2.3058252427184466e-05,
      "loss": 0.7369,
      "step": 1280
    },
    {
      "epoch": 2.798250410060142,
      "grad_norm": 0.07809579372406006,
      "learning_rate": 2.281553398058252e-05,
      "loss": 0.743,
      "step": 1281
    },
    {
      "epoch": 2.8004373974849646,
      "grad_norm": 0.10320547968149185,
      "learning_rate": 2.2572815533980577e-05,
      "loss": 0.722,
      "step": 1282
    },
    {
      "epoch": 2.8026243849097865,
      "grad_norm": 0.09415378421545029,
      "learning_rate": 2.233009708737864e-05,
      "loss": 0.7528,
      "step": 1283
    },
    {
      "epoch": 2.8048113723346093,
      "grad_norm": 0.13103462755680084,
      "learning_rate": 2.2087378640776695e-05,
      "loss": 0.6785,
      "step": 1284
    },
    {
      "epoch": 2.8069983597594312,
      "grad_norm": 0.09520339220762253,
      "learning_rate": 2.1844660194174753e-05,
      "loss": 0.7469,
      "step": 1285
    },
    {
      "epoch": 2.8091853471842536,
      "grad_norm": 0.06156136468052864,
      "learning_rate": 2.1601941747572817e-05,
      "loss": 0.7619,
      "step": 1286
    },
    {
      "epoch": 2.811372334609076,
      "grad_norm": 0.07038269191980362,
      "learning_rate": 2.135922330097087e-05,
      "loss": 0.452,
      "step": 1287
    },
    {
      "epoch": 2.8135593220338984,
      "grad_norm": 0.07831636816263199,
      "learning_rate": 2.1116504854368928e-05,
      "loss": 0.7101,
      "step": 1288
    },
    {
      "epoch": 2.8157463094587207,
      "grad_norm": 0.08735121041536331,
      "learning_rate": 2.087378640776699e-05,
      "loss": 0.8627,
      "step": 1289
    },
    {
      "epoch": 2.8179332968835427,
      "grad_norm": 0.05308119207620621,
      "learning_rate": 2.0631067961165046e-05,
      "loss": 0.5359,
      "step": 1290
    },
    {
      "epoch": 2.8201202843083655,
      "grad_norm": 0.053742680698633194,
      "learning_rate": 2.0388349514563103e-05,
      "loss": 0.4116,
      "step": 1291
    },
    {
      "epoch": 2.8223072717331874,
      "grad_norm": 0.1100885346531868,
      "learning_rate": 2.014563106796116e-05,
      "loss": 0.9184,
      "step": 1292
    },
    {
      "epoch": 2.82449425915801,
      "grad_norm": 0.04482419416308403,
      "learning_rate": 1.990291262135922e-05,
      "loss": 0.4212,
      "step": 1293
    },
    {
      "epoch": 2.826681246582832,
      "grad_norm": 0.07495293021202087,
      "learning_rate": 1.966019417475728e-05,
      "loss": 0.8411,
      "step": 1294
    },
    {
      "epoch": 2.8288682340076545,
      "grad_norm": 0.06448440998792648,
      "learning_rate": 1.9417475728155336e-05,
      "loss": 0.6008,
      "step": 1295
    },
    {
      "epoch": 2.831055221432477,
      "grad_norm": 0.0787663534283638,
      "learning_rate": 1.9174757281553396e-05,
      "loss": 0.6465,
      "step": 1296
    },
    {
      "epoch": 2.833242208857299,
      "grad_norm": 0.08537033200263977,
      "learning_rate": 1.8932038834951454e-05,
      "loss": 0.7017,
      "step": 1297
    },
    {
      "epoch": 2.8354291962821216,
      "grad_norm": 0.058448903262615204,
      "learning_rate": 1.8689320388349514e-05,
      "loss": 0.4895,
      "step": 1298
    },
    {
      "epoch": 2.8376161837069436,
      "grad_norm": 0.05903905630111694,
      "learning_rate": 1.844660194174757e-05,
      "loss": 0.5857,
      "step": 1299
    },
    {
      "epoch": 2.839803171131766,
      "grad_norm": 0.1151174008846283,
      "learning_rate": 1.820388349514563e-05,
      "loss": 0.7771,
      "step": 1300
    },
    {
      "epoch": 2.8419901585565883,
      "grad_norm": 0.07729051262140274,
      "learning_rate": 1.7961165048543686e-05,
      "loss": 0.7188,
      "step": 1301
    },
    {
      "epoch": 2.8441771459814107,
      "grad_norm": 0.05661698058247566,
      "learning_rate": 1.7718446601941747e-05,
      "loss": 0.679,
      "step": 1302
    },
    {
      "epoch": 2.846364133406233,
      "grad_norm": 0.07814532518386841,
      "learning_rate": 1.7475728155339804e-05,
      "loss": 0.7158,
      "step": 1303
    },
    {
      "epoch": 2.848551120831055,
      "grad_norm": 0.08739367127418518,
      "learning_rate": 1.723300970873786e-05,
      "loss": 0.7517,
      "step": 1304
    },
    {
      "epoch": 2.8507381082558774,
      "grad_norm": 0.11283941566944122,
      "learning_rate": 1.6990291262135922e-05,
      "loss": 0.5734,
      "step": 1305
    },
    {
      "epoch": 2.8529250956806997,
      "grad_norm": 0.0747821256518364,
      "learning_rate": 1.674757281553398e-05,
      "loss": 0.5776,
      "step": 1306
    },
    {
      "epoch": 2.855112083105522,
      "grad_norm": 0.13224893808364868,
      "learning_rate": 1.6504854368932037e-05,
      "loss": 0.9154,
      "step": 1307
    },
    {
      "epoch": 2.8572990705303445,
      "grad_norm": 0.0987900048494339,
      "learning_rate": 1.6262135922330097e-05,
      "loss": 0.7162,
      "step": 1308
    },
    {
      "epoch": 2.859486057955167,
      "grad_norm": 0.11385104060173035,
      "learning_rate": 1.6019417475728155e-05,
      "loss": 0.6492,
      "step": 1309
    },
    {
      "epoch": 2.8616730453799892,
      "grad_norm": 0.0830446407198906,
      "learning_rate": 1.5776699029126212e-05,
      "loss": 0.5508,
      "step": 1310
    },
    {
      "epoch": 2.863860032804811,
      "grad_norm": 0.0805460587143898,
      "learning_rate": 1.5533980582524273e-05,
      "loss": 0.729,
      "step": 1311
    },
    {
      "epoch": 2.8660470202296335,
      "grad_norm": 0.04700075089931488,
      "learning_rate": 1.5291262135922327e-05,
      "loss": 0.4615,
      "step": 1312
    },
    {
      "epoch": 2.868234007654456,
      "grad_norm": 0.061646781861782074,
      "learning_rate": 1.5048543689320387e-05,
      "loss": 0.5481,
      "step": 1313
    },
    {
      "epoch": 2.8704209950792783,
      "grad_norm": 0.08027023822069168,
      "learning_rate": 1.4805825242718445e-05,
      "loss": 0.5735,
      "step": 1314
    },
    {
      "epoch": 2.8726079825041007,
      "grad_norm": 0.03735914081335068,
      "learning_rate": 1.4563106796116503e-05,
      "loss": 0.7346,
      "step": 1315
    },
    {
      "epoch": 2.874794969928923,
      "grad_norm": 0.12285737693309784,
      "learning_rate": 1.4320388349514562e-05,
      "loss": 0.4963,
      "step": 1316
    },
    {
      "epoch": 2.8769819573537454,
      "grad_norm": 0.07507744431495667,
      "learning_rate": 1.407766990291262e-05,
      "loss": 0.6827,
      "step": 1317
    },
    {
      "epoch": 2.8791689447785673,
      "grad_norm": 0.05776430293917656,
      "learning_rate": 1.3834951456310679e-05,
      "loss": 0.5822,
      "step": 1318
    },
    {
      "epoch": 2.8813559322033897,
      "grad_norm": 0.12840424478054047,
      "learning_rate": 1.3592233009708738e-05,
      "loss": 0.8266,
      "step": 1319
    },
    {
      "epoch": 2.883542919628212,
      "grad_norm": 0.06003335863351822,
      "learning_rate": 1.3349514563106795e-05,
      "loss": 0.9205,
      "step": 1320
    },
    {
      "epoch": 2.8857299070530344,
      "grad_norm": 0.12210050225257874,
      "learning_rate": 1.3106796116504854e-05,
      "loss": 0.6197,
      "step": 1321
    },
    {
      "epoch": 2.887916894477857,
      "grad_norm": 0.12230640649795532,
      "learning_rate": 1.2864077669902911e-05,
      "loss": 0.6101,
      "step": 1322
    },
    {
      "epoch": 2.890103881902679,
      "grad_norm": 0.11287673562765121,
      "learning_rate": 1.262135922330097e-05,
      "loss": 0.6778,
      "step": 1323
    },
    {
      "epoch": 2.8922908693275016,
      "grad_norm": 0.051977936178445816,
      "learning_rate": 1.237864077669903e-05,
      "loss": 0.5601,
      "step": 1324
    },
    {
      "epoch": 2.8944778567523235,
      "grad_norm": 0.07573337107896805,
      "learning_rate": 1.2135922330097087e-05,
      "loss": 0.6853,
      "step": 1325
    },
    {
      "epoch": 2.896664844177146,
      "grad_norm": 0.04770318791270256,
      "learning_rate": 1.1893203883495146e-05,
      "loss": 0.5123,
      "step": 1326
    },
    {
      "epoch": 2.8988518316019682,
      "grad_norm": 0.052877768874168396,
      "learning_rate": 1.1650485436893201e-05,
      "loss": 0.6725,
      "step": 1327
    },
    {
      "epoch": 2.9010388190267906,
      "grad_norm": 0.13692569732666016,
      "learning_rate": 1.140776699029126e-05,
      "loss": 0.6215,
      "step": 1328
    },
    {
      "epoch": 2.903225806451613,
      "grad_norm": 0.12803183495998383,
      "learning_rate": 1.116504854368932e-05,
      "loss": 0.7568,
      "step": 1329
    },
    {
      "epoch": 2.9054127938764354,
      "grad_norm": 0.11710110306739807,
      "learning_rate": 1.0922330097087376e-05,
      "loss": 0.5461,
      "step": 1330
    },
    {
      "epoch": 2.9075997813012577,
      "grad_norm": 0.05682337284088135,
      "learning_rate": 1.0679611650485435e-05,
      "loss": 0.4522,
      "step": 1331
    },
    {
      "epoch": 2.9097867687260797,
      "grad_norm": 0.07984013855457306,
      "learning_rate": 1.0436893203883494e-05,
      "loss": 0.7247,
      "step": 1332
    },
    {
      "epoch": 2.911973756150902,
      "grad_norm": 0.07054444402456284,
      "learning_rate": 1.0194174757281552e-05,
      "loss": 0.6919,
      "step": 1333
    },
    {
      "epoch": 2.9141607435757244,
      "grad_norm": 0.13668102025985718,
      "learning_rate": 9.95145631067961e-06,
      "loss": 0.6537,
      "step": 1334
    },
    {
      "epoch": 2.916347731000547,
      "grad_norm": 0.07120262086391449,
      "learning_rate": 9.708737864077668e-06,
      "loss": 0.6164,
      "step": 1335
    },
    {
      "epoch": 2.918534718425369,
      "grad_norm": 0.07631273567676544,
      "learning_rate": 9.466019417475727e-06,
      "loss": 0.5199,
      "step": 1336
    },
    {
      "epoch": 2.920721705850191,
      "grad_norm": 0.09133388847112656,
      "learning_rate": 9.223300970873786e-06,
      "loss": 0.6461,
      "step": 1337
    },
    {
      "epoch": 2.922908693275014,
      "grad_norm": 0.08395536988973618,
      "learning_rate": 8.980582524271843e-06,
      "loss": 0.8084,
      "step": 1338
    },
    {
      "epoch": 2.925095680699836,
      "grad_norm": 0.08669275045394897,
      "learning_rate": 8.737864077669902e-06,
      "loss": 0.6684,
      "step": 1339
    },
    {
      "epoch": 2.927282668124658,
      "grad_norm": 0.10852108895778656,
      "learning_rate": 8.495145631067961e-06,
      "loss": 0.9161,
      "step": 1340
    },
    {
      "epoch": 2.9294696555494806,
      "grad_norm": 0.05524509400129318,
      "learning_rate": 8.252427184466018e-06,
      "loss": 0.5381,
      "step": 1341
    },
    {
      "epoch": 2.931656642974303,
      "grad_norm": 0.08389575779438019,
      "learning_rate": 8.009708737864077e-06,
      "loss": 0.5115,
      "step": 1342
    },
    {
      "epoch": 2.9338436303991253,
      "grad_norm": 0.0699021965265274,
      "learning_rate": 7.766990291262136e-06,
      "loss": 0.791,
      "step": 1343
    },
    {
      "epoch": 2.9360306178239473,
      "grad_norm": 0.13843934237957,
      "learning_rate": 7.524271844660194e-06,
      "loss": 0.7158,
      "step": 1344
    },
    {
      "epoch": 2.93821760524877,
      "grad_norm": 0.07031416893005371,
      "learning_rate": 7.281553398058252e-06,
      "loss": 0.9197,
      "step": 1345
    },
    {
      "epoch": 2.940404592673592,
      "grad_norm": 0.07277525216341019,
      "learning_rate": 7.03883495145631e-06,
      "loss": 0.7927,
      "step": 1346
    },
    {
      "epoch": 2.9425915800984144,
      "grad_norm": 0.06796955317258835,
      "learning_rate": 6.796116504854369e-06,
      "loss": 0.5561,
      "step": 1347
    },
    {
      "epoch": 2.9447785675232367,
      "grad_norm": 0.09654916077852249,
      "learning_rate": 6.553398058252427e-06,
      "loss": 0.6991,
      "step": 1348
    },
    {
      "epoch": 2.946965554948059,
      "grad_norm": 0.05071938782930374,
      "learning_rate": 6.310679611650485e-06,
      "loss": 0.4628,
      "step": 1349
    },
    {
      "epoch": 2.9491525423728815,
      "grad_norm": 0.0948738306760788,
      "learning_rate": 6.067961165048543e-06,
      "loss": 0.5872,
      "step": 1350
    },
    {
      "epoch": 2.9513395297977034,
      "grad_norm": 0.09830582141876221,
      "learning_rate": 5.8252427184466006e-06,
      "loss": 0.6827,
      "step": 1351
    },
    {
      "epoch": 2.9535265172225262,
      "grad_norm": 0.06940705329179764,
      "learning_rate": 5.58252427184466e-06,
      "loss": 0.6454,
      "step": 1352
    },
    {
      "epoch": 2.955713504647348,
      "grad_norm": 0.10547400265932083,
      "learning_rate": 5.339805825242718e-06,
      "loss": 0.6769,
      "step": 1353
    },
    {
      "epoch": 2.9579004920721705,
      "grad_norm": 0.07688979804515839,
      "learning_rate": 5.097087378640776e-06,
      "loss": 0.6048,
      "step": 1354
    },
    {
      "epoch": 2.960087479496993,
      "grad_norm": 0.2706167995929718,
      "learning_rate": 4.854368932038834e-06,
      "loss": 0.9213,
      "step": 1355
    },
    {
      "epoch": 2.9622744669218153,
      "grad_norm": 0.07999067008495331,
      "learning_rate": 4.611650485436893e-06,
      "loss": 0.7272,
      "step": 1356
    },
    {
      "epoch": 2.9644614543466377,
      "grad_norm": 0.10268191993236542,
      "learning_rate": 4.368932038834951e-06,
      "loss": 0.7823,
      "step": 1357
    },
    {
      "epoch": 2.9666484417714596,
      "grad_norm": 0.2441248893737793,
      "learning_rate": 4.126213592233009e-06,
      "loss": 0.8201,
      "step": 1358
    },
    {
      "epoch": 2.968835429196282,
      "grad_norm": 0.07264109700918198,
      "learning_rate": 3.883495145631068e-06,
      "loss": 0.6676,
      "step": 1359
    },
    {
      "epoch": 2.9710224166211043,
      "grad_norm": 0.04458310455083847,
      "learning_rate": 3.640776699029126e-06,
      "loss": 0.6127,
      "step": 1360
    },
    {
      "epoch": 2.9732094040459267,
      "grad_norm": 0.08155495673418045,
      "learning_rate": 3.3980582524271844e-06,
      "loss": 0.5568,
      "step": 1361
    },
    {
      "epoch": 2.975396391470749,
      "grad_norm": 0.08346610516309738,
      "learning_rate": 3.1553398058252426e-06,
      "loss": 0.745,
      "step": 1362
    },
    {
      "epoch": 2.9775833788955715,
      "grad_norm": 0.0532771572470665,
      "learning_rate": 2.9126213592233003e-06,
      "loss": 0.5768,
      "step": 1363
    },
    {
      "epoch": 2.979770366320394,
      "grad_norm": 0.07556641101837158,
      "learning_rate": 2.669902912621359e-06,
      "loss": 0.5983,
      "step": 1364
    },
    {
      "epoch": 2.9819573537452158,
      "grad_norm": 0.09355679899454117,
      "learning_rate": 2.427184466019417e-06,
      "loss": 0.7703,
      "step": 1365
    },
    {
      "epoch": 2.984144341170038,
      "grad_norm": 0.11744330078363419,
      "learning_rate": 2.1844660194174755e-06,
      "loss": 0.8041,
      "step": 1366
    },
    {
      "epoch": 2.9863313285948605,
      "grad_norm": 0.08273312449455261,
      "learning_rate": 1.941747572815534e-06,
      "loss": 0.6304,
      "step": 1367
    },
    {
      "epoch": 2.988518316019683,
      "grad_norm": 0.05637767165899277,
      "learning_rate": 1.6990291262135922e-06,
      "loss": 0.5463,
      "step": 1368
    },
    {
      "epoch": 2.9907053034445052,
      "grad_norm": 0.05977838486433029,
      "learning_rate": 1.4563106796116501e-06,
      "loss": 0.713,
      "step": 1369
    },
    {
      "epoch": 2.9928922908693276,
      "grad_norm": 0.12927646934986115,
      "learning_rate": 1.2135922330097085e-06,
      "loss": 0.7835,
      "step": 1370
    },
    {
      "epoch": 2.99507927829415,
      "grad_norm": 0.06799733638763428,
      "learning_rate": 9.70873786407767e-07,
      "loss": 0.5287,
      "step": 1371
    },
    {
      "epoch": 2.997266265718972,
      "grad_norm": 0.1508604884147644,
      "learning_rate": 7.281553398058251e-07,
      "loss": 0.736,
      "step": 1372
    },
    {
      "epoch": 2.9994532531437943,
      "grad_norm": 0.07225324213504791,
      "learning_rate": 4.854368932038835e-07,
      "loss": 0.6506,
      "step": 1373
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.6369936466217041,
      "learning_rate": 2.4271844660194176e-07,
      "loss": 0.5374,
      "step": 1374
    },
    {
      "epoch": 3.0,
      "step": 1374,
      "total_flos": 8.559517791567217e+17,
      "train_loss": 0.6696628328667631,
      "train_runtime": 2857.8712,
      "train_samples_per_second": 7.68,
      "train_steps_per_second": 0.481
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1374,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 8.559517791567217e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}